fix: set event hook for wrapping values into SensitiveValue (#9177)

Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>

.cursor/skills/onyx-cli/SKILL.md

@@ -0,0 +1,186 @@
+---
+name: onyx-cli
+description: Query the Onyx knowledge base using the onyx-cli command. Use when the user wants to search company documents, ask questions about internal knowledge, query connected data sources, or look up information stored in Onyx.
+---
+
+# Onyx CLI — Agent Tool
+
+Onyx is an enterprise search and Gen-AI platform that connects to company documents, apps, and people. The `onyx-cli` CLI provides non-interactive commands to query the Onyx knowledge base and list available agents.
+
+## Prerequisites
+
+### 1. Check if installed
+
+```bash
+which onyx-cli
+```
+
+### 2. Install (if needed)
+
+**Primary — pip:**
+
+```bash
+pip install onyx-cli
+```
+
+**From source (Go):**
+
+```bash
+cd cli && go build -o onyx-cli . && sudo mv onyx-cli /usr/local/bin/
+```
+
+### 3. Check if configured
+
+```bash
+onyx-cli validate-config
+```
+
+This checks the config file exists, API key is present, and tests the server connection via `/api/me`. Exit code 0 on success, non-zero with a descriptive error on failure.
+
+If unconfigured, you have two options:
+
+**Option A — Interactive setup (requires user input):**
+
+```bash
+onyx-cli configure
+```
+
+This prompts for the Onyx server URL and API key, tests the connection, and saves config.
+
+**Option B — Environment variables (non-interactive, preferred for agents):**
+
+```bash
+export ONYX_SERVER_URL="https://your-onyx-server.com"  # default: https://cloud.onyx.app
+export ONYX_API_KEY="your-api-key"
+```
+
+Environment variables override the config file. If these are set, no config file is needed.
+
+| Variable | Required | Description |
+|----------|----------|-------------|
+| `ONYX_SERVER_URL` | No | Onyx server base URL (default: `https://cloud.onyx.app`) |
+| `ONYX_API_KEY` | Yes | API key for authentication |
+| `ONYX_PERSONA_ID` | No | Default agent/persona ID |
+
+If neither the config file nor environment variables are set, tell the user that `onyx-cli` needs to be configured and ask them to either:
+- Run `onyx-cli configure` interactively, or
+- Set `ONYX_SERVER_URL` and `ONYX_API_KEY` environment variables
+
+## Commands
+
+### Validate configuration
+
+```bash
+onyx-cli validate-config
+```
+
+Checks config file exists, API key is present, and tests the server connection. Use this before `ask` or `agents` to confirm the CLI is properly set up.
+
+### List available agents
+
+```bash
+onyx-cli agents
+```
+
+Prints a table of agent IDs, names, and descriptions. Use `--json` for structured output:
+
+```bash
+onyx-cli agents --json
+```
+
+Use agent IDs with `ask --agent-id` to query a specific agent.
+
+### Basic query (plain text output)
+
+```bash
+onyx-cli ask "What is our company's PTO policy?"
+```
+
+Streams the answer as plain text to stdout. Exit code 0 on success, non-zero on error.
+
+### JSON output (structured events)
+
+```bash
+onyx-cli ask --json "What authentication methods do we support?"
+```
+
+Outputs JSON-encoded parsed stream events (one object per line). Key event objects include message deltas, stop, errors, search-start, and citation payloads.
+
+Each line is a JSON object with this envelope:
+
+```json
+{"type": "<event_type>", "event": { ... }}
+```
+
+| Event Type | Description |
+|------------|-------------|
+| `message_delta` | Content token — concatenate all `content` fields for the full answer |
+| `stop` | Stream complete |
+| `error` | Error with `error` message field |
+| `search_tool_start` | Onyx started searching documents |
+| `citation_info` | Source citation — see shape below |
+
+`citation_info` event shape:
+
+```json
+{
+  "type": "citation_info",
+  "event": {
+    "citation_number": 1,
+    "document_id": "abc123def456",
+    "placement": {"turn_index": 0, "tab_index": 0, "sub_turn_index": null}
+  }
+}
+```
+
+`placement` is metadata about where in the conversation the citation appeared and can be ignored for most use cases.
+
+### Specify an agent
+
+```bash
+onyx-cli ask --agent-id 5 "Summarize our Q4 roadmap"
+```
+
+Uses a specific Onyx agent/persona instead of the default.
+
+### All flags
+
+| Flag | Type | Description |
+|------|------|-------------|
+| `--agent-id` | int | Agent ID to use (overrides default) |
+| `--json` | bool | Output raw NDJSON events instead of plain text |
+
+## Statelessness
+
+Each `onyx-cli ask` call creates an independent chat session. There is no built-in way to chain context across multiple `ask` invocations — every call starts fresh. If you need multi-turn conversation with memory, use the interactive TUI (`onyx-cli` or `onyx-cli chat`) instead.
+
+## When to Use
+
+Use `onyx-cli ask` when:
+
+- The user asks about company-specific information (policies, docs, processes)
+- You need to search internal knowledge bases or connected data sources
+- The user references Onyx, asks you to "search Onyx", or wants to query their documents
+- You need context from company wikis, Confluence, Google Drive, Slack, or other connected sources
+
+Do NOT use when:
+
+- The question is about general programming knowledge (use your own knowledge)
+- The user is asking about code in the current repository (use grep/read tools)
+- The user hasn't mentioned Onyx and the question doesn't require internal company data
+
+## Examples
+
+```bash
+# Simple question
+onyx-cli ask "What are the steps to deploy to production?"
+
+# Get structured output for parsing
+onyx-cli ask --json "List all active API integrations"
+
+# Use a specialized agent
+onyx-cli ask --agent-id 3 "What were the action items from last week's standup?"
+
+# Pipe the answer into another command
+onyx-cli ask "What is the database schema for users?" | head -20
+```

.github/workflows/deployment.yml

@@ -182,9 +182,53 @@ jobs:
           title: "🚨 Version Tag Check Failed"
           ref-name: ${{ github.ref_name }}
 
-  build-desktop:
+  # Create GitHub release first, before desktop builds start.
+  # This ensures all desktop matrix jobs upload to the same release instead of
+  # racing to create duplicate releases.
+  create-release:
     needs: determine-builds
     if: needs.determine-builds.outputs.build-desktop == 'true'
+    runs-on: ubuntu-slim
+    timeout-minutes: 10
+    permissions:
+      contents: write
+    outputs:
+      release-id: ${{ steps.create-release.outputs.id }}
+    steps:
+      - name: Checkout
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # ratchet:actions/checkout@v6
+        with:
+          persist-credentials: false
+
+      - name: Determine release tag
+        id: release-tag
+        env:
+          IS_TEST_RUN: ${{ needs.determine-builds.outputs.is-test-run }}
+          SHORT_SHA: ${{ needs.determine-builds.outputs.short-sha }}
+        run: |
+          if [ "${IS_TEST_RUN}" == "true" ]; then
+            echo "tag=v0.0.0-dev+${SHORT_SHA}" >> "$GITHUB_OUTPUT"
+          else
+            echo "tag=${GITHUB_REF_NAME}" >> "$GITHUB_OUTPUT"
+          fi
+
+      - name: Create GitHub Release
+        id: create-release
+        uses: softprops/action-gh-release@da05d552573ad5aba039eaac05058a918a7bf631 # ratchet:softprops/action-gh-release@v2
+        with:
+          tag_name: ${{ steps.release-tag.outputs.tag }}
+          name: ${{ steps.release-tag.outputs.tag }}
+          body: "See the assets to download this version and install."
+          draft: true
+          prerelease: false
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+  build-desktop:
+    needs:
+      - determine-builds
+      - create-release
+    if: needs.determine-builds.outputs.build-desktop == 'true'
     permissions:
       id-token: write
       contents: write
@@ -208,12 +252,12 @@ jobs:
     steps:
       - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # ratchet:actions/checkout@v6.0.2
         with:
-          # NOTE: persist-credentials is needed for tauri-action to create GitHub releases.
+          # NOTE: persist-credentials is needed for tauri-action to upload assets to GitHub releases.
           persist-credentials: true # zizmor: ignore[artipacked]
 
       - name: Configure AWS credentials
         if: startsWith(matrix.platform, 'macos-')
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -353,11 +397,9 @@ jobs:
           APPLE_SIGNING_IDENTITY: ${{ env.CERT_ID }}
           APPLE_TEAM_ID: ${{ env.APPLE_TEAM_ID }}
         with:
-          tagName: ${{ needs.determine-builds.outputs.is-test-run != 'true' && 'v__VERSION__' || format('v0.0.0-dev+{0}', needs.determine-builds.outputs.short-sha) }}
-          releaseName: ${{ needs.determine-builds.outputs.is-test-run != 'true' && 'v__VERSION__' || format('v0.0.0-dev+{0}', needs.determine-builds.outputs.short-sha) }}
-          releaseBody: "See the assets to download this version and install."
-          releaseDraft: true
-          prerelease: false
+          # Use the release created by the create-release job to avoid race conditions
+          # when multiple matrix jobs try to create/update the same release simultaneously
+          releaseId: ${{ needs.create-release.outputs.release-id }}
           assetNamePattern: "[name]_[arch][ext]"
           args: ${{ matrix.args }}
 
@@ -384,7 +426,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -458,7 +500,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -527,7 +569,7 @@ jobs:
       - uses: runs-on/action@cd2b598b0515d39d78c38a02d529db87d2196d1e # ratchet:runs-on/action@v2
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -597,7 +639,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -679,7 +721,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -756,7 +798,7 @@ jobs:
       - uses: runs-on/action@cd2b598b0515d39d78c38a02d529db87d2196d1e # ratchet:runs-on/action@v2
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -823,7 +865,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -896,7 +938,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -964,7 +1006,7 @@ jobs:
       - uses: runs-on/action@cd2b598b0515d39d78c38a02d529db87d2196d1e # ratchet:runs-on/action@v2
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -1034,7 +1076,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -1107,7 +1149,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -1176,7 +1218,7 @@ jobs:
       - uses: runs-on/action@cd2b598b0515d39d78c38a02d529db87d2196d1e # ratchet:runs-on/action@v2
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -1246,7 +1288,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -1326,7 +1368,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -1400,7 +1442,7 @@ jobs:
       - uses: runs-on/action@cd2b598b0515d39d78c38a02d529db87d2196d1e # ratchet:runs-on/action@v2
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -1465,7 +1507,7 @@ jobs:
       - uses: runs-on/action@cd2b598b0515d39d78c38a02d529db87d2196d1e # ratchet:runs-on/action@v2
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -1520,7 +1562,7 @@ jobs:
       - uses: runs-on/action@cd2b598b0515d39d78c38a02d529db87d2196d1e # ratchet:runs-on/action@v2
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -1580,7 +1622,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -1637,7 +1679,7 @@ jobs:
       - uses: runs-on/action@cd2b598b0515d39d78c38a02d529db87d2196d1e # ratchet:runs-on/action@v2
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2

.github/workflows/nightly-llm-provider-chat.yml

@@ -15,7 +15,8 @@ permissions:
 jobs:
   provider-chat-test:
     uses: ./.github/workflows/reusable-nightly-llm-provider-chat.yml
-    secrets: inherit
+    secrets:
+      AWS_OIDC_ROLE_ARN: ${{ secrets.AWS_OIDC_ROLE_ARN }}
     permissions:
       contents: read
       id-token: write

.github/workflows/post-merge-beta-cherry-pick.yml

@@ -6,11 +6,13 @@ on:
       - main
 
 permissions:
-  contents: write
-  pull-requests: write
+  contents: read
 
 jobs:
   cherry-pick-to-latest-release:
+    permissions:
+      contents: write
+      pull-requests: write
     outputs:
       should_cherrypick: ${{ steps.gate.outputs.should_cherrypick }}
       pr_number: ${{ steps.gate.outputs.pr_number }}

.github/workflows/pr-desktop-build.yml

@@ -57,7 +57,7 @@ jobs:
           cache-dependency-path: ./desktop/package-lock.json
 
       - name: Setup Rust
-        uses: dtolnay/rust-toolchain@4be9e76fd7c4901c61fb841f559994984270fce7
+        uses: dtolnay/rust-toolchain@efa25f7f19611383d5b0ccf2d1c8914531636bf9
         with:
           toolchain: stable
           targets: ${{ matrix.target }}

.github/workflows/pr-golang-tests.yml

@@ -0,0 +1,56 @@
+name: Golang Tests
+concurrency:
+  group: Golang-Tests-${{ github.workflow }}-${{ github.head_ref || github.event.workflow_run.head_branch || github.run_id }}
+  cancel-in-progress: true
+
+on:
+  merge_group:
+  pull_request:
+    branches:
+      - main
+      - "release/**"
+  push:
+    tags:
+      - "v*.*.*"
+
+permissions: {}
+
+env:
+  GO_VERSION: "1.26"
+
+jobs:
+  detect-modules:
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    outputs:
+      modules: ${{ steps.set-modules.outputs.modules }}
+    steps:
+      - uses: actions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8
+        with:
+          persist-credentials: false
+      - id: set-modules
+        run: echo "modules=$(find . -name 'go.mod' -exec dirname {} \; | jq -Rc '[.,inputs]')" >> "$GITHUB_OUTPUT"
+
+  golang:
+    needs: detect-modules
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    strategy:
+      matrix:
+        modules: ${{ fromJSON(needs.detect-modules.outputs.modules) }}
+    steps:
+      - uses: actions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8 # ratchet:actions/checkout@v6
+        with:
+          persist-credentials: false
+      - uses: actions/setup-go@4dc6199c7b1a012772edbd06daecab0f50c9053c # zizmor: ignore[cache-poisoning]
+        with:
+          go-version: ${{ env.GO_VERSION }}
+          cache-dependency-path: "**/go.sum"
+
+      - run: go mod tidy
+        working-directory: ${{ matrix.modules }}
+      - run: git diff --exit-code go.mod go.sum
+        working-directory: ${{ matrix.modules }}
+
+      - run: go test ./...
+        working-directory: ${{ matrix.modules }}

.github/workflows/pr-helm-chart-testing.yml

@@ -71,7 +71,7 @@ jobs:
 
       - name: Create kind cluster
         if: steps.list-changed.outputs.changed == 'true'
-        uses: helm/kind-action@92086f6be054225fa813e0a4b13787fc9088faab # ratchet:helm/kind-action@v1.13.0
+        uses: helm/kind-action@ef37e7f390d99f746eb8b610417061a60e82a6cc # ratchet:helm/kind-action@v1.14.0
 
       - name: Pre-install cluster status check
         if: steps.list-changed.outputs.changed == 'true'

.github/workflows/pr-jest-tests.yml

@@ -31,7 +31,7 @@ jobs:
         uses: actions/setup-node@6044e13b5dc448c55e2357c09f80417699197238 # ratchet:actions/setup-node@v4
         with:
           node-version: 22
-          cache: "npm"
+          cache: "npm" # zizmor: ignore[cache-poisoning] test-only workflow; no deploy artifacts
           cache-dependency-path: ./web/package-lock.json
 
       - name: Install node dependencies

.github/workflows/pr-playwright-tests.yml

@@ -461,7 +461,7 @@ jobs:
       # --- Visual Regression Diff ---
       - name: Configure AWS credentials
         if: always()
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2

.github/workflows/pr-quality-checks.yml

@@ -38,9 +38,9 @@ jobs:
       - name: Install node dependencies
         working-directory: ./web
         run: npm ci
-      - uses: j178/prek-action@9d6a3097e0c1865ecce00cfb89fe80f2ee91b547 # ratchet:j178/prek-action@v1
+      - uses: j178/prek-action@0bb87d7f00b0c99306c8bcb8b8beba1eb581c037 # ratchet:j178/prek-action@v1
         with:
-          prek-version: '0.2.21'
+          prek-version: '0.3.4'
           extra-args: ${{ github.event_name == 'pull_request' && format('--from-ref {0} --to-ref {1}', github.event.pull_request.base.sha, github.event.pull_request.head.sha) || github.event_name == 'merge_group' && format('--from-ref {0} --to-ref {1}', github.event.merge_group.base_sha, github.event.merge_group.head_sha) || github.ref_name == 'main' && '--all-files' || '' }}
       - name: Check Actions
         uses: giner/check-actions@28d366c7cbbe235f9624a88aa31a628167eee28c # ratchet:giner/check-actions@v1.0.1

.github/workflows/release-cli.yml

@@ -0,0 +1,214 @@
+name: Release CLI
+
+on:
+  push:
+    tags:
+      - "cli/v*.*.*"
+
+jobs:
+  pypi:
+    runs-on: ubuntu-latest
+    environment:
+      name: release-cli
+    permissions:
+      id-token: write
+    timeout-minutes: 10
+    strategy:
+      matrix:
+        os-arch:
+          - { goos: "linux", goarch: "amd64" }
+          - { goos: "linux", goarch: "arm64" }
+          - { goos: "windows", goarch: "amd64" }
+          - { goos: "windows", goarch: "arm64" }
+          - { goos: "darwin", goarch: "amd64" }
+          - { goos: "darwin", goarch: "arm64" }
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # ratchet:actions/checkout@v6
+        with:
+          persist-credentials: false
+      - uses: astral-sh/setup-uv@5a095e7a2014a4212f075830d4f7277575a9d098 # ratchet:astral-sh/setup-uv@v7
+        with:
+          enable-cache: false
+          version: "0.9.9"
+      - run: |
+          GOOS="${{ matrix.os-arch.goos }}" \
+          GOARCH="${{ matrix.os-arch.goarch }}" \
+          uv build --wheel
+        working-directory: cli
+      - run: uv publish
+        working-directory: cli
+
+  docker-amd64:
+    runs-on:
+      - runs-on
+      - runner=2cpu-linux-x64
+      - run-id=${{ github.run_id }}-cli-amd64
+      - extras=ecr-cache
+    environment: deploy
+    permissions:
+      id-token: write
+    timeout-minutes: 30
+    outputs:
+      digest: ${{ steps.build.outputs.digest }}
+    env:
+      REGISTRY_IMAGE: onyxdotapp/onyx-cli
+    steps:
+      - uses: runs-on/action@cd2b598b0515d39d78c38a02d529db87d2196d1e # ratchet:runs-on/action@v2
+
+      - name: Checkout
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # ratchet:actions/checkout@v6
+        with:
+          persist-credentials: false
+
+      - name: Configure AWS credentials
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7 # ratchet:aws-actions/configure-aws-credentials@v6.0.0
+        with:
+          role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
+          aws-region: us-east-2
+
+      - name: Get AWS Secrets
+        uses: aws-actions/aws-secretsmanager-get-secrets@a9a7eb4e2f2871d30dc5b892576fde60a2ecc802 # ratchet:aws-actions/aws-secretsmanager-get-secrets@v2.0.10
+        with:
+          secret-ids: |
+            DOCKER_USERNAME, deploy/docker-username
+            DOCKER_TOKEN, deploy/docker-token
+          parse-json-secrets: true
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@4d04d5d9486b7bd6fa91e7baf45bbb4f8b9deedd # ratchet:docker/setup-buildx-action@v4
+
+      - name: Login to Docker Hub
+        uses: docker/login-action@b45d80f862d83dbcd57f89517bcf500b2ab88fb2 # ratchet:docker/login-action@v4
+        with:
+          username: ${{ env.DOCKER_USERNAME }}
+          password: ${{ env.DOCKER_TOKEN }}
+
+      - name: Build and push AMD64
+        id: build
+        uses: docker/build-push-action@d08e5c354a6adb9ed34480a06d141179aa583294 # ratchet:docker/build-push-action@v7
+        with:
+          context: ./cli
+          file: ./cli/Dockerfile
+          platforms: linux/amd64
+          cache-from: type=registry,ref=${{ env.REGISTRY_IMAGE }}:latest
+          cache-to: type=inline
+          outputs: type=image,name=${{ env.REGISTRY_IMAGE }},push-by-digest=true,name-canonical=true,push=true
+
+  docker-arm64:
+    runs-on:
+      - runs-on
+      - runner=2cpu-linux-arm64
+      - run-id=${{ github.run_id }}-cli-arm64
+      - extras=ecr-cache
+    environment: deploy
+    permissions:
+      id-token: write
+    timeout-minutes: 30
+    outputs:
+      digest: ${{ steps.build.outputs.digest }}
+    env:
+      REGISTRY_IMAGE: onyxdotapp/onyx-cli
+    steps:
+      - uses: runs-on/action@cd2b598b0515d39d78c38a02d529db87d2196d1e # ratchet:runs-on/action@v2
+
+      - name: Checkout
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # ratchet:actions/checkout@v6
+        with:
+          persist-credentials: false
+
+      - name: Configure AWS credentials
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7 # ratchet:aws-actions/configure-aws-credentials@v6.0.0
+        with:
+          role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
+          aws-region: us-east-2
+
+      - name: Get AWS Secrets
+        uses: aws-actions/aws-secretsmanager-get-secrets@a9a7eb4e2f2871d30dc5b892576fde60a2ecc802 # ratchet:aws-actions/aws-secretsmanager-get-secrets@v2.0.10
+        with:
+          secret-ids: |
+            DOCKER_USERNAME, deploy/docker-username
+            DOCKER_TOKEN, deploy/docker-token
+          parse-json-secrets: true
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@4d04d5d9486b7bd6fa91e7baf45bbb4f8b9deedd # ratchet:docker/setup-buildx-action@v4
+
+      - name: Login to Docker Hub
+        uses: docker/login-action@b45d80f862d83dbcd57f89517bcf500b2ab88fb2 # ratchet:docker/login-action@v4
+        with:
+          username: ${{ env.DOCKER_USERNAME }}
+          password: ${{ env.DOCKER_TOKEN }}
+
+      - name: Build and push ARM64
+        id: build
+        uses: docker/build-push-action@d08e5c354a6adb9ed34480a06d141179aa583294 # ratchet:docker/build-push-action@v7
+        with:
+          context: ./cli
+          file: ./cli/Dockerfile
+          platforms: linux/arm64
+          cache-from: type=registry,ref=${{ env.REGISTRY_IMAGE }}:latest
+          cache-to: type=inline
+          outputs: type=image,name=${{ env.REGISTRY_IMAGE }},push-by-digest=true,name-canonical=true,push=true
+
+  merge-docker:
+    needs:
+      - docker-amd64
+      - docker-arm64
+    runs-on:
+      - runs-on
+      - runner=2cpu-linux-x64
+      - run-id=${{ github.run_id }}-cli-merge
+    environment: deploy
+    permissions:
+      id-token: write
+    timeout-minutes: 10
+    env:
+      REGISTRY_IMAGE: onyxdotapp/onyx-cli
+    steps:
+      - uses: runs-on/action@cd2b598b0515d39d78c38a02d529db87d2196d1e # ratchet:runs-on/action@v2
+
+      - name: Configure AWS credentials
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7 # ratchet:aws-actions/configure-aws-credentials@v6.0.0
+        with:
+          role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
+          aws-region: us-east-2
+
+      - name: Get AWS Secrets
+        uses: aws-actions/aws-secretsmanager-get-secrets@a9a7eb4e2f2871d30dc5b892576fde60a2ecc802 # ratchet:aws-actions/aws-secretsmanager-get-secrets@v2.0.10
+        with:
+          secret-ids: |
+            DOCKER_USERNAME, deploy/docker-username
+            DOCKER_TOKEN, deploy/docker-token
+          parse-json-secrets: true
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@4d04d5d9486b7bd6fa91e7baf45bbb4f8b9deedd # ratchet:docker/setup-buildx-action@v4
+
+      - name: Login to Docker Hub
+        uses: docker/login-action@b45d80f862d83dbcd57f89517bcf500b2ab88fb2 # ratchet:docker/login-action@v4
+        with:
+          username: ${{ env.DOCKER_USERNAME }}
+          password: ${{ env.DOCKER_TOKEN }}
+
+      - name: Create and push manifest
+        env:
+          AMD64_DIGEST: ${{ needs.docker-amd64.outputs.digest }}
+          ARM64_DIGEST: ${{ needs.docker-arm64.outputs.digest }}
+          TAG: ${{ github.ref_name }}
+        run: |
+          SANITIZED_TAG="${TAG#cli/}"
+          IMAGES=(
+            "${REGISTRY_IMAGE}@${AMD64_DIGEST}"
+            "${REGISTRY_IMAGE}@${ARM64_DIGEST}"
+          )
+
+          if [[ "$TAG" =~ ^cli/v[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
+            docker buildx imagetools create \
+              -t "${REGISTRY_IMAGE}:${SANITIZED_TAG}" \
+              -t "${REGISTRY_IMAGE}:latest" \
+              "${IMAGES[@]}"
+          else
+            docker buildx imagetools create \
+              -t "${REGISTRY_IMAGE}:${SANITIZED_TAG}" \
+              "${IMAGES[@]}"
+          fi

.github/workflows/release-devtools.yml

@@ -22,12 +22,10 @@ jobs:
           - { goos: "windows", goarch: "arm64" }
           - { goos: "darwin", goarch: "amd64" }
           - { goos: "darwin", goarch: "arm64" }
-          - { goos: "", goarch: "" }
     steps:
       - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # ratchet:actions/checkout@v6
         with:
           persist-credentials: false
-          fetch-depth: 0
       - uses: astral-sh/setup-uv@61cb8a9741eeb8a550a1b8544337180c0fc8476b # ratchet:astral-sh/setup-uv@v7
         with:
           enable-cache: false

.github/workflows/reusable-nightly-llm-provider-chat.yml

@@ -48,6 +48,10 @@ on:
         required: false
         default: true
         type: boolean
+    secrets:
+      AWS_OIDC_ROLE_ARN:
+        description: "AWS role ARN for OIDC auth"
+        required: true
 
 permissions:
   contents: read
@@ -73,7 +77,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -116,7 +120,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -158,7 +162,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -264,7 +268,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2

.github/workflows/sandbox-deployment.yml

@@ -110,7 +110,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -180,7 +180,7 @@ jobs:
           persist-credentials: false
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2
@@ -244,7 +244,7 @@ jobs:
       - uses: runs-on/action@cd2b598b0515d39d78c38a02d529db87d2196d1e # ratchet:runs-on/action@v2
 
       - name: Configure AWS credentials
-        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708
+        uses: aws-actions/configure-aws-credentials@8df5847569e6427dd6c4fb1cf565c83acfa8afa7
         with:
           role-to-assume: ${{ secrets.AWS_OIDC_ROLE_ARN }}
           aws-region: us-east-2

.pre-commit-config.yaml

@@ -119,10 +119,11 @@ repos:
           ]
 
   - repo: https://github.com/golangci/golangci-lint
-    rev: 9f61b0f53f80672872fced07b6874397c3ed197b # frozen: v2.7.2
+    rev: 5d1e709b7be35cb2025444e19de266b056b7b7ee # frozen: v2.10.1
     hooks:
       - id: golangci-lint
-        entry: bash -c "find tools/ -name go.mod -print0 | xargs -0 -I{} bash -c 'cd \"$(dirname {})\" && golangci-lint run ./...'"
+        language_version: "1.26.0"
+        entry: bash -c "find . -name go.mod -not -path './.venv/*' -print0 | xargs -0 -I{} bash -c 'cd \"$(dirname {})\" && golangci-lint run ./...'"
 
   - repo: https://github.com/astral-sh/ruff-pre-commit
     # Ruff version.

AGENTS.md

@@ -104,6 +104,10 @@ Onyx uses Celery for asynchronous task processing with multiple specialized work
 
 - Always use `@shared_task` rather than `@celery_app`
 - Put tasks under `background/celery/tasks/` or `ee/background/celery/tasks`
+- Never enqueue a task without an expiration. Always supply `expires=` when
+  sending tasks, either from the beat schedule or directly from another task. It
+  should never be acceptable to submit code which enqueues tasks without an
+  expiration, as doing so can lead to unbounded task queue growth.
 
 **Defining APIs**:
 When creating new FastAPI APIs, do NOT use the `response_model` field. Instead, just type the
@@ -540,6 +544,8 @@ To run them:
 npx playwright test <TEST_NAME>
 ```
 
+For shared fixtures, best practices, and detailed guidance, see `backend/tests/README.md`.
+
 ## Logs
 
 When (1) writing integration tests or (2) doing live tests (e.g. curl / playwright) you can get access

backend/Dockerfile

@@ -141,6 +141,7 @@ COPY --chown=onyx:onyx ./scripts/debugging /app/scripts/debugging
 COPY --chown=onyx:onyx ./scripts/force_delete_connector_by_id.py /app/scripts/force_delete_connector_by_id.py
 COPY --chown=onyx:onyx ./scripts/supervisord_entrypoint.sh /app/scripts/supervisord_entrypoint.sh
 COPY --chown=onyx:onyx ./scripts/setup_craft_templates.sh /app/scripts/setup_craft_templates.sh
+COPY --chown=onyx:onyx ./scripts/reencrypt_secrets.py /app/scripts/reencrypt_secrets.py
 RUN chmod +x /app/scripts/supervisord_entrypoint.sh /app/scripts/setup_craft_templates.sh
 
 # Run Craft template setup at build time when ENABLE_CRAFT=true

backend/alembic_tenants/versions/3b9f09038764_add_read_only_kg_user.py

@@ -11,7 +11,6 @@ from sqlalchemy import text
 from alembic import op
 from onyx.configs.app_configs import DB_READONLY_PASSWORD
 from onyx.configs.app_configs import DB_READONLY_USER
-from shared_configs.configs import MULTI_TENANT
 
 
 # revision identifiers, used by Alembic.
@@ -22,59 +21,52 @@ depends_on = None
 
 
 def upgrade() -> None:
-    if MULTI_TENANT:
+    # Enable pg_trgm extension if not already enabled
+    op.execute("CREATE EXTENSION IF NOT EXISTS pg_trgm")
 
-        # Enable pg_trgm extension if not already enabled
-        op.execute("CREATE EXTENSION IF NOT EXISTS pg_trgm")
+    # Create the read-only db user if it does not already exist.
+    if not (DB_READONLY_USER and DB_READONLY_PASSWORD):
+        raise Exception("DB_READONLY_USER or DB_READONLY_PASSWORD is not set")
 
-        # Create read-only db user here only in multi-tenant mode. For single-tenant mode,
-        # the user is created in the standard migration.
-        if not (DB_READONLY_USER and DB_READONLY_PASSWORD):
-            raise Exception("DB_READONLY_USER or DB_READONLY_PASSWORD is not set")
-
-        op.execute(
-            text(
-                f"""
-                DO $$
-                BEGIN
-                    -- Check if the read-only user already exists
-                    IF NOT EXISTS (SELECT FROM pg_catalog.pg_roles WHERE rolname = '{DB_READONLY_USER}') THEN
-                        -- Create the read-only user with the specified password
-                        EXECUTE format('CREATE USER %I WITH PASSWORD %L', '{DB_READONLY_USER}', '{DB_READONLY_PASSWORD}');
-                        -- First revoke all privileges to ensure a clean slate
-                        EXECUTE format('REVOKE ALL ON DATABASE %I FROM %I', current_database(), '{DB_READONLY_USER}');
-                        -- Grant only the CONNECT privilege to allow the user to connect to the database
-                        -- but not perform any operations without additional specific grants
-                        EXECUTE format('GRANT CONNECT ON DATABASE %I TO %I', current_database(), '{DB_READONLY_USER}');
-                    END IF;
-                END
-                $$;
-                """
-            )
-        )
-
-
-def downgrade() -> None:
-    if MULTI_TENANT:
-        # Drop read-only db user here only in single tenant mode. For multi-tenant mode,
-        # the user is dropped in the alembic_tenants migration.
-
-        op.execute(
-            text(
-                f"""
+    op.execute(
+        text(
+            f"""
             DO $$
             BEGIN
-                IF EXISTS (SELECT FROM pg_catalog.pg_roles WHERE rolname = '{DB_READONLY_USER}') THEN
-                    -- First revoke all privileges from the database
+                -- Check if the read-only user already exists
+                IF NOT EXISTS (SELECT FROM pg_catalog.pg_roles WHERE rolname = '{DB_READONLY_USER}') THEN
+                    -- Create the read-only user with the specified password
+                    EXECUTE format('CREATE USER %I WITH PASSWORD %L', '{DB_READONLY_USER}', '{DB_READONLY_PASSWORD}');
+                    -- First revoke all privileges to ensure a clean slate
                     EXECUTE format('REVOKE ALL ON DATABASE %I FROM %I', current_database(), '{DB_READONLY_USER}');
-                    -- Then revoke all privileges from the public schema
-                    EXECUTE format('REVOKE ALL ON SCHEMA public FROM %I', '{DB_READONLY_USER}');
-                    -- Then drop the user
-                    EXECUTE format('DROP USER %I', '{DB_READONLY_USER}');
+                    -- Grant only the CONNECT privilege to allow the user to connect to the database
+                    -- but not perform any operations without additional specific grants
+                    EXECUTE format('GRANT CONNECT ON DATABASE %I TO %I', current_database(), '{DB_READONLY_USER}');
                 END IF;
             END
             $$;
-        """
-            )
+            """
         )
-        op.execute(text("DROP EXTENSION IF EXISTS pg_trgm"))
+    )
+
+
+def downgrade() -> None:
+    op.execute(
+        text(
+            f"""
+        DO $$
+        BEGIN
+            IF EXISTS (SELECT FROM pg_catalog.pg_roles WHERE rolname = '{DB_READONLY_USER}') THEN
+                -- First revoke all privileges from the database
+                EXECUTE format('REVOKE ALL ON DATABASE %I FROM %I', current_database(), '{DB_READONLY_USER}');
+                -- Then revoke all privileges from the public schema
+                EXECUTE format('REVOKE ALL ON SCHEMA public FROM %I', '{DB_READONLY_USER}');
+                -- Then drop the user
+                EXECUTE format('DROP USER %I', '{DB_READONLY_USER}');
+            END IF;
+        END
+        $$;
+    """
+        )
+    )
+    op.execute(text("DROP EXTENSION IF EXISTS pg_trgm"))

backend/ee/onyx/server/billing/api.py

@@ -246,7 +246,11 @@ async def get_billing_information(
         )
     except OnyxError as e:
         # Open circuit breaker on connection failures (self-hosted only)
-        if e.status_code in (502, 503, 504):
+        if e.status_code in (
+            OnyxErrorCode.BAD_GATEWAY.status_code,
+            OnyxErrorCode.SERVICE_UNAVAILABLE.status_code,
+            OnyxErrorCode.GATEWAY_TIMEOUT.status_code,
+        ):
             _open_billing_circuit()
         raise
 

backend/ee/onyx/utils/encryption.py

@@ -14,67 +14,91 @@ from onyx.utils.variable_functionality import fetch_versioned_implementation
 logger = setup_logger()
 
 
-@lru_cache(maxsize=1)
+@lru_cache(maxsize=2)
 def _get_trimmed_key(key: str) -> bytes:
     encoded_key = key.encode()
     key_length = len(encoded_key)
     if key_length < 16:
         raise RuntimeError("Invalid ENCRYPTION_KEY_SECRET - too short")
-    elif key_length > 32:
-        key = key[:32]
-    elif key_length not in (16, 24, 32):
-        valid_lengths = [16, 24, 32]
-        key = key[: min(valid_lengths, key=lambda x: abs(x - key_length))]
 
-    return encoded_key
+    # Trim to the largest valid AES key size that fits
+    valid_lengths = [32, 24, 16]
+    for size in valid_lengths:
+        if key_length >= size:
+            return encoded_key[:size]
+
+    raise AssertionError("unreachable")
 
 
-def _encrypt_string(input_str: str) -> bytes:
-    if not ENCRYPTION_KEY_SECRET:
+def _encrypt_string(input_str: str, key: str | None = None) -> bytes:
+    effective_key = key if key is not None else ENCRYPTION_KEY_SECRET
+    if not effective_key:
         return input_str.encode()
 
-    key = _get_trimmed_key(ENCRYPTION_KEY_SECRET)
+    trimmed = _get_trimmed_key(effective_key)
     iv = urandom(16)
     padder = padding.PKCS7(algorithms.AES.block_size).padder()
     padded_data = padder.update(input_str.encode()) + padder.finalize()
 
-    cipher = Cipher(algorithms.AES(key), modes.CBC(iv), backend=default_backend())
+    cipher = Cipher(algorithms.AES(trimmed), modes.CBC(iv), backend=default_backend())
     encryptor = cipher.encryptor()
     encrypted_data = encryptor.update(padded_data) + encryptor.finalize()
 
     return iv + encrypted_data
 
 
-def _decrypt_bytes(input_bytes: bytes) -> str:
-    if not ENCRYPTION_KEY_SECRET:
+def _decrypt_bytes(input_bytes: bytes, key: str | None = None) -> str:
+    effective_key = key if key is not None else ENCRYPTION_KEY_SECRET
+    if not effective_key:
         return input_bytes.decode()
 
-    key = _get_trimmed_key(ENCRYPTION_KEY_SECRET)
-    iv = input_bytes[:16]
-    encrypted_data = input_bytes[16:]
+    trimmed = _get_trimmed_key(effective_key)
+    try:
+        iv = input_bytes[:16]
+        encrypted_data = input_bytes[16:]
 
-    cipher = Cipher(algorithms.AES(key), modes.CBC(iv), backend=default_backend())
-    decryptor = cipher.decryptor()
-    decrypted_padded_data = decryptor.update(encrypted_data) + decryptor.finalize()
+        cipher = Cipher(
+            algorithms.AES(trimmed), modes.CBC(iv), backend=default_backend()
+        )
+        decryptor = cipher.decryptor()
+        decrypted_padded_data = decryptor.update(encrypted_data) + decryptor.finalize()
 
-    unpadder = padding.PKCS7(algorithms.AES.block_size).unpadder()
-    decrypted_data = unpadder.update(decrypted_padded_data) + unpadder.finalize()
+        unpadder = padding.PKCS7(algorithms.AES.block_size).unpadder()
+        decrypted_data = unpadder.update(decrypted_padded_data) + unpadder.finalize()
 
-    return decrypted_data.decode()
+        return decrypted_data.decode()
+    except (ValueError, UnicodeDecodeError):
+        if key is not None:
+            # Explicit key was provided — don't fall back silently
+            raise
+        # Read path: attempt raw UTF-8 decode as a fallback for legacy data.
+        # Does NOT handle data encrypted with a different key — that
+        # ciphertext is not valid UTF-8 and will raise below.
+        logger.warning(
+            "AES decryption failed — falling back to raw decode. "
+            "Run the re-encrypt secrets script to rotate to the current key."
+        )
+        try:
+            return input_bytes.decode()
+        except UnicodeDecodeError:
+            raise ValueError(
+                "Data is not valid UTF-8 — likely encrypted with a different key. "
+                "Run the re-encrypt secrets script to rotate to the current key."
+            ) from None
 
 
-def encrypt_string_to_bytes(input_str: str) -> bytes:
+def encrypt_string_to_bytes(input_str: str, key: str | None = None) -> bytes:
     versioned_encryption_fn = fetch_versioned_implementation(
         "onyx.utils.encryption", "_encrypt_string"
     )
-    return versioned_encryption_fn(input_str)
+    return versioned_encryption_fn(input_str, key=key)
 
 
-def decrypt_bytes_to_string(input_bytes: bytes) -> str:
+def decrypt_bytes_to_string(input_bytes: bytes, key: str | None = None) -> str:
     versioned_decryption_fn = fetch_versioned_implementation(
         "onyx.utils.encryption", "_decrypt_bytes"
     )
-    return versioned_decryption_fn(input_bytes)
+    return versioned_decryption_fn(input_bytes, key=key)
 
 
 def test_encryption() -> None:

backend/onyx/background/indexing/run_docfetching.py

@@ -58,8 +58,6 @@ from onyx.file_store.document_batch_storage import DocumentBatchStorage
 from onyx.file_store.document_batch_storage import get_document_batch_storage
 from onyx.indexing.indexing_heartbeat import IndexingHeartbeatInterface
 from onyx.indexing.indexing_pipeline import index_doc_batch_prepare
-from onyx.indexing.postgres_sanitization import sanitize_document_for_postgres
-from onyx.indexing.postgres_sanitization import sanitize_hierarchy_nodes_for_postgres
 from onyx.redis.redis_hierarchy import cache_hierarchy_nodes_batch
 from onyx.redis.redis_hierarchy import ensure_source_node_exists
 from onyx.redis.redis_hierarchy import get_node_id_from_raw_id
@@ -71,6 +69,8 @@ from onyx.server.features.build.indexing.persistent_document_writer import (
 )
 from onyx.utils.logger import setup_logger
 from onyx.utils.middleware import make_randomized_onyx_request_id
+from onyx.utils.postgres_sanitization import sanitize_document_for_postgres
+from onyx.utils.postgres_sanitization import sanitize_hierarchy_nodes_for_postgres
 from onyx.utils.variable_functionality import global_version
 from shared_configs.configs import MULTI_TENANT
 from shared_configs.contextvars import INDEX_ATTEMPT_INFO_CONTEXTVAR

backend/onyx/chat/llm_loop.py

@@ -36,7 +36,6 @@ from onyx.db.memory import add_memory
 from onyx.db.memory import update_memory_at_index
 from onyx.db.memory import UserMemoryContext
 from onyx.db.models import Persona
-from onyx.llm.constants import LlmProviderNames
 from onyx.llm.interfaces import LLM
 from onyx.llm.interfaces import LLMUserIdentity
 from onyx.llm.interfaces import ToolChoiceOptions
@@ -84,28 +83,6 @@ def _looks_like_xml_tool_call_payload(text: str | None) -> bool:
     )
 
 
-def _should_keep_bedrock_tool_definitions(
-    llm: object, simple_chat_history: list[ChatMessageSimple]
-) -> bool:
-    """Bedrock requires tool config when history includes toolUse/toolResult blocks."""
-    model_provider = getattr(getattr(llm, "config", None), "model_provider", None)
-    if model_provider not in {
-        LlmProviderNames.BEDROCK,
-        LlmProviderNames.BEDROCK_CONVERSE,
-    }:
-        return False
-
-    return any(
-        (
-            msg.message_type == MessageType.ASSISTANT
-            and msg.tool_calls
-            and len(msg.tool_calls) > 0
-        )
-        or msg.message_type == MessageType.TOOL_CALL_RESPONSE
-        for msg in simple_chat_history
-    )
-
-
 def _try_fallback_tool_extraction(
     llm_step_result: LlmStepResult,
     tool_choice: ToolChoiceOptions,
@@ -686,12 +663,7 @@ def run_llm_loop(
             elif out_of_cycles or ran_image_gen:
                 # Last cycle, no tools allowed, just answer!
                 tool_choice = ToolChoiceOptions.NONE
-                # Bedrock requires tool config in requests that include toolUse/toolResult history.
-                final_tools = (
-                    tools
-                    if _should_keep_bedrock_tool_definitions(llm, simple_chat_history)
-                    else []
-                )
+                final_tools = []
             else:
                 tool_choice = ToolChoiceOptions.AUTO
                 final_tools = tools

backend/onyx/chat/llm_step.py

@@ -15,6 +15,7 @@ from onyx.chat.citation_processor import DynamicCitationProcessor
 from onyx.chat.emitter import Emitter
 from onyx.chat.models import ChatMessageSimple
 from onyx.chat.models import LlmStepResult
+from onyx.chat.tool_call_args_streaming import maybe_emit_argument_delta
 from onyx.configs.app_configs import LOG_ONYX_MODEL_INTERACTIONS
 from onyx.configs.app_configs import PROMPT_CACHE_CHAT_HISTORY
 from onyx.configs.constants import MessageType
@@ -54,7 +55,9 @@ from onyx.server.query_and_chat.streaming_models import ReasoningStart
 from onyx.tools.models import ToolCallKickoff
 from onyx.tracing.framework.create import generation_span
 from onyx.utils.b64 import get_image_type_from_bytes
+from onyx.utils.jsonriver import Parser
 from onyx.utils.logger import setup_logger
+from onyx.utils.postgres_sanitization import sanitize_string
 from onyx.utils.text_processing import find_all_json_objects
 
 logger = setup_logger()
@@ -166,15 +169,6 @@ def _find_function_calls_open_marker(text_lower: str) -> int:
         search_from = idx + 1
 
 
-def _sanitize_llm_output(value: str) -> str:
-    """Remove characters that PostgreSQL's text/JSONB types cannot store.
-
-    - NULL bytes (\x00): Not allowed in PostgreSQL text types
-    - UTF-16 surrogates (\ud800-\udfff): Invalid in UTF-8 encoding
-    """
-    return "".join(c for c in value if c != "\x00" and not ("\ud800" <= c <= "\udfff"))
-
-
 def _try_parse_json_string(value: Any) -> Any:
     """Attempt to parse a JSON string value into its Python equivalent.
 
@@ -222,9 +216,7 @@ def _parse_tool_args_to_dict(raw_args: Any) -> dict[str, Any]:
     if isinstance(raw_args, dict):
         # Parse any string values that look like JSON arrays/objects
         return {
-            k: _try_parse_json_string(
-                _sanitize_llm_output(v) if isinstance(v, str) else v
-            )
+            k: _try_parse_json_string(sanitize_string(v) if isinstance(v, str) else v)
             for k, v in raw_args.items()
         }
 
@@ -232,7 +224,7 @@ def _parse_tool_args_to_dict(raw_args: Any) -> dict[str, Any]:
         return {}
 
     # Sanitize before parsing to remove NULL bytes and surrogates
-    raw_args = _sanitize_llm_output(raw_args)
+    raw_args = sanitize_string(raw_args)
 
     try:
         parsed1: Any = json.loads(raw_args)
@@ -545,12 +537,12 @@ def _extract_xml_attribute(attrs: str, attr_name: str) -> str | None:
     )
     if not attr_match:
         return None
-    return _sanitize_llm_output(unescape(attr_match.group(2).strip()))
+    return sanitize_string(unescape(attr_match.group(2).strip()))
 
 
 def _parse_xml_parameter_value(raw_value: str, string_attr: str | None) -> Any:
     """Parse a parameter value from XML-style tool call payloads."""
-    value = _sanitize_llm_output(unescape(raw_value).strip())
+    value = sanitize_string(unescape(raw_value).strip())
 
     if string_attr and string_attr.lower() == "true":
         return value
@@ -569,6 +561,7 @@ def _resolve_tool_arguments(obj: dict[str, Any]) -> dict[str, Any] | None:
     """
     arguments = obj.get("arguments", obj.get("parameters", {}))
     if isinstance(arguments, str):
+        arguments = sanitize_string(arguments)
         try:
             arguments = json.loads(arguments)
         except json.JSONDecodeError:
@@ -1018,6 +1011,7 @@ def run_llm_step_pkt_generator(
         )
 
     id_to_tool_call_map: dict[int, dict[str, Any]] = {}
+    arg_parsers: dict[int, Parser] = {}
     reasoning_start = False
     answer_start = False
     accumulated_reasoning = ""
@@ -1224,7 +1218,14 @@ def run_llm_step_pkt_generator(
                 yield from _close_reasoning_if_active()
 
                 for tool_call_delta in delta.tool_calls:
+                    # maybe_emit depends and update being called first and attaching the delta
                     _update_tool_call_with_delta(id_to_tool_call_map, tool_call_delta)
+                    yield from maybe_emit_argument_delta(
+                        tool_calls_in_progress=id_to_tool_call_map,
+                        tool_call_delta=tool_call_delta,
+                        placement=_current_placement(),
+                        parsers=arg_parsers,
+                    )
 
         # Flush any tail text buffered while checking for split "<function_calls" markers.
         filtered_content_tail = xml_tool_call_content_filter.flush()

backend/onyx/chat/save_chat.py

@@ -19,6 +19,7 @@ from onyx.natural_language_processing.utils import get_tokenizer
 from onyx.server.query_and_chat.chat_utils import mime_type_to_chat_file_type
 from onyx.tools.models import ToolCallInfo
 from onyx.utils.logger import setup_logger
+from onyx.utils.postgres_sanitization import sanitize_string
 
 logger = setup_logger()
 
@@ -201,8 +202,13 @@ def save_chat_turn(
         pre_answer_processing_time: Duration of processing before answer starts (in seconds)
     """
     # 1. Update ChatMessage with message content, reasoning tokens, and token count
-    assistant_message.message = message_text
-    assistant_message.reasoning_tokens = reasoning_tokens
+    sanitized_message_text = (
+        sanitize_string(message_text) if message_text else message_text
+    )
+    assistant_message.message = sanitized_message_text
+    assistant_message.reasoning_tokens = (
+        sanitize_string(reasoning_tokens) if reasoning_tokens else reasoning_tokens
+    )
     assistant_message.is_clarification = is_clarification
 
     # Use pre-answer processing time (captured when MESSAGE_START was emitted)
@@ -212,8 +218,10 @@ def save_chat_turn(
     # Calculate token count using default tokenizer, when storing, this should not use the LLM
     # specific one so we use a system default tokenizer here.
     default_tokenizer = get_tokenizer(None, None)
-    if message_text:
-        assistant_message.token_count = len(default_tokenizer.encode(message_text))
+    if sanitized_message_text:
+        assistant_message.token_count = len(
+            default_tokenizer.encode(sanitized_message_text)
+        )
     else:
         assistant_message.token_count = 0
 
@@ -328,8 +336,10 @@ def save_chat_turn(
     # 8. Attach code interpreter generated files that the assistant actually
     # referenced in its response, so they are available via load_all_chat_files
     # on subsequent turns. Files not mentioned are intermediate artifacts.
-    if message_text:
-        referenced = _extract_referenced_file_descriptors(tool_calls, message_text)
+    if sanitized_message_text:
+        referenced = _extract_referenced_file_descriptors(
+            tool_calls, sanitized_message_text
+        )
         if referenced:
             existing_files = assistant_message.files or []
             assistant_message.files = existing_files + referenced

backend/onyx/chat/tool_call_args_streaming.py

@@ -0,0 +1,77 @@
+from collections.abc import Generator
+from collections.abc import Mapping
+from typing import Any
+from typing import Type
+
+from onyx.llm.model_response import ChatCompletionDeltaToolCall
+from onyx.server.query_and_chat.placement import Placement
+from onyx.server.query_and_chat.streaming_models import Packet
+from onyx.server.query_and_chat.streaming_models import ToolCallArgumentDelta
+from onyx.tools.built_in_tools import TOOL_NAME_TO_CLASS
+from onyx.tools.interface import Tool
+from onyx.utils.jsonriver import Parser
+
+
+def _get_tool_class(
+    tool_calls_in_progress: Mapping[int, Mapping[str, Any]],
+    tool_call_delta: ChatCompletionDeltaToolCall,
+) -> Type[Tool] | None:
+    """Look up the Tool subclass for a streaming tool call delta."""
+    tool_name = tool_calls_in_progress.get(tool_call_delta.index, {}).get("name")
+    if not tool_name:
+        return None
+    return TOOL_NAME_TO_CLASS.get(tool_name)
+
+
+def maybe_emit_argument_delta(
+    tool_calls_in_progress: Mapping[int, Mapping[str, Any]],
+    tool_call_delta: ChatCompletionDeltaToolCall,
+    placement: Placement,
+    parsers: dict[int, Parser],
+) -> Generator[Packet, None, None]:
+    """Emit decoded tool-call argument deltas to the frontend.
+
+    Uses a ``jsonriver.Parser`` per tool-call index to incrementally parse
+    the JSON argument string and extract only the newly-appended content
+    for each string-valued argument.
+
+    NOTE: Non-string arguments (numbers, booleans, null, arrays, objects)
+    are skipped — they are available in the final tool-call kickoff packet.
+
+    ``parsers`` is a mutable dict keyed by tool-call index. A new
+    ``Parser`` is created automatically for each new index.
+    """
+    tool_cls = _get_tool_class(tool_calls_in_progress, tool_call_delta)
+    if not tool_cls or not tool_cls.should_emit_argument_deltas():
+        return
+
+    fn = tool_call_delta.function
+    delta_fragment = fn.arguments if fn else None
+    if not delta_fragment:
+        return
+
+    idx = tool_call_delta.index
+    if idx not in parsers:
+        parsers[idx] = Parser()
+    parser = parsers[idx]
+
+    deltas = parser.feed(delta_fragment)
+
+    argument_deltas: dict[str, str] = {}
+    for delta in deltas:
+        if isinstance(delta, dict):
+            for key, value in delta.items():
+                if isinstance(value, str):
+                    argument_deltas[key] = argument_deltas.get(key, "") + value
+
+    if not argument_deltas:
+        return
+
+    tc_data = tool_calls_in_progress[tool_call_delta.index]
+    yield Packet(
+        placement=placement,
+        obj=ToolCallArgumentDelta(
+            tool_type=tc_data.get("name", ""),
+            argument_deltas=argument_deltas,
+        ),
+    )

backend/onyx/configs/app_configs.py

@@ -68,6 +68,10 @@ FILE_TOKEN_COUNT_THRESHOLD = int(
     os.environ.get("FILE_TOKEN_COUNT_THRESHOLD", str(_DEFAULT_FILE_TOKEN_LIMIT))
 )
 
+# Maximum upload size for a single user file (chat/projects) in MB.
+USER_FILE_MAX_UPLOAD_SIZE_MB = int(os.environ.get("USER_FILE_MAX_UPLOAD_SIZE_MB") or 50)
+USER_FILE_MAX_UPLOAD_SIZE_BYTES = USER_FILE_MAX_UPLOAD_SIZE_MB * 1024 * 1024
+
 # If set to true, will show extra/uncommon connectors in the "Other" category
 SHOW_EXTRA_CONNECTORS = os.environ.get("SHOW_EXTRA_CONNECTORS", "").lower() == "true"
 

backend/onyx/db/chat.py

@@ -38,6 +38,7 @@ from onyx.llm.override_models import LLMOverride
 from onyx.llm.override_models import PromptOverride
 from onyx.server.query_and_chat.models import ChatMessageDetail
 from onyx.utils.logger import setup_logger
+from onyx.utils.postgres_sanitization import sanitize_string
 
 
 logger = setup_logger()
@@ -675,58 +676,43 @@ def set_as_latest_chat_message(
     db_session.commit()
 
 
-def _sanitize_for_postgres(value: str) -> str:
-    """Remove NUL (0x00) characters from strings as PostgreSQL doesn't allow them."""
-    sanitized = value.replace("\x00", "")
-    if value and not sanitized:
-        logger.warning("Sanitization removed all characters from string")
-    return sanitized
-
-
-def _sanitize_list_for_postgres(values: list[str]) -> list[str]:
-    """Remove NUL (0x00) characters from all strings in a list."""
-    return [_sanitize_for_postgres(v) for v in values]
-
-
 def create_db_search_doc(
     server_search_doc: ServerSearchDoc,
     db_session: Session,
     commit: bool = True,
 ) -> DBSearchDoc:
-    # Sanitize string fields to remove NUL characters (PostgreSQL doesn't allow them)
     db_search_doc = DBSearchDoc(
-        document_id=_sanitize_for_postgres(server_search_doc.document_id),
+        document_id=sanitize_string(server_search_doc.document_id),
         chunk_ind=server_search_doc.chunk_ind,
-        semantic_id=_sanitize_for_postgres(server_search_doc.semantic_identifier),
+        semantic_id=sanitize_string(server_search_doc.semantic_identifier),
         link=(
-            _sanitize_for_postgres(server_search_doc.link)
+            sanitize_string(server_search_doc.link)
             if server_search_doc.link is not None
             else None
         ),
-        blurb=_sanitize_for_postgres(server_search_doc.blurb),
+        blurb=sanitize_string(server_search_doc.blurb),
         source_type=server_search_doc.source_type,
         boost=server_search_doc.boost,
         hidden=server_search_doc.hidden,
         doc_metadata=server_search_doc.metadata,
         is_relevant=server_search_doc.is_relevant,
         relevance_explanation=(
-            _sanitize_for_postgres(server_search_doc.relevance_explanation)
+            sanitize_string(server_search_doc.relevance_explanation)
             if server_search_doc.relevance_explanation is not None
             else None
         ),
-        # For docs further down that aren't reranked, we can't use the retrieval score
         score=server_search_doc.score or 0.0,
-        match_highlights=_sanitize_list_for_postgres(
-            server_search_doc.match_highlights
-        ),
+        match_highlights=[
+            sanitize_string(h) for h in server_search_doc.match_highlights
+        ],
         updated_at=server_search_doc.updated_at,
         primary_owners=(
-            _sanitize_list_for_postgres(server_search_doc.primary_owners)
+            [sanitize_string(o) for o in server_search_doc.primary_owners]
             if server_search_doc.primary_owners is not None
             else None
         ),
         secondary_owners=(
-            _sanitize_list_for_postgres(server_search_doc.secondary_owners)
+            [sanitize_string(o) for o in server_search_doc.secondary_owners]
             if server_search_doc.secondary_owners is not None
             else None
         ),

backend/onyx/db/llm.py

@@ -25,8 +25,11 @@ from onyx.server.manage.embedding.models import CloudEmbeddingProvider
 from onyx.server.manage.embedding.models import CloudEmbeddingProviderCreationRequest
 from onyx.server.manage.llm.models import LLMProviderUpsertRequest
 from onyx.server.manage.llm.models import LLMProviderView
+from onyx.utils.logger import setup_logger
 from shared_configs.enums import EmbeddingProvider
 
+logger = setup_logger()
+
 
 def update_group_llm_provider_relationships__no_commit(
     llm_provider_id: int,
@@ -812,6 +815,43 @@ def sync_auto_mode_models(
             changes += 1
 
     db_session.commit()
+
+    # Update the default if this provider currently holds the global CHAT default
+    recommended_default = llm_recommendations.get_default_model(provider.provider)
+    if recommended_default:
+        current_default_name = db_session.scalar(
+            select(ModelConfiguration.name)
+            .join(
+                LLMModelFlow,
+                LLMModelFlow.model_configuration_id == ModelConfiguration.id,
+            )
+            .where(
+                ModelConfiguration.llm_provider_id == provider.id,
+                LLMModelFlow.llm_model_flow_type == LLMModelFlowType.CHAT,
+                LLMModelFlow.is_default == True,  # noqa: E712
+            )
+        )
+
+        if (
+            current_default_name is not None
+            and current_default_name != recommended_default.name
+        ):
+            try:
+                _update_default_model(
+                    db_session=db_session,
+                    provider_id=provider.id,
+                    model=recommended_default.name,
+                    flow_type=LLMModelFlowType.CHAT,
+                )
+                changes += 1
+            except ValueError:
+                logger.warning(
+                    "Recommended default model '%s' not found "
+                    "for provider_id=%s; skipping default update.",
+                    recommended_default.name,
+                    provider.id,
+                )
+
     return changes
 
 

backend/onyx/db/models.py

@@ -36,9 +36,11 @@ from sqlalchemy import Text
 from sqlalchemy import text
 from sqlalchemy import UniqueConstraint
 from sqlalchemy.dialects import postgresql
+from sqlalchemy import event
 from sqlalchemy.engine.interfaces import Dialect
 from sqlalchemy.orm import DeclarativeBase
 from sqlalchemy.orm import Mapped
+from sqlalchemy.orm import Mapper
 from sqlalchemy.orm import mapped_column
 from sqlalchemy.orm import relationship
 from sqlalchemy.types import LargeBinary
@@ -117,10 +119,50 @@ class Base(DeclarativeBase):
     __abstract__ = True
 
 
-class EncryptedString(TypeDecorator):
+class _EncryptedBase(TypeDecorator):
+    """Base for encrypted column types that wrap values in SensitiveValue."""
+
     impl = LargeBinary
-    # This type's behavior is fully deterministic and doesn't depend on any external factors.
     cache_ok = True
+    _is_json: bool = False
+
+    def wrap_raw(self, value: Any) -> SensitiveValue:
+        """Encrypt a raw value and wrap it in SensitiveValue.
+
+        Called by the attribute set event so the Python-side type is always
+        SensitiveValue, regardless of whether the value was loaded from the DB
+        or assigned in application code.
+        """
+        if self._is_json:
+            if not isinstance(value, dict):
+                raise TypeError(
+                    f"EncryptedJson column expected dict, got {type(value).__name__}"
+                )
+            raw_str = json.dumps(value)
+        else:
+            if not isinstance(value, str):
+                raise TypeError(
+                    f"EncryptedString column expected str, got {type(value).__name__}"
+                )
+            raw_str = value
+        return SensitiveValue(
+            encrypted_bytes=encrypt_string_to_bytes(raw_str),
+            decrypt_fn=decrypt_bytes_to_string,
+            is_json=self._is_json,
+        )
+
+    def compare_values(self, x: Any, y: Any) -> bool:
+        if x is None or y is None:
+            return x == y
+        if isinstance(x, SensitiveValue):
+            x = x.get_value(apply_mask=False)
+        if isinstance(y, SensitiveValue):
+            y = y.get_value(apply_mask=False)
+        return x == y
+
+
+class EncryptedString(_EncryptedBase):
+    _is_json: bool = False
 
     def process_bind_param(
         self, value: str | SensitiveValue[str] | None, dialect: Dialect  # noqa: ARG002
@@ -144,20 +186,9 @@ class EncryptedString(TypeDecorator):
             )
         return None
 
-    def compare_values(self, x: Any, y: Any) -> bool:
-        if x is None or y is None:
-            return x == y
-        if isinstance(x, SensitiveValue):
-            x = x.get_value(apply_mask=False)
-        if isinstance(y, SensitiveValue):
-            y = y.get_value(apply_mask=False)
-        return x == y
 
-
-class EncryptedJson(TypeDecorator):
-    impl = LargeBinary
-    # This type's behavior is fully deterministic and doesn't depend on any external factors.
-    cache_ok = True
+class EncryptedJson(_EncryptedBase):
+    _is_json: bool = True
 
     def process_bind_param(
         self,
@@ -165,9 +196,7 @@ class EncryptedJson(TypeDecorator):
         dialect: Dialect,  # noqa: ARG002
     ) -> bytes | None:
         if value is not None:
-            # Handle both raw dicts and SensitiveValue wrappers
             if isinstance(value, SensitiveValue):
-                # Get raw value for storage
                 value = value.get_value(apply_mask=False)
             json_str = json.dumps(value)
             return encrypt_string_to_bytes(json_str)
@@ -184,14 +213,40 @@ class EncryptedJson(TypeDecorator):
             )
         return None
 
-    def compare_values(self, x: Any, y: Any) -> bool:
-        if x is None or y is None:
-            return x == y
-        if isinstance(x, SensitiveValue):
-            x = x.get_value(apply_mask=False)
-        if isinstance(y, SensitiveValue):
-            y = y.get_value(apply_mask=False)
-        return x == y
+
+_REGISTERED_ATTRS: set[str] = set()
+
+
+@event.listens_for(Mapper, "mapper_configured")
+def _register_sensitive_value_set_events(
+    mapper: Mapper,
+    class_: type,
+) -> None:
+    """Auto-wrap raw values in SensitiveValue when assigned to encrypted columns."""
+    for prop in mapper.column_attrs:
+        for col in prop.columns:
+            if isinstance(col.type, _EncryptedBase):
+                col_type = col.type
+                attr = getattr(class_, prop.key)
+
+                # Guard against double-registration (e.g. if mapper is
+                # re-configured in test setups)
+                attr_key = f"{class_.__qualname__}.{prop.key}"
+                if attr_key in _REGISTERED_ATTRS:
+                    continue
+                _REGISTERED_ATTRS.add(attr_key)
+
+                @event.listens_for(attr, "set", retval=True)
+                def _wrap_value(
+                    target: Any,  # noqa: ARG001
+                    value: Any,
+                    oldvalue: Any,  # noqa: ARG001
+                    initiator: Any,  # noqa: ARG001
+                    _col_type: _EncryptedBase = col_type,
+                ) -> Any:
+                    if value is not None and not isinstance(value, SensitiveValue):
+                        return _col_type.wrap_raw(value)
+                    return value
 
 
 class NullFilteredString(TypeDecorator):

backend/onyx/db/rotate_encryption_key.py

@@ -0,0 +1,161 @@
+"""Rotate encryption key for all encrypted columns.
+
+Dynamically discovers all columns using EncryptedString / EncryptedJson,
+decrypts each value with the old key, and re-encrypts with the current
+ENCRYPTION_KEY_SECRET.
+
+The operation is idempotent: rows already encrypted with the current key
+are skipped. Commits are made in batches so a crash mid-rotation can be
+safely resumed by re-running.
+"""
+
+import json
+from typing import Any
+
+from sqlalchemy import LargeBinary
+from sqlalchemy import select
+from sqlalchemy import update
+from sqlalchemy.orm import Session
+
+from onyx.configs.app_configs import ENCRYPTION_KEY_SECRET
+from onyx.db.models import Base
+from onyx.db.models import EncryptedJson
+from onyx.db.models import EncryptedString
+from onyx.utils.encryption import decrypt_bytes_to_string
+from onyx.utils.logger import setup_logger
+from onyx.utils.variable_functionality import global_version
+
+logger = setup_logger()
+
+_BATCH_SIZE = 500
+
+
+def _can_decrypt_with_current_key(data: bytes) -> bool:
+    """Check if data is already encrypted with the current key.
+
+    Passes the key explicitly so the fallback-to-raw-decode path in
+    _decrypt_bytes is NOT triggered — a clean success/failure signal.
+    """
+    try:
+        decrypt_bytes_to_string(data, key=ENCRYPTION_KEY_SECRET)
+        return True
+    except Exception:
+        return False
+
+
+def _discover_encrypted_columns() -> list[tuple[type, str, list[str], bool]]:
+    """Walk all ORM models and find columns using EncryptedString/EncryptedJson.
+
+    Returns list of (ModelClass, column_attr_name, [pk_attr_names], is_json).
+    """
+    results: list[tuple[type, str, list[str], bool]] = []
+
+    for mapper in Base.registry.mappers:
+        model_cls = mapper.class_
+        pk_names = [col.key for col in mapper.primary_key]
+
+        for prop in mapper.column_attrs:
+            for col in prop.columns:
+                if isinstance(col.type, EncryptedJson):
+                    results.append((model_cls, prop.key, pk_names, True))
+                elif isinstance(col.type, EncryptedString):
+                    results.append((model_cls, prop.key, pk_names, False))
+
+    return results
+
+
+def rotate_encryption_key(
+    db_session: Session,
+    old_key: str | None,
+    dry_run: bool = False,
+) -> dict[str, int]:
+    """Decrypt all encrypted columns with old_key and re-encrypt with the current key.
+
+    Args:
+        db_session: Active database session.
+        old_key: The previous encryption key. Pass None or "" if values were
+                 not previously encrypted with a key.
+        dry_run: If True, count rows that need rotation without modifying data.
+
+    Returns:
+        Dict of "table.column" -> number of rows re-encrypted (or would be).
+
+    Commits every _BATCH_SIZE rows so that locks are held briefly and progress
+    is preserved on crash. Already-rotated rows are detected and skipped,
+    making the operation safe to re-run.
+    """
+    if not global_version.is_ee_version():
+        raise RuntimeError("EE mode is not enabled — rotation requires EE encryption.")
+
+    if not ENCRYPTION_KEY_SECRET:
+        raise RuntimeError(
+            "ENCRYPTION_KEY_SECRET is not set — cannot rotate. "
+            "Set the target encryption key in the environment before running."
+        )
+
+    encrypted_columns = _discover_encrypted_columns()
+    totals: dict[str, int] = {}
+
+    for model_cls, col_name, pk_names, is_json in encrypted_columns:
+        table_name: str = model_cls.__tablename__  # type: ignore[attr-defined]
+        col_attr = getattr(model_cls, col_name)
+        pk_attrs = [getattr(model_cls, pk) for pk in pk_names]
+
+        # Read raw bytes directly, bypassing the TypeDecorator
+        raw_col = col_attr.property.columns[0]
+
+        stmt = select(*pk_attrs, raw_col.cast(LargeBinary)).where(col_attr.is_not(None))
+        rows = db_session.execute(stmt).all()
+
+        reencrypted = 0
+        batch_pending = 0
+        for row in rows:
+            raw_bytes: bytes | None = row[-1]
+            if raw_bytes is None:
+                continue
+
+            if _can_decrypt_with_current_key(raw_bytes):
+                continue
+
+            try:
+                if not old_key:
+                    decrypted_str = raw_bytes.decode("utf-8")
+                else:
+                    decrypted_str = decrypt_bytes_to_string(raw_bytes, key=old_key)
+
+                # For EncryptedJson, parse back to dict so the TypeDecorator
+                # can json.dumps() it cleanly (avoids double-encoding).
+                value: Any = json.loads(decrypted_str) if is_json else decrypted_str
+            except (ValueError, UnicodeDecodeError) as e:
+                pk_vals = [row[i] for i in range(len(pk_names))]
+                logger.warning(
+                    f"Could not decrypt/parse {table_name}.{col_name} "
+                    f"row {pk_vals} — skipping: {e}"
+                )
+                continue
+
+            if not dry_run:
+                pk_filters = [pk_attr == row[i] for i, pk_attr in enumerate(pk_attrs)]
+                update_stmt = (
+                    update(model_cls).where(*pk_filters).values({col_name: value})
+                )
+                db_session.execute(update_stmt)
+                batch_pending += 1
+
+                if batch_pending >= _BATCH_SIZE:
+                    db_session.commit()
+                    batch_pending = 0
+            reencrypted += 1
+
+        # Flush remaining rows in this column
+        if batch_pending > 0:
+            db_session.commit()
+
+        if reencrypted > 0:
+            totals[f"{table_name}.{col_name}"] = reencrypted
+            logger.info(
+                f"{'[DRY RUN] Would re-encrypt' if dry_run else 'Re-encrypted'} "
+                f"{reencrypted} value(s) in {table_name}.{col_name}"
+            )
+
+    return totals

backend/onyx/db/tools.py

@@ -13,12 +13,15 @@ from onyx.db.constants import UNSET
 from onyx.db.constants import UnsetType
 from onyx.db.enums import MCPServerStatus
 from onyx.db.models import MCPServer
+from onyx.db.models import OAuthConfig
 from onyx.db.models import Tool
 from onyx.db.models import ToolCall
 from onyx.server.features.tool.models import Header
 from onyx.tools.built_in_tools import BUILT_IN_TOOL_TYPES
 from onyx.utils.headers import HeaderItemDict
 from onyx.utils.logger import setup_logger
+from onyx.utils.postgres_sanitization import sanitize_json_like
+from onyx.utils.postgres_sanitization import sanitize_string
 
 if TYPE_CHECKING:
     pass
@@ -159,10 +162,26 @@ def update_tool(
         ]
     if passthrough_auth is not None:
         tool.passthrough_auth = passthrough_auth
+    old_oauth_config_id = tool.oauth_config_id
     if not isinstance(oauth_config_id, UnsetType):
         tool.oauth_config_id = oauth_config_id
-    db_session.commit()
+        db_session.flush()
 
+    # Clean up orphaned OAuthConfig if the oauth_config_id was changed
+    if (
+        old_oauth_config_id is not None
+        and not isinstance(oauth_config_id, UnsetType)
+        and old_oauth_config_id != oauth_config_id
+    ):
+        other_tools = db_session.scalars(
+            select(Tool).where(Tool.oauth_config_id == old_oauth_config_id)
+        ).all()
+        if not other_tools:
+            oauth_config = db_session.get(OAuthConfig, old_oauth_config_id)
+            if oauth_config:
+                db_session.delete(oauth_config)
+
+    db_session.commit()
     return tool
 
 
@@ -171,8 +190,21 @@ def delete_tool__no_commit(tool_id: int, db_session: Session) -> None:
     if tool is None:
         raise ValueError(f"Tool with ID {tool_id} does not exist")
 
+    oauth_config_id = tool.oauth_config_id
+
     db_session.delete(tool)
-    db_session.flush()  # Don't commit yet, let caller decide when to commit
+    db_session.flush()
+
+    # Clean up orphaned OAuthConfig if no other tools reference it
+    if oauth_config_id is not None:
+        other_tools = db_session.scalars(
+            select(Tool).where(Tool.oauth_config_id == oauth_config_id)
+        ).all()
+        if not other_tools:
+            oauth_config = db_session.get(OAuthConfig, oauth_config_id)
+            if oauth_config:
+                db_session.delete(oauth_config)
+                db_session.flush()
 
 
 def get_builtin_tool(
@@ -256,11 +288,13 @@ def create_tool_call_no_commit(
         tab_index=tab_index,
         tool_id=tool_id,
         tool_call_id=tool_call_id,
-        reasoning_tokens=reasoning_tokens,
-        tool_call_arguments=tool_call_arguments,
-        tool_call_response=tool_call_response,
+        reasoning_tokens=(
+            sanitize_string(reasoning_tokens) if reasoning_tokens else reasoning_tokens
+        ),
+        tool_call_arguments=sanitize_json_like(tool_call_arguments),
+        tool_call_response=sanitize_json_like(tool_call_response),
         tool_call_tokens=tool_call_tokens,
-        generated_images=generated_images,
+        generated_images=sanitize_json_like(generated_images),
     )
 
     db_session.add(tool_call)

backend/onyx/document_index/FILTER_SEMANTICS.md

@@ -0,0 +1,103 @@
+# Vector DB Filter Semantics
+
+How `IndexFilters` fields combine into the final query filter. Applies to both Vespa and OpenSearch.
+
+## Filter categories
+
+| Category | Fields | Join logic |
+|---|---|---|
+| **Visibility** | `hidden` | Always applied (unless `include_hidden`) |
+| **Tenant** | `tenant_id` | AND (multi-tenant only) |
+| **ACL** | `access_control_list` | OR within, AND with rest |
+| **Narrowing** | `source_type`, `tags`, `time_cutoff` | Each OR within, AND with rest |
+| **Knowledge scope** | `document_set`, `user_file_ids`, `attached_document_ids`, `hierarchy_node_ids` | OR within group, AND with rest |
+| **Additive scope** | `project_id`, `persona_id` | OR'd into knowledge scope **only when** a knowledge scope filter already exists |
+
+## How filters combine
+
+All categories are AND'd together. Within the knowledge scope category, individual filters are OR'd.
+
+```
+NOT hidden
+AND tenant = T                          -- if multi-tenant
+AND (acl contains A1 OR acl contains A2)
+AND (source_type = S1 OR ...)           -- if set
+AND (tag = T1 OR ...)                   -- if set
+AND <knowledge scope>                   -- see below
+AND time >= cutoff                      -- if set
+```
+
+## Knowledge scope rules
+
+The knowledge scope filter controls **what knowledge an assistant can access**.
+
+### No explicit knowledge attached
+
+When `document_set`, `user_file_ids`, `attached_document_ids`, and `hierarchy_node_ids` are all empty/None:
+
+- **No knowledge scope filter is applied.** The assistant can see everything (subject to ACL).
+- `project_id` and `persona_id` are ignored — they never restrict on their own.
+
+### One explicit knowledge type
+
+```
+-- Only document sets
+AND (document_sets contains "Engineering" OR document_sets contains "Legal")
+
+-- Only user files
+AND (document_id = "uuid-1" OR document_id = "uuid-2")
+```
+
+### Multiple explicit knowledge types (OR'd)
+
+```
+-- Document sets + user files
+AND (
+    document_sets contains "Engineering"
+    OR document_id = "uuid-1"
+)
+```
+
+### Explicit knowledge + overflowing user files
+
+When an explicit knowledge restriction is in effect **and** `project_id` or `persona_id` is set (user files overflowed the LLM context window), the additive scopes widen the filter:
+
+```
+-- Document sets + persona user files overflowed
+AND (
+    document_sets contains "Engineering"
+    OR personas contains 42
+)
+
+-- User files + project files overflowed
+AND (
+    document_id = "uuid-1"
+    OR user_project contains 7
+)
+```
+
+### Only project_id or persona_id (no explicit knowledge)
+
+No knowledge scope filter. The assistant searches everything.
+
+```
+-- Just ACL, no restriction
+NOT hidden
+AND (acl contains ...)
+```
+
+## Field reference
+
+| Filter field | Vespa field | Vespa type | Purpose |
+|---|---|---|---|
+| `document_set` | `document_sets` | `weightedset<string>` | Connector doc sets attached to assistant |
+| `user_file_ids` | `document_id` | `string` | User files uploaded to assistant |
+| `attached_document_ids` | `document_id` | `string` | Documents explicitly attached (OpenSearch only) |
+| `hierarchy_node_ids` | `ancestor_hierarchy_node_ids` | `array<int>` | Folder/space nodes (OpenSearch only) |
+| `project_id` | `user_project` | `array<int>` | Project tag for overflowing user files |
+| `persona_id` | `personas` | `array<int>` | Persona tag for overflowing user files |
+| `access_control_list` | `access_control_list` | `weightedset<string>` | ACL entries for the requesting user |
+| `source_type` | `source_type` | `string` | Connector source type (e.g. `web`, `jira`) |
+| `tags` | `metadata_list` | `array<string>` | Document metadata tags |
+| `time_cutoff` | `doc_updated_at` | `long` | Minimum document update timestamp |
+| `tenant_id` | `tenant_id` | `string` | Tenant isolation (multi-tenant) |

backend/onyx/document_index/opensearch/client.py

@@ -61,6 +61,25 @@ class SearchHit(BaseModel, Generic[SchemaDocumentModel]):
     explanation: dict[str, Any] | None = None
 
 
+class IndexInfo(BaseModel):
+    """
+    Represents information about an OpenSearch index.
+    """
+
+    model_config = {"frozen": True}
+
+    name: str
+    health: str
+    status: str
+    num_primary_shards: str
+    num_replica_shards: str
+    docs_count: str
+    docs_deleted: str
+    created_at: str
+    total_size: str
+    primary_shards_size: str
+
+
 def get_new_body_without_vectors(body: dict[str, Any]) -> dict[str, Any]:
     """Recursively replaces vectors in the body with their length.
 
@@ -159,8 +178,8 @@ class OpenSearchClient(AbstractContextManager):
         Raises:
             Exception: There was an error creating the search pipeline.
         """
-        result = self._client.search_pipeline.put(id=pipeline_id, body=pipeline_body)
-        if not result.get("acknowledged", False):
+        response = self._client.search_pipeline.put(id=pipeline_id, body=pipeline_body)
+        if not response.get("acknowledged", False):
             raise RuntimeError(f"Failed to create search pipeline {pipeline_id}.")
 
     @log_function_time(print_only=True, debug_only=True, include_args=True)
@@ -173,8 +192,8 @@ class OpenSearchClient(AbstractContextManager):
         Raises:
             Exception: There was an error deleting the search pipeline.
         """
-        result = self._client.search_pipeline.delete(id=pipeline_id)
-        if not result.get("acknowledged", False):
+        response = self._client.search_pipeline.delete(id=pipeline_id)
+        if not response.get("acknowledged", False):
             raise RuntimeError(f"Failed to delete search pipeline {pipeline_id}.")
 
     @log_function_time(print_only=True, debug_only=True, include_args=True)
@@ -198,6 +217,34 @@ class OpenSearchClient(AbstractContextManager):
             logger.error(f"Failed to put cluster settings: {response}.")
             return False
 
+    @log_function_time(print_only=True, debug_only=True)
+    def list_indices_with_info(self) -> list[IndexInfo]:
+        """
+        Lists the indices in the OpenSearch cluster with information about each
+        index.
+
+        Returns:
+            A list of IndexInfo objects for each index.
+        """
+        response = self._client.cat.indices(format="json")
+        indices: list[IndexInfo] = []
+        for raw_index_info in response:
+            indices.append(
+                IndexInfo(
+                    name=raw_index_info.get("index", ""),
+                    health=raw_index_info.get("health", ""),
+                    status=raw_index_info.get("status", ""),
+                    num_primary_shards=raw_index_info.get("pri", ""),
+                    num_replica_shards=raw_index_info.get("rep", ""),
+                    docs_count=raw_index_info.get("docs.count", ""),
+                    docs_deleted=raw_index_info.get("docs.deleted", ""),
+                    created_at=raw_index_info.get("creation.date.string", ""),
+                    total_size=raw_index_info.get("store.size", ""),
+                    primary_shards_size=raw_index_info.get("pri.store.size", ""),
+                )
+            )
+        return indices
+
     @log_function_time(print_only=True, debug_only=True)
     def ping(self) -> bool:
         """Pings the OpenSearch cluster.

backend/onyx/document_index/opensearch/opensearch_document_index.py

@@ -739,7 +739,8 @@ class OpenSearchDocumentIndex(DocumentIndex):
             The number of chunks successfully deleted.
         """
         logger.debug(
-            f"[OpenSearchDocumentIndex] Deleting document {document_id} from index {self._index_name}."
+            f"[OpenSearchDocumentIndex] Deleting document {document_id} from index "
+            f"{self._index_name}."
         )
         query_body = DocumentQuery.delete_from_document_id_query(
             document_id=document_id,
@@ -775,7 +776,8 @@ class OpenSearchDocumentIndex(DocumentIndex):
                 specified documents.
         """
         logger.debug(
-            f"[OpenSearchDocumentIndex] Updating {len(update_requests)} chunks for index {self._index_name}."
+            f"[OpenSearchDocumentIndex] Updating {len(update_requests)} chunks for index "
+            f"{self._index_name}."
         )
         for update_request in update_requests:
             properties_to_update: dict[str, Any] = dict()
@@ -831,9 +833,11 @@ class OpenSearchDocumentIndex(DocumentIndex):
                     # here.
                     # TODO(andrei): Fix the aforementioned race condition.
                     raise ChunkCountNotFoundError(
-                        f"Tried to update document {doc_id} but its chunk count is not known. Older versions of the "
-                        "application used to permit this but is not a supported state for a document when using OpenSearch. "
-                        "The document was likely just added to the indexing pipeline and the chunk count will be updated shortly."
+                        f"Tried to update document {doc_id} but its chunk count is not known. "
+                        "Older versions of the application used to permit this but is not a "
+                        "supported state for a document when using OpenSearch. The document was "
+                        "likely just added to the indexing pipeline and the chunk count will be "
+                        "updated shortly."
                     )
                 if doc_chunk_count == 0:
                     raise ValueError(
@@ -865,7 +869,8 @@ class OpenSearchDocumentIndex(DocumentIndex):
         chunk IDs vs querying for matching document chunks.
         """
         logger.debug(
-            f"[OpenSearchDocumentIndex] Retrieving {len(chunk_requests)} chunks for index {self._index_name}."
+            f"[OpenSearchDocumentIndex] Retrieving {len(chunk_requests)} chunks for index "
+            f"{self._index_name}."
         )
         results: list[InferenceChunk] = []
         for chunk_request in chunk_requests:
@@ -912,7 +917,8 @@ class OpenSearchDocumentIndex(DocumentIndex):
         num_to_retrieve: int,
     ) -> list[InferenceChunk]:
         logger.debug(
-            f"[OpenSearchDocumentIndex] Hybrid retrieving {num_to_retrieve} chunks for index {self._index_name}."
+            f"[OpenSearchDocumentIndex] Hybrid retrieving {num_to_retrieve} chunks for index "
+            f"{self._index_name}."
         )
         # TODO(andrei): This could be better, the caller should just make this
         # decision when passing in the query param. See the above comment in the
@@ -932,8 +938,10 @@ class OpenSearchDocumentIndex(DocumentIndex):
             index_filters=filters,
             include_hidden=False,
         )
-        # NOTE: Using z-score normalization here because it's better for hybrid search from a theoretical standpoint.
-        # Empirically on a small dataset of up to 10K docs, it's not very different. Likely more impactful at scale.
+        # NOTE: Using z-score normalization here because it's better for hybrid
+        # search from a theoretical standpoint. Empirically on a small dataset
+        # of up to 10K docs, it's not very different. Likely more impactful at
+        # scale.
         # https://opensearch.org/blog/introducing-the-z-score-normalization-technique-for-hybrid-search/
         search_hits: list[SearchHit[DocumentChunk]] = self._client.search(
             body=query_body,
@@ -960,7 +968,8 @@ class OpenSearchDocumentIndex(DocumentIndex):
         dirty: bool | None = None,  # noqa: ARG002
     ) -> list[InferenceChunk]:
         logger.debug(
-            f"[OpenSearchDocumentIndex] Randomly retrieving {num_to_retrieve} chunks for index {self._index_name}."
+            f"[OpenSearchDocumentIndex] Randomly retrieving {num_to_retrieve} chunks for index "
+            f"{self._index_name}."
         )
         query_body = DocumentQuery.get_random_search_query(
             tenant_state=self._tenant_state,
@@ -990,7 +999,8 @@ class OpenSearchDocumentIndex(DocumentIndex):
         complete.
         """
         logger.debug(
-            f"[OpenSearchDocumentIndex] Indexing {len(chunks)} raw chunks for index {self._index_name}."
+            f"[OpenSearchDocumentIndex] Indexing {len(chunks)} raw chunks for index "
+            f"{self._index_name}."
         )
         # Do not raise if the document already exists, just update. This is
         # because the document may already have been indexed during the

backend/onyx/document_index/opensearch/schema.py

@@ -243,7 +243,8 @@ class DocumentChunk(BaseModel):
             return value
         if not isinstance(value, int):
             raise ValueError(
-                f"Bug: Expected an int for the last_updated property from OpenSearch, got {type(value)} instead."
+                f"Bug: Expected an int for the last_updated property from OpenSearch, got "
+                f"{type(value)} instead."
             )
         return datetime.fromtimestamp(value, tz=timezone.utc)
 
@@ -284,19 +285,22 @@ class DocumentChunk(BaseModel):
         elif isinstance(value, TenantState):
             if MULTI_TENANT != value.multitenant:
                 raise ValueError(
-                    f"Bug: An existing TenantState object was supplied to the DocumentChunk model but its multi-tenant mode "
-                    f"({value.multitenant}) does not match the program's current global tenancy state."
+                    f"Bug: An existing TenantState object was supplied to the DocumentChunk model "
+                    f"but its multi-tenant mode ({value.multitenant}) does not match the program's "
+                    "current global tenancy state."
                 )
             return value
         elif not isinstance(value, str):
             raise ValueError(
-                f"Bug: Expected a str for the tenant_id property from OpenSearch, got {type(value)} instead."
+                f"Bug: Expected a str for the tenant_id property from OpenSearch, got "
+                f"{type(value)} instead."
             )
         else:
             if not MULTI_TENANT:
                 raise ValueError(
-                    "Bug: Got a non-null str for the tenant_id property from OpenSearch but multi-tenant mode is not enabled. "
-                    "This is unexpected because in single-tenant mode we don't expect to see a tenant_id."
+                    "Bug: Got a non-null str for the tenant_id property from OpenSearch but "
+                    "multi-tenant mode is not enabled. This is unexpected because in single-tenant "
+                    "mode we don't expect to see a tenant_id."
                 )
             return TenantState(tenant_id=value, multitenant=MULTI_TENANT)
 
@@ -352,8 +356,10 @@ class DocumentSchema:
             "properties": {
                 TITLE_FIELD_NAME: {
                     "type": "text",
-                    # Language analyzer (e.g. english) stems at index and search time for variant matching.
-                    # Configure via OPENSEARCH_TEXT_ANALYZER. Existing indices need reindexing after a change.
+                    # Language analyzer (e.g. english) stems at index and search
+                    # time for variant matching. Configure via
+                    # OPENSEARCH_TEXT_ANALYZER. Existing indices need reindexing
+                    # after a change.
                     "analyzer": OPENSEARCH_TEXT_ANALYZER,
                     "fields": {
                         # Subfield accessed as title.keyword. Not indexed for

backend/onyx/document_index/opensearch/search.py

@@ -698,41 +698,6 @@ class DocumentQuery:
             """
             return {"terms": {ANCESTOR_HIERARCHY_NODE_IDS_FIELD_NAME: node_ids}}
 
-        def _get_assistant_knowledge_filter(
-            attached_doc_ids: list[str] | None,
-            node_ids: list[int] | None,
-            file_ids: list[UUID] | None,
-            document_sets: list[str] | None,
-        ) -> dict[str, Any]:
-            """Combined filter for assistant knowledge.
-
-            When an assistant has attached knowledge, search should be scoped to:
-            - Documents explicitly attached (by document ID), OR
-            - Documents under attached hierarchy nodes (by ancestor node IDs), OR
-            - User-uploaded files attached to the assistant, OR
-            - Documents in the assistant's document sets (if any)
-            """
-            knowledge_filter: dict[str, Any] = {
-                "bool": {"should": [], "minimum_should_match": 1}
-            }
-            if attached_doc_ids:
-                knowledge_filter["bool"]["should"].append(
-                    _get_attached_document_id_filter(attached_doc_ids)
-                )
-            if node_ids:
-                knowledge_filter["bool"]["should"].append(
-                    _get_hierarchy_node_filter(node_ids)
-                )
-            if file_ids:
-                knowledge_filter["bool"]["should"].append(
-                    _get_user_file_id_filter(file_ids)
-                )
-            if document_sets:
-                knowledge_filter["bool"]["should"].append(
-                    _get_document_set_filter(document_sets)
-                )
-            return knowledge_filter
-
         filter_clauses: list[dict[str, Any]] = []
 
         if not include_hidden:
@@ -758,41 +723,53 @@ class DocumentQuery:
             # document's metadata list.
             filter_clauses.append(_get_tag_filter(tags))
 
-        # Check if this is an assistant knowledge search (has any assistant-scoped knowledge)
-        has_assistant_knowledge = (
+        # Knowledge scope: explicit knowledge attachments restrict what
+        # an assistant can see.  When none are set the assistant
+        # searches everything.
+        #
+        # project_id / persona_id are additive: they make overflowing
+        # user files findable but must NOT trigger the restriction on
+        # their own (an agent with no explicit knowledge should search
+        # everything).
+        has_knowledge_scope = (
             attached_document_ids
             or hierarchy_node_ids
             or user_file_ids
             or document_sets
         )
 
-        if has_assistant_knowledge:
-            # If assistant has attached knowledge, scope search to that knowledge.
-            # Document sets are included in the OR filter so directly attached
-            # docs are always findable even if not in the document sets.
-            filter_clauses.append(
-                _get_assistant_knowledge_filter(
-                    attached_document_ids,
-                    hierarchy_node_ids,
-                    user_file_ids,
-                    document_sets,
+        if has_knowledge_scope:
+            knowledge_filter: dict[str, Any] = {
+                "bool": {"should": [], "minimum_should_match": 1}
+            }
+            if attached_document_ids:
+                knowledge_filter["bool"]["should"].append(
+                    _get_attached_document_id_filter(attached_document_ids)
                 )
-            )
-        elif user_file_ids:
-            # Fallback for non-assistant user file searches (e.g., project searches)
-            # If at least one user file ID is provided, the caller will only
-            # retrieve documents where the document ID is in this input list of
-            # file IDs.
-            filter_clauses.append(_get_user_file_id_filter(user_file_ids))
-
-        if project_id is not None:
-            # If a project ID is provided, the caller will only retrieve
-            # documents where the project ID provided here is present in the
-            # document's user projects list.
-            filter_clauses.append(_get_user_project_filter(project_id))
-
-        if persona_id is not None:
-            filter_clauses.append(_get_persona_filter(persona_id))
+            if hierarchy_node_ids:
+                knowledge_filter["bool"]["should"].append(
+                    _get_hierarchy_node_filter(hierarchy_node_ids)
+                )
+            if user_file_ids:
+                knowledge_filter["bool"]["should"].append(
+                    _get_user_file_id_filter(user_file_ids)
+                )
+            if document_sets:
+                knowledge_filter["bool"]["should"].append(
+                    _get_document_set_filter(document_sets)
+                )
+            # Additive: widen scope to also cover overflowing user
+            # files, but only when an explicit restriction is already
+            # in effect.
+            if project_id is not None:
+                knowledge_filter["bool"]["should"].append(
+                    _get_user_project_filter(project_id)
+                )
+            if persona_id is not None:
+                knowledge_filter["bool"]["should"].append(
+                    _get_persona_filter(persona_id)
+                )
+            filter_clauses.append(knowledge_filter)
 
         if time_cutoff is not None:
             # If a time cutoff is provided, the caller will only retrieve

backend/onyx/document_index/vespa/shared_utils/vespa_request_builders.py

@@ -23,11 +23,8 @@ from shared_configs.configs import MULTI_TENANT
 logger = setup_logger()
 
 
-def build_tenant_id_filter(tenant_id: str, include_trailing_and: bool = False) -> str:
-    filter_str = f'({TENANT_ID} contains "{tenant_id}")'
-    if include_trailing_and:
-        filter_str += " and "
-    return filter_str
+def build_tenant_id_filter(tenant_id: str) -> str:
+    return f'({TENANT_ID} contains "{tenant_id}")'
 
 
 def build_vespa_filters(
@@ -37,30 +34,22 @@ def build_vespa_filters(
     remove_trailing_and: bool = False,  # Set to True when using as a complete Vespa query
 ) -> str:
     def _build_or_filters(key: str, vals: list[str] | None) -> str:
-        """For string-based 'contains' filters, e.g. WSET fields or array<string> fields."""
+        """For string-based 'contains' filters, e.g. WSET fields or array<string> fields.
+        Returns a bare clause like '(key contains "v1" or key contains "v2")' or ""."""
         if not key or not vals:
             return ""
         eq_elems = [f'{key} contains "{val}"' for val in vals if val]
         if not eq_elems:
             return ""
-        or_clause = " or ".join(eq_elems)
-        return f"({or_clause}) and "
+        return f"({' or '.join(eq_elems)})"
 
     def _build_int_or_filters(key: str, vals: list[int] | None) -> str:
-        """
-        For an integer field filter.
-        If vals is not None, we want *only* docs whose key matches one of vals.
-        """
-        # If `vals` is None => skip the filter entirely
+        """For an integer field filter.
+        Returns a bare clause or ""."""
         if vals is None or not vals:
             return ""
-
-        # Otherwise build the OR filter
         eq_elems = [f"{key} = {val}" for val in vals]
-        or_clause = " or ".join(eq_elems)
-        result = f"({or_clause}) and "
-
-        return result
+        return f"({' or '.join(eq_elems)})"
 
     def _build_kg_filter(
         kg_entities: list[str] | None,
@@ -73,16 +62,12 @@ def build_vespa_filters(
         combined_filter_parts = []
 
         def _build_kge(entity: str) -> str:
-            # TYPE-SUBTYPE::ID -> "TYPE-SUBTYPE::ID"
-            # TYPE-SUBTYPE::*  -> ({prefix: true}"TYPE-SUBTYPE")
-            # TYPE::*          -> ({prefix: true}"TYPE")
             GENERAL = "::*"
             if entity.endswith(GENERAL):
                 return f'({{prefix: true}}"{entity.split(GENERAL, 1)[0]}")'
             else:
                 return f'"{entity}"'
 
-        # OR the entities (give new design)
         if kg_entities:
             filter_parts = []
             for kg_entity in kg_entities:
@@ -104,8 +89,7 @@ def build_vespa_filters(
 
         # TODO: remove kg terms entirely from prompts and codebase
 
-        # AND the combined filter parts
-        return f"({' and '.join(combined_filter_parts)}) and "
+        return f"({' and '.join(combined_filter_parts)})"
 
     def _build_kg_source_filters(
         kg_sources: list[str] | None,
@@ -114,16 +98,14 @@ def build_vespa_filters(
             return ""
 
         source_phrases = [f'{DOCUMENT_ID} contains "{source}"' for source in kg_sources]
-
-        return f"({' or '.join(source_phrases)}) and "
+        return f"({' or '.join(source_phrases)})"
 
     def _build_kg_chunk_id_zero_only_filter(
         kg_chunk_id_zero_only: bool,
     ) -> str:
         if not kg_chunk_id_zero_only:
             return ""
-
-        return "(chunk_id = 0 ) and "
+        return "(chunk_id = 0)"
 
     def _build_time_filter(
         cutoff: datetime | None,
@@ -135,8 +117,8 @@ def build_vespa_filters(
         cutoff_secs = int(cutoff.timestamp())
 
         if include_untimed:
-            return f"!({DOC_UPDATED_AT} < {cutoff_secs}) and "
-        return f"({DOC_UPDATED_AT} >= {cutoff_secs}) and "
+            return f"!({DOC_UPDATED_AT} < {cutoff_secs})"
+        return f"({DOC_UPDATED_AT} >= {cutoff_secs})"
 
     def _build_user_project_filter(
         project_id: int | None,
@@ -147,8 +129,7 @@ def build_vespa_filters(
             pid = int(project_id)
         except Exception:
             return ""
-        # Vespa YQL 'contains' expects a string literal; quote the integer
-        return f'({USER_PROJECT} contains "{pid}") and '
+        return f'({USER_PROJECT} contains "{pid}")'
 
     def _build_persona_filter(
         persona_id: int | None,
@@ -160,73 +141,94 @@ def build_vespa_filters(
         except Exception:
             logger.warning(f"Invalid persona ID: {persona_id}")
             return ""
-        return f'({PERSONAS} contains "{pid}") and '
+        return f'({PERSONAS} contains "{pid}")'
 
-    # Start building the filter string
-    filter_str = f"!({HIDDEN}=true) and " if not include_hidden else ""
+    def _append(parts: list[str], clause: str) -> None:
+        if clause:
+            parts.append(clause)
+
+    # Collect all top-level filter clauses, then join with " and " at the end.
+    filter_parts: list[str] = []
+
+    if not include_hidden:
+        filter_parts.append(f"!({HIDDEN}=true)")
 
     # TODO: add error condition if MULTI_TENANT and no tenant_id filter is set
-    # If running in multi-tenant mode
     if filters.tenant_id and MULTI_TENANT:
-        filter_str += build_tenant_id_filter(
-            filters.tenant_id, include_trailing_and=True
-        )
+        filter_parts.append(build_tenant_id_filter(filters.tenant_id))
 
     # ACL filters
     if filters.access_control_list is not None:
-        filter_str += _build_or_filters(
-            ACCESS_CONTROL_LIST, filters.access_control_list
+        _append(
+            filter_parts,
+            _build_or_filters(ACCESS_CONTROL_LIST, filters.access_control_list),
         )
 
     # Source type filters
     source_strs = (
         [s.value for s in filters.source_type] if filters.source_type else None
     )
-    filter_str += _build_or_filters(SOURCE_TYPE, source_strs)
+    _append(filter_parts, _build_or_filters(SOURCE_TYPE, source_strs))
 
     # Tag filters
     tag_attributes = None
     if filters.tags:
-        # build e.g. "tag_key|tag_value"
         tag_attributes = [
             f"{tag.tag_key}{INDEX_SEPARATOR}{tag.tag_value}" for tag in filters.tags
         ]
-    filter_str += _build_or_filters(METADATA_LIST, tag_attributes)
+    _append(filter_parts, _build_or_filters(METADATA_LIST, tag_attributes))
 
-    # Document sets
-    filter_str += _build_or_filters(DOCUMENT_SETS, filters.document_set)
+    # Knowledge scope: explicit knowledge attachments (document_sets,
+    # user_file_ids) restrict what an assistant can see.  When none are
+    # set, the assistant can see everything.
+    #
+    # project_id / persona_id are additive: they make overflowing user
+    # files findable in Vespa but must NOT trigger the restriction on
+    # their own (an agent with no explicit knowledge should search
+    # everything).
+    knowledge_scope_parts: list[str] = []
+
+    _append(
+        knowledge_scope_parts, _build_or_filters(DOCUMENT_SETS, filters.document_set)
+    )
 
-    # Convert UUIDs to strings for user_file_ids
     user_file_ids_str = (
         [str(uuid) for uuid in filters.user_file_ids] if filters.user_file_ids else None
     )
-    filter_str += _build_or_filters(DOCUMENT_ID, user_file_ids_str)
+    _append(knowledge_scope_parts, _build_or_filters(DOCUMENT_ID, user_file_ids_str))
 
-    # User project filter (array<int> attribute membership)
-    filter_str += _build_user_project_filter(filters.project_id)
+    # Only include project/persona scopes when an explicit knowledge
+    # restriction is already in effect — they widen the scope to also
+    # cover overflowing user files but never restrict on their own.
+    if knowledge_scope_parts:
+        _append(knowledge_scope_parts, _build_user_project_filter(filters.project_id))
+        _append(knowledge_scope_parts, _build_persona_filter(filters.persona_id))
 
-    # Persona filter (array<int> attribute membership)
-    filter_str += _build_persona_filter(filters.persona_id)
+    if len(knowledge_scope_parts) > 1:
+        filter_parts.append("(" + " or ".join(knowledge_scope_parts) + ")")
+    elif len(knowledge_scope_parts) == 1:
+        filter_parts.append(knowledge_scope_parts[0])
 
     # Time filter
-    filter_str += _build_time_filter(filters.time_cutoff)
+    _append(filter_parts, _build_time_filter(filters.time_cutoff))
 
     # # Knowledge Graph Filters
-    # filter_str += _build_kg_filter(
+    # _append(filter_parts, _build_kg_filter(
     #     kg_entities=filters.kg_entities,
     #     kg_relationships=filters.kg_relationships,
     #     kg_terms=filters.kg_terms,
-    # )
+    # ))
 
-    # filter_str += _build_kg_source_filters(filters.kg_sources)
+    # _append(filter_parts, _build_kg_source_filters(filters.kg_sources))
 
-    # filter_str += _build_kg_chunk_id_zero_only_filter(
+    # _append(filter_parts, _build_kg_chunk_id_zero_only_filter(
     #     filters.kg_chunk_id_zero_only or False
-    # )
+    # ))
 
-    # Trim trailing " and "
-    if remove_trailing_and and filter_str.endswith(" and "):
-        filter_str = filter_str[:-5]
+    filter_str = " and ".join(filter_parts)
+
+    if filter_str and not remove_trailing_and:
+        filter_str += " and "
 
     return filter_str
 

backend/onyx/error_handling/exceptions.py

@@ -48,10 +48,11 @@ class OnyxError(Exception):
         *,
         status_code_override: int | None = None,
     ) -> None:
+        resolved_message = message or error_code.code
+        super().__init__(resolved_message)
         self.error_code = error_code
-        self.message = message or error_code.code
+        self.message = resolved_message
         self._status_code_override = status_code_override
-        super().__init__(self.message)
 
     @property
     def status_code(self) -> int:

backend/onyx/indexing/indexing_pipeline.py

@@ -49,7 +49,6 @@ from onyx.indexing.embedder import IndexingEmbedder
 from onyx.indexing.models import DocAwareChunk
 from onyx.indexing.models import IndexingBatchAdapter
 from onyx.indexing.models import UpdatableChunkData
-from onyx.indexing.postgres_sanitization import sanitize_documents_for_postgres
 from onyx.indexing.vector_db_insertion import write_chunks_to_vector_db_with_backoff
 from onyx.llm.factory import get_default_llm_with_vision
 from onyx.llm.factory import get_llm_for_contextual_rag
@@ -65,6 +64,7 @@ from onyx.prompts.contextual_retrieval import CONTEXTUAL_RAG_PROMPT1
 from onyx.prompts.contextual_retrieval import CONTEXTUAL_RAG_PROMPT2
 from onyx.prompts.contextual_retrieval import DOCUMENT_SUMMARY_PROMPT
 from onyx.utils.logger import setup_logger
+from onyx.utils.postgres_sanitization import sanitize_documents_for_postgres
 from onyx.utils.threadpool_concurrency import run_functions_tuples_in_parallel
 from onyx.utils.timing import log_function_time
 

backend/onyx/llm/constants.py

@@ -22,6 +22,7 @@ class LlmProviderNames(str, Enum):
     OPENROUTER = "openrouter"
     AZURE = "azure"
     OLLAMA_CHAT = "ollama_chat"
+    LM_STUDIO = "lm_studio"
     MISTRAL = "mistral"
     LITELLM_PROXY = "litellm_proxy"
 
@@ -41,6 +42,7 @@ WELL_KNOWN_PROVIDER_NAMES = [
     LlmProviderNames.OPENROUTER,
     LlmProviderNames.AZURE,
     LlmProviderNames.OLLAMA_CHAT,
+    LlmProviderNames.LM_STUDIO,
 ]
 
 
@@ -56,6 +58,7 @@ PROVIDER_DISPLAY_NAMES: dict[str, str] = {
     LlmProviderNames.AZURE: "Azure",
     "ollama": "Ollama",
     LlmProviderNames.OLLAMA_CHAT: "Ollama",
+    LlmProviderNames.LM_STUDIO: "LM Studio",
     "groq": "Groq",
     "anyscale": "Anyscale",
     "deepseek": "DeepSeek",
@@ -103,6 +106,7 @@ AGGREGATOR_PROVIDERS: set[str] = {
     LlmProviderNames.BEDROCK_CONVERSE,
     LlmProviderNames.OPENROUTER,
     LlmProviderNames.OLLAMA_CHAT,
+    LlmProviderNames.LM_STUDIO,
     LlmProviderNames.VERTEX_AI,
     LlmProviderNames.AZURE,
 }

backend/onyx/llm/factory.py

@@ -20,7 +20,9 @@ from onyx.llm.multi_llm import LitellmLLM
 from onyx.llm.override_models import LLMOverride
 from onyx.llm.utils import get_max_input_tokens_from_llm_provider
 from onyx.llm.utils import model_supports_image_input
-from onyx.llm.well_known_providers.constants import OLLAMA_API_KEY_CONFIG_KEY
+from onyx.llm.well_known_providers.constants import (
+    PROVIDERS_WITH_SPECIAL_API_KEY_HANDLING,
+)
 from onyx.natural_language_processing.utils import get_tokenizer
 from onyx.server.manage.llm.models import LLMProviderView
 from onyx.utils.headers import build_llm_extra_headers
@@ -32,14 +34,18 @@ logger = setup_logger()
 def _build_provider_extra_headers(
     provider: str, custom_config: dict[str, str] | None
 ) -> dict[str, str]:
-    if provider == LlmProviderNames.OLLAMA_CHAT and custom_config:
-        raw_api_key = custom_config.get(OLLAMA_API_KEY_CONFIG_KEY)
-        api_key = raw_api_key.strip() if raw_api_key else None
+    if provider in PROVIDERS_WITH_SPECIAL_API_KEY_HANDLING and custom_config:
+        raw = custom_config.get(PROVIDERS_WITH_SPECIAL_API_KEY_HANDLING[provider])
+        api_key = raw.strip() if raw else None
         if not api_key:
             return {}
-        if not api_key.lower().startswith("bearer "):
-            api_key = f"Bearer {api_key}"
-        return {"Authorization": api_key}
+        return {
+            "Authorization": (
+                api_key
+                if api_key.lower().startswith("bearer ")
+                else f"Bearer {api_key}"
+            )
+        }
 
     # Passing these will put Onyx on the OpenRouter leaderboard
     elif provider == LlmProviderNames.OPENROUTER:

backend/onyx/llm/model_metadata_enrichments.json

@@ -1512,6 +1512,10 @@
     "display_name": "Claude Opus 4.5",
     "model_vendor": "anthropic"
   },
+  "claude-opus-4-6": {
+    "display_name": "Claude Opus 4.6",
+    "model_vendor": "anthropic"
+  },
   "claude-opus-4-5-20251101": {
     "display_name": "Claude Opus 4.5",
     "model_vendor": "anthropic",
@@ -1526,6 +1530,10 @@
     "display_name": "Claude Sonnet 4.5",
     "model_vendor": "anthropic"
   },
+  "claude-sonnet-4-6": {
+    "display_name": "Claude Sonnet 4.6",
+    "model_vendor": "anthropic"
+  },
   "claude-sonnet-4-5-20250929": {
     "display_name": "Claude Sonnet 4.5",
     "model_vendor": "anthropic",
@@ -2516,6 +2524,10 @@
     "model_vendor": "openai",
     "model_version": "2025-10-06"
   },
+  "gpt-5.4": {
+    "display_name": "GPT-5.4",
+    "model_vendor": "openai"
+  },
   "gpt-5.2-pro-2025-12-11": {
     "display_name": "GPT-5.2 Pro",
     "model_vendor": "openai",

backend/onyx/llm/multi_llm.py

@@ -42,6 +42,7 @@ from onyx.llm.well_known_providers.constants import AWS_SECRET_ACCESS_KEY_KWARG
 from onyx.llm.well_known_providers.constants import (
     AWS_SECRET_ACCESS_KEY_KWARG_ENV_VAR_FORMAT,
 )
+from onyx.llm.well_known_providers.constants import LM_STUDIO_API_KEY_CONFIG_KEY
 from onyx.llm.well_known_providers.constants import OLLAMA_API_KEY_CONFIG_KEY
 from onyx.llm.well_known_providers.constants import VERTEX_CREDENTIALS_FILE_KWARG
 from onyx.llm.well_known_providers.constants import (
@@ -92,6 +93,98 @@ def _prompt_to_dicts(prompt: LanguageModelInput) -> list[dict[str, Any]]:
     return [prompt.model_dump(exclude_none=True)]
 
 
+def _normalize_content(raw: Any) -> str:
+    """Normalize a message content field to a plain string.
+
+    Content can be a string, None, or a list of content-block dicts
+    (e.g. [{"type": "text", "text": "..."}]).
+    """
+    if raw is None:
+        return ""
+    if isinstance(raw, str):
+        return raw
+    if isinstance(raw, list):
+        return "\n".join(
+            block.get("text", "") if isinstance(block, dict) else str(block)
+            for block in raw
+        )
+    return str(raw)
+
+
+def _strip_tool_content_from_messages(
+    messages: list[dict[str, Any]],
+) -> list[dict[str, Any]]:
+    """Convert tool-related messages to plain text.
+
+    Bedrock's Converse API requires toolConfig when messages contain
+    toolUse/toolResult content blocks. When no tools are provided for the
+    current request, we must convert any tool-related history into plain text
+    to avoid the "toolConfig field must be defined" error.
+
+    This is the same approach used by _OllamaHistoryMessageFormatter.
+    """
+    result: list[dict[str, Any]] = []
+    for msg in messages:
+        role = msg.get("role")
+        tool_calls = msg.get("tool_calls")
+
+        if role == "assistant" and tool_calls:
+            # Convert structured tool calls to text representation
+            tool_call_lines = []
+            for tc in tool_calls:
+                func = tc.get("function", {})
+                name = func.get("name", "unknown")
+                args = func.get("arguments", "{}")
+                tc_id = tc.get("id", "")
+                tool_call_lines.append(
+                    f"[Tool Call] name={name} id={tc_id} args={args}"
+                )
+
+            existing_content = _normalize_content(msg.get("content"))
+            parts = (
+                [existing_content] + tool_call_lines
+                if existing_content
+                else tool_call_lines
+            )
+            new_msg = {
+                "role": "assistant",
+                "content": "\n".join(parts),
+            }
+            result.append(new_msg)
+
+        elif role == "tool":
+            # Convert tool response to user message with text content
+            tool_call_id = msg.get("tool_call_id", "")
+            content = _normalize_content(msg.get("content"))
+            tool_result_text = f"[Tool Result] id={tool_call_id}\n{content}"
+            # Merge into previous user message if it is also a converted
+            # tool result to avoid consecutive user messages (Bedrock requires
+            # strict user/assistant alternation).
+            if (
+                result
+                and result[-1]["role"] == "user"
+                and "[Tool Result]" in result[-1].get("content", "")
+            ):
+                result[-1]["content"] += "\n\n" + tool_result_text
+            else:
+                result.append({"role": "user", "content": tool_result_text})
+
+        else:
+            result.append(msg)
+
+    return result
+
+
+def _messages_contain_tool_content(messages: list[dict[str, Any]]) -> bool:
+    """Check if any messages contain tool-related content blocks."""
+    for msg in messages:
+        if msg.get("role") == "tool":
+            return True
+        if msg.get("role") == "assistant" and msg.get("tool_calls"):
+            return True
+    return False
+
+
 def _is_vertex_model_rejecting_output_config(model_name: str) -> bool:
     normalized_model_name = model_name.lower()
     return any(
@@ -157,6 +250,9 @@ class LitellmLLM(LLM):
                 elif model_provider == LlmProviderNames.OLLAMA_CHAT:
                     if k == OLLAMA_API_KEY_CONFIG_KEY:
                         model_kwargs["api_key"] = v
+                elif model_provider == LlmProviderNames.LM_STUDIO:
+                    if k == LM_STUDIO_API_KEY_CONFIG_KEY:
+                        model_kwargs["api_key"] = v
                 elif model_provider == LlmProviderNames.BEDROCK:
                     if k == AWS_REGION_NAME_KWARG:
                         model_kwargs[k] = v
@@ -173,6 +269,19 @@ class LitellmLLM(LLM):
                     elif k == AWS_SECRET_ACCESS_KEY_KWARG_ENV_VAR_FORMAT:
                         model_kwargs[AWS_SECRET_ACCESS_KEY_KWARG] = v
 
+        # LM Studio: LiteLLM defaults to "fake-api-key" when no key is provided,
+        # which LM Studio rejects. Ensure we always pass an explicit key (or empty
+        # string) to prevent LiteLLM from injecting its fake default.
+        if model_provider == LlmProviderNames.LM_STUDIO:
+            model_kwargs.setdefault("api_key", "")
+
+            # Users provide the server root (e.g. http://localhost:1234) but LiteLLM
+            # needs /v1 for OpenAI-compatible calls.
+            if self._api_base is not None:
+                base = self._api_base.rstrip("/")
+                self._api_base = base if base.endswith("/v1") else f"{base}/v1"
+                model_kwargs["api_base"] = self._api_base
+
         # Default vertex_location to "global" if not provided for Vertex AI
         # Latest gemini models are only available through the global region
         if (
@@ -404,13 +513,30 @@ class LitellmLLM(LLM):
                 else nullcontext()
             )
             with env_ctx:
+                messages = _prompt_to_dicts(prompt)
+
+                # Bedrock's Converse API requires toolConfig when messages
+                # contain toolUse/toolResult content blocks. When no tools are
+                # provided for this request but the history contains tool
+                # content from previous turns, strip it to plain text.
+                is_bedrock = self._model_provider in {
+                    LlmProviderNames.BEDROCK,
+                    LlmProviderNames.BEDROCK_CONVERSE,
+                }
+                if (
+                    is_bedrock
+                    and not tools
+                    and _messages_contain_tool_content(messages)
+                ):
+                    messages = _strip_tool_content_from_messages(messages)
+
                 response = litellm.completion(
                     mock_response=get_llm_mock_response() or MOCK_LLM_RESPONSE,
                     model=model,
                     base_url=self._api_base or None,
                     api_version=self._api_version or None,
                     custom_llm_provider=self._custom_llm_provider or None,
-                    messages=_prompt_to_dicts(prompt),
+                    messages=messages,
                     tools=tools,
                     tool_choice=tool_choice,
                     stream=stream,

backend/onyx/llm/utils.py

@@ -322,7 +322,7 @@ def test_llm(llm: LLM) -> str | None:
     error_msg = None
     for _ in range(2):
         try:
-            llm.invoke(UserMessage(content="Do not respond"))
+            llm.invoke(UserMessage(content="Do not respond"), max_tokens=50)
             return None
         except Exception as e:
             error_msg = str(e)

backend/onyx/llm/well_known_providers/constants.py

@@ -1,52 +1,27 @@
+from onyx.llm.constants import LlmProviderNames
+
 OPENAI_PROVIDER_NAME = "openai"
-# Curated list of OpenAI models to show by default in the UI
-OPENAI_VISIBLE_MODEL_NAMES = {
-    "gpt-5",
-    "gpt-5-mini",
-    "o1",
-    "o3-mini",
-    "gpt-4o",
-    "gpt-4o-mini",
-}
 
 BEDROCK_PROVIDER_NAME = "bedrock"
-BEDROCK_DEFAULT_MODEL = "anthropic.claude-3-5-sonnet-20241022-v2:0"
-
-
-def _fallback_bedrock_regions() -> list[str]:
-    # Fall back to a conservative set of well-known Bedrock regions if boto3 data isn't available.
-    return [
-        "us-east-1",
-        "us-east-2",
-        "us-gov-east-1",
-        "us-gov-west-1",
-        "us-west-2",
-        "ap-northeast-1",
-        "ap-south-1",
-        "ap-southeast-1",
-        "ap-southeast-2",
-        "ap-east-1",
-        "ca-central-1",
-        "eu-central-1",
-        "eu-west-2",
-    ]
 
 
 OLLAMA_PROVIDER_NAME = "ollama_chat"
 OLLAMA_API_KEY_CONFIG_KEY = "OLLAMA_API_KEY"
 
+LM_STUDIO_PROVIDER_NAME = "lm_studio"
+LM_STUDIO_API_KEY_CONFIG_KEY = "LM_STUDIO_API_KEY"
+
+# Providers that use optional Bearer auth from custom_config
+PROVIDERS_WITH_SPECIAL_API_KEY_HANDLING: dict[str, str] = {
+    LlmProviderNames.OLLAMA_CHAT: OLLAMA_API_KEY_CONFIG_KEY,
+    LlmProviderNames.LM_STUDIO: LM_STUDIO_API_KEY_CONFIG_KEY,
+}
+
 # OpenRouter
 OPENROUTER_PROVIDER_NAME = "openrouter"
 
 ANTHROPIC_PROVIDER_NAME = "anthropic"
 
-# Curated list of Anthropic models to show by default in the UI
-ANTHROPIC_VISIBLE_MODEL_NAMES = {
-    "claude-opus-4-5",
-    "claude-sonnet-4-5",
-    "claude-haiku-4-5",
-}
-
 AZURE_PROVIDER_NAME = "azure"
 
 
@@ -54,13 +29,6 @@ VERTEXAI_PROVIDER_NAME = "vertex_ai"
 VERTEX_CREDENTIALS_FILE_KWARG = "vertex_credentials"
 VERTEX_CREDENTIALS_FILE_KWARG_ENV_VAR_FORMAT = "CREDENTIALS_FILE"
 VERTEX_LOCATION_KWARG = "vertex_location"
-VERTEXAI_DEFAULT_MODEL = "gemini-2.5-flash"
-# Curated list of Vertex AI models to show by default in the UI
-VERTEXAI_VISIBLE_MODEL_NAMES = {
-    "gemini-2.5-flash",
-    "gemini-2.5-flash-lite",
-    "gemini-2.5-pro",
-}
 
 AWS_REGION_NAME_KWARG = "aws_region_name"
 AWS_REGION_NAME_KWARG_ENV_VAR_FORMAT = "AWS_REGION_NAME"

backend/onyx/llm/well_known_providers/llm_provider_options.py

@@ -15,6 +15,7 @@ from onyx.llm.well_known_providers.auto_update_service import (
 from onyx.llm.well_known_providers.constants import ANTHROPIC_PROVIDER_NAME
 from onyx.llm.well_known_providers.constants import AZURE_PROVIDER_NAME
 from onyx.llm.well_known_providers.constants import BEDROCK_PROVIDER_NAME
+from onyx.llm.well_known_providers.constants import LM_STUDIO_PROVIDER_NAME
 from onyx.llm.well_known_providers.constants import OLLAMA_PROVIDER_NAME
 from onyx.llm.well_known_providers.constants import OPENAI_PROVIDER_NAME
 from onyx.llm.well_known_providers.constants import OPENROUTER_PROVIDER_NAME
@@ -44,6 +45,7 @@ def _get_provider_to_models_map() -> dict[str, list[str]]:
         ANTHROPIC_PROVIDER_NAME: get_anthropic_model_names(),
         VERTEXAI_PROVIDER_NAME: get_vertexai_model_names(),
         OLLAMA_PROVIDER_NAME: [],  # Dynamic - fetched from Ollama API
+        LM_STUDIO_PROVIDER_NAME: [],  # Dynamic - fetched from LM Studio API
         OPENROUTER_PROVIDER_NAME: [],  # Dynamic - fetched from OpenRouter API
     }
 
@@ -323,6 +325,7 @@ def get_provider_display_name(provider_name: str) -> str:
     _ONYX_PROVIDER_DISPLAY_NAMES: dict[str, str] = {
         OPENAI_PROVIDER_NAME: "ChatGPT (OpenAI)",
         OLLAMA_PROVIDER_NAME: "Ollama",
+        LM_STUDIO_PROVIDER_NAME: "LM Studio",
         ANTHROPIC_PROVIDER_NAME: "Claude (Anthropic)",
         AZURE_PROVIDER_NAME: "Azure OpenAI",
         BEDROCK_PROVIDER_NAME: "Amazon Bedrock",

backend/onyx/llm/well_known_providers/recommended-models.json

@@ -1,12 +1,12 @@
 {
   "version": "1.1",
-  "updated_at": "2026-02-05T00:00:00Z",
+  "updated_at": "2026-03-05T00:00:00Z",
   "providers": {
     "openai": {
-      "default_model": { "name": "gpt-5.2" },
+      "default_model": { "name": "gpt-5.4" },
       "additional_visible_models": [
-        { "name": "gpt-5-mini" },
-        { "name": "gpt-4.1" }
+        { "name": "gpt-5.4" },
+        { "name": "gpt-5.2" }
       ]
     },
     "anthropic": {
@@ -16,6 +16,10 @@
           "name": "claude-opus-4-6",
           "display_name": "Claude Opus 4.6"
         },
+        {
+          "name": "claude-sonnet-4-6",
+          "display_name": "Claude Sonnet 4.6"
+        },
         {
           "name": "claude-opus-4-5",
           "display_name": "Claude Opus 4.5"

backend/onyx/server/features/build/api/api.py

@@ -5,6 +5,7 @@ from uuid import UUID
 import httpx
 from fastapi import APIRouter
 from fastapi import Depends
+from fastapi import HTTPException
 from fastapi import Request
 from fastapi import Response
 from fastapi.responses import RedirectResponse
@@ -23,8 +24,6 @@ from onyx.db.enums import SharingScope
 from onyx.db.index_attempt import get_latest_index_attempt_for_cc_pair_id
 from onyx.db.models import BuildSession
 from onyx.db.models import User
-from onyx.error_handling.error_codes import OnyxErrorCode
-from onyx.error_handling.exceptions import OnyxError
 from onyx.server.features.build.api.messages_api import router as messages_router
 from onyx.server.features.build.api.models import BuildConnectorInfo
 from onyx.server.features.build.api.models import BuildConnectorListResponse
@@ -45,10 +44,13 @@ logger = setup_logger()
 def require_onyx_craft_enabled(user: User = Depends(current_user)) -> User:
     """
     Dependency that checks if Onyx Craft is enabled for the user.
-    Raises OnyxError(UNAUTHORIZED) if Onyx Craft is disabled via feature flag.
+    Raises HTTP 403 if Onyx Craft is disabled via feature flag.
     """
     if not is_onyx_craft_enabled(user):
-        raise OnyxError(OnyxErrorCode.UNAUTHORIZED, "Onyx Craft is not available")
+        raise HTTPException(
+            status_code=403,
+            detail="Onyx Craft is not available",
+        )
     return user
 
 
@@ -288,20 +290,20 @@ def _get_sandbox_url(session_id: UUID, db_session: Session) -> str:
         Internal URL to proxy requests to
 
     Raises:
-        OnyxError: If session not found, port not allocated, or sandbox not found
+        HTTPException: If session not found, port not allocated, or sandbox not found
     """
 
     session = db_session.get(BuildSession, session_id)
     if not session:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
     if session.nextjs_port is None:
-        raise OnyxError(OnyxErrorCode.SERVICE_UNAVAILABLE, "Session port not allocated")
+        raise HTTPException(status_code=503, detail="Session port not allocated")
     if session.user_id is None:
-        raise OnyxError(OnyxErrorCode.USER_NOT_FOUND, "User not found")
+        raise HTTPException(status_code=404, detail="User not found")
 
     sandbox = get_sandbox_by_user_id(db_session, session.user_id)
     if sandbox is None:
-        raise OnyxError(OnyxErrorCode.NOT_FOUND, "Sandbox not found")
+        raise HTTPException(status_code=404, detail="Sandbox not found")
 
     sandbox_manager = get_sandbox_manager()
     return sandbox_manager.get_webapp_url(sandbox.id, session.nextjs_port)
@@ -362,10 +364,10 @@ def _proxy_request(
 
     except httpx.TimeoutException:
         logger.error(f"Timeout while proxying request to {target_url}")
-        raise OnyxError(OnyxErrorCode.GATEWAY_TIMEOUT, "Gateway timeout")
+        raise HTTPException(status_code=504, detail="Gateway timeout")
     except httpx.RequestError as e:
         logger.error(f"Error proxying request to {target_url}: {e}")
-        raise OnyxError(OnyxErrorCode.BAD_GATEWAY, "Bad gateway")
+        raise HTTPException(status_code=502, detail="Bad gateway")
 
 
 def _check_webapp_access(
@@ -379,13 +381,13 @@ def _check_webapp_access(
     """
     session = db_session.get(BuildSession, session_id)
     if not session:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
     if session.sharing_scope == SharingScope.PUBLIC_GLOBAL:
         return session
     if user is None:
-        raise OnyxError(OnyxErrorCode.UNAUTHENTICATED, "Authentication required")
+        raise HTTPException(status_code=401, detail="Authentication required")
     if session.sharing_scope == SharingScope.PRIVATE and session.user_id != user.id:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
     return session
 
 
@@ -425,13 +427,13 @@ def get_webapp(
     """
     try:
         _check_webapp_access(session_id, user, db_session)
-    except OnyxError as e:
+    except HTTPException as e:
         if e.status_code == 401:
             return RedirectResponse(url="/auth/login", status_code=302)
         raise
     try:
         return _proxy_request(path, request, session_id, db_session)
-    except OnyxError as e:
+    except HTTPException as e:
         if e.status_code in (502, 503, 504):
             return _offline_html_response()
         raise
@@ -460,13 +462,19 @@ def reset_sandbox(
     try:
         success = session_manager.terminate_user_sandbox(user.id)
         if not success:
-            raise OnyxError(OnyxErrorCode.NOT_FOUND, "No sandbox found for user")
+            raise HTTPException(
+                status_code=404,
+                detail="No sandbox found for user",
+            )
         db_session.commit()
-    except OnyxError:
+    except HTTPException:
         raise
     except Exception as e:
         db_session.rollback()
         logger.error(f"Failed to reset sandbox for user {user.id}: {e}")
-        raise OnyxError(OnyxErrorCode.INTERNAL_ERROR, f"Failed to reset sandbox: {e}")
+        raise HTTPException(
+            status_code=500,
+            detail=f"Failed to reset sandbox: {e}",
+        )
 
     return Response(status_code=204)

backend/onyx/server/features/build/api/messages_api.py

@@ -5,6 +5,7 @@ from uuid import UUID
 
 from fastapi import APIRouter
 from fastapi import Depends
+from fastapi import HTTPException
 from fastapi.responses import StreamingResponse
 from sqlalchemy.orm import Session
 
@@ -13,8 +14,6 @@ from onyx.configs.constants import PUBLIC_API_TAGS
 from onyx.db.engine.sql_engine import get_session
 from onyx.db.engine.sql_engine import get_session_with_current_tenant
 from onyx.db.models import User
-from onyx.error_handling.error_codes import OnyxErrorCode
-from onyx.error_handling.exceptions import OnyxError
 from onyx.server.features.build.api.models import MessageListResponse
 from onyx.server.features.build.api.models import MessageRequest
 from onyx.server.features.build.api.models import MessageResponse
@@ -37,7 +36,7 @@ def check_build_rate_limits(
     """
     Dependency to check build mode rate limits before processing the request.
 
-    Raises OnyxError(RATE_LIMITED) if rate limit is exceeded.
+    Raises HTTPException(429) if rate limit is exceeded.
     Follows the same pattern as chat's check_token_rate_limits.
     """
     session_manager = SessionManager(db_session)
@@ -45,7 +44,10 @@ def check_build_rate_limits(
     try:
         session_manager.check_rate_limit(user)
     except RateLimitError as e:
-        raise OnyxError(OnyxErrorCode.RATE_LIMITED, str(e))
+        raise HTTPException(
+            status_code=429,
+            detail=str(e),
+        )
 
 
 @router.get("/sessions/{session_id}/messages", tags=PUBLIC_API_TAGS)
@@ -56,14 +58,14 @@ def list_messages(
 ) -> MessageListResponse:
     """Get all messages for a build session."""
     if user is None:
-        raise OnyxError(OnyxErrorCode.UNAUTHENTICATED, "Authentication required")
+        raise HTTPException(status_code=401, detail="Authentication required")
 
     session_manager = SessionManager(db_session)
 
     messages = session_manager.list_messages(session_id, user.id)
 
     if messages is None:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
 
     return MessageListResponse(
         messages=[MessageResponse.from_model(msg) for msg in messages]

backend/onyx/server/features/build/api/sessions_api.py

@@ -5,6 +5,7 @@ from uuid import UUID
 from fastapi import APIRouter
 from fastapi import Depends
 from fastapi import File
+from fastapi import HTTPException
 from fastapi import Response
 from fastapi import UploadFile
 from sqlalchemy import exists
@@ -16,8 +17,6 @@ from onyx.db.enums import BuildSessionStatus
 from onyx.db.enums import SandboxStatus
 from onyx.db.models import BuildMessage
 from onyx.db.models import User
-from onyx.error_handling.error_codes import OnyxErrorCode
-from onyx.error_handling.exceptions import OnyxError
 from onyx.redis.redis_pool import get_redis_client
 from onyx.server.features.build.api.models import ArtifactResponse
 from onyx.server.features.build.api.models import DetailedSessionResponse
@@ -118,9 +117,9 @@ def create_session(
         blocking=True, blocking_timeout=SESSION_CREATE_LOCK_TIMEOUT_SECONDS
     )
     if not acquired:
-        raise OnyxError(
-            OnyxErrorCode.SERVICE_UNAVAILABLE,
-            "Session creation timed out waiting for lock",
+        raise HTTPException(
+            status_code=503,
+            detail="Session creation timed out waiting for lock",
         )
 
     try:
@@ -145,11 +144,11 @@ def create_session(
     except ValueError as e:
         logger.exception("Session creation failed")
         db_session.rollback()
-        raise OnyxError(OnyxErrorCode.RATE_LIMITED, str(e))
+        raise HTTPException(status_code=429, detail=str(e))
     except Exception as e:
         db_session.rollback()
         logger.error(f"Session creation failed: {e}")
-        raise OnyxError(OnyxErrorCode.INTERNAL_ERROR, f"Session creation failed: {e}")
+        raise HTTPException(status_code=500, detail=f"Session creation failed: {e}")
     finally:
         if lock.owned():
             lock.release()
@@ -172,7 +171,7 @@ def get_session_details(
     session = session_manager.get_session(session_id, user.id)
 
     if session is None:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
 
     # Get the user's sandbox to include in response
     sandbox = get_sandbox_by_user_id(db_session, user.id)
@@ -238,7 +237,7 @@ def generate_session_name(
     generated_name = session_manager.generate_session_name(session_id, user.id)
 
     if generated_name is None:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
 
     return SessionNameGenerateResponse(name=generated_name)
 
@@ -258,7 +257,7 @@ def generate_suggestions(
     # Verify session exists and belongs to user
     session = session_manager.get_session(session_id, user.id)
     if session is None:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
 
     # Generate suggestions
     suggestions_data = session_manager.generate_followup_suggestions(
@@ -291,7 +290,7 @@ def update_session_name(
     session = session_manager.update_session_name(session_id, user.id, request.name)
 
     if session is None:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
 
     # Get the user's sandbox to include in response
     sandbox = get_sandbox_by_user_id(db_session, user.id)
@@ -310,7 +309,7 @@ def set_session_public(
         session_id, user.id, request.sharing_scope, db_session
     )
     if not updated:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
     return SetSessionSharingResponse(
         session_id=str(session_id),
         sharing_scope=updated.sharing_scope,
@@ -333,18 +332,18 @@ def delete_session(
     try:
         success = session_manager.delete_session(session_id, user.id)
         if not success:
-            raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+            raise HTTPException(status_code=404, detail="Session not found")
         db_session.commit()
-    except OnyxError:
-        # Re-raise OnyxError exceptions (like 404) without rollback
+    except HTTPException:
+        # Re-raise HTTP exceptions (like 404) without rollback
         raise
     except Exception as e:
         # Sandbox termination failed - rollback to preserve session
         db_session.rollback()
         logger.error(f"Failed to delete session {session_id}: {e}")
-        raise OnyxError(
-            OnyxErrorCode.INTERNAL_ERROR,
-            f"Failed to delete session: {e}",
+        raise HTTPException(
+            status_code=500,
+            detail=f"Failed to delete session: {e}",
         )
 
     return Response(status_code=204)
@@ -374,11 +373,11 @@ def restore_session(
     """
     session = get_build_session(session_id, user.id, db_session)
     if not session:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
 
     sandbox = get_sandbox_by_user_id(db_session, user.id)
     if not sandbox:
-        raise OnyxError(OnyxErrorCode.NOT_FOUND, "Sandbox not found")
+        raise HTTPException(status_code=404, detail="Sandbox not found")
 
     # If sandbox is already running, check if session workspace exists
     sandbox_manager = get_sandbox_manager()
@@ -393,7 +392,10 @@ def restore_session(
     # instead of making the user wait. The frontend will retry.
     acquired = lock.acquire(blocking=False)
     if not acquired:
-        raise OnyxError(OnyxErrorCode.CONFLICT, "Restore already in progress")
+        raise HTTPException(
+            status_code=409,
+            detail="Restore already in progress",
+        )
 
     try:
         # Re-fetch sandbox status (may have changed while waiting for lock)
@@ -508,9 +510,9 @@ def restore_session(
 
     except Exception as e:
         logger.error(f"Failed to restore session {session_id}: {e}", exc_info=True)
-        raise OnyxError(
-            OnyxErrorCode.INTERNAL_ERROR,
-            f"Failed to restore session: {e}",
+        raise HTTPException(
+            status_code=500,
+            detail=f"Failed to restore session: {e}",
         )
     finally:
         if lock.owned():
@@ -545,7 +547,7 @@ def list_artifacts(
 
     artifacts = session_manager.list_artifacts(session_id, user_id)
     if artifacts is None:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
 
     return artifacts
 
@@ -575,15 +577,15 @@ def list_directory(
     except ValueError as e:
         error_message = str(e)
         if "path traversal" in error_message.lower():
-            raise OnyxError(OnyxErrorCode.UNAUTHORIZED, "Access denied")
+            raise HTTPException(status_code=403, detail="Access denied")
         elif "not found" in error_message.lower():
-            raise OnyxError(OnyxErrorCode.NOT_FOUND, "Directory not found")
+            raise HTTPException(status_code=404, detail="Directory not found")
         elif "not a directory" in error_message.lower():
-            raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, "Path is not a directory")
-        raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, error_message)
+            raise HTTPException(status_code=400, detail="Path is not a directory")
+        raise HTTPException(status_code=400, detail=error_message)
 
     if listing is None:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
 
     return listing
 
@@ -607,13 +609,13 @@ def download_artifact(
             "path traversal" in error_message.lower()
             or "access denied" in error_message.lower()
         ):
-            raise OnyxError(OnyxErrorCode.UNAUTHORIZED, "Access denied")
+            raise HTTPException(status_code=403, detail="Access denied")
         elif "directory" in error_message.lower():
-            raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, "Cannot download directory")
-        raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, error_message)
+            raise HTTPException(status_code=400, detail="Cannot download directory")
+        raise HTTPException(status_code=400, detail=error_message)
 
     if result is None:
-        raise OnyxError(OnyxErrorCode.NOT_FOUND, "Artifact not found")
+        raise HTTPException(status_code=404, detail="Artifact not found")
 
     content, mime_type, filename = result
 
@@ -657,11 +659,11 @@ def export_docx(
             "path traversal" in error_message.lower()
             or "access denied" in error_message.lower()
         ):
-            raise OnyxError(OnyxErrorCode.UNAUTHORIZED, "Access denied")
-        raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, error_message)
+            raise HTTPException(status_code=403, detail="Access denied")
+        raise HTTPException(status_code=400, detail=error_message)
 
     if result is None:
-        raise OnyxError(OnyxErrorCode.NOT_FOUND, "File not found")
+        raise HTTPException(status_code=404, detail="File not found")
 
     docx_bytes, filename = result
 
@@ -699,11 +701,11 @@ def get_pptx_preview(
             "path traversal" in error_message.lower()
             or "access denied" in error_message.lower()
         ):
-            raise OnyxError(OnyxErrorCode.UNAUTHORIZED, "Access denied")
-        raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, error_message)
+            raise HTTPException(status_code=403, detail="Access denied")
+        raise HTTPException(status_code=400, detail=error_message)
 
     if result is None:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
 
     return PptxPreviewResponse(**result)
 
@@ -725,7 +727,7 @@ def get_webapp_info(
     webapp_info = session_manager.get_webapp_info(session_id, user_id)
 
     if webapp_info is None:
-        raise OnyxError(OnyxErrorCode.SESSION_NOT_FOUND, "Session not found")
+        raise HTTPException(status_code=404, detail="Session not found")
 
     return WebappInfo(**webapp_info)
 
@@ -747,7 +749,7 @@ def download_webapp(
     result = session_manager.download_webapp_zip(session_id, user_id)
 
     if result is None:
-        raise OnyxError(OnyxErrorCode.NOT_FOUND, "Webapp not found")
+        raise HTTPException(status_code=404, detail="Webapp not found")
 
     zip_bytes, filename = result
 
@@ -780,11 +782,11 @@ def download_directory(
     except ValueError as e:
         error_message = str(e)
         if "path traversal" in error_message.lower():
-            raise OnyxError(OnyxErrorCode.UNAUTHORIZED, "Access denied")
-        raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, error_message)
+            raise HTTPException(status_code=403, detail="Access denied")
+        raise HTTPException(status_code=400, detail=error_message)
 
     if result is None:
-        raise OnyxError(OnyxErrorCode.NOT_FOUND, "Directory not found")
+        raise HTTPException(status_code=404, detail="Directory not found")
 
     zip_bytes, filename = result
 
@@ -812,7 +814,7 @@ def upload_file_endpoint(
     session_manager = SessionManager(db_session)
 
     if not file.filename:
-        raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, "File has no filename")
+        raise HTTPException(status_code=400, detail="File has no filename")
 
     # Read file content (use sync file interface)
     content = file.file.read()
@@ -820,7 +822,7 @@ def upload_file_endpoint(
     # Validate file (extension, mime type, size)
     is_valid, error = validate_file(file.filename, file.content_type, len(content))
     if not is_valid:
-        raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, error)
+        raise HTTPException(status_code=400, detail=error)
 
     # Sanitize filename
     safe_filename = sanitize_filename(file.filename)
@@ -834,12 +836,12 @@ def upload_file_endpoint(
         )
     except UploadLimitExceededError as e:
         # Return 429 for limit exceeded errors
-        raise OnyxError(OnyxErrorCode.RATE_LIMITED, str(e))
+        raise HTTPException(status_code=429, detail=str(e))
     except ValueError as e:
         error_message = str(e)
         if "not found" in error_message.lower():
-            raise OnyxError(OnyxErrorCode.NOT_FOUND, error_message)
-        raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, error_message)
+            raise HTTPException(status_code=404, detail=error_message)
+        raise HTTPException(status_code=400, detail=error_message)
 
     return UploadResponse(
         filename=safe_filename,
@@ -869,14 +871,14 @@ def delete_file_endpoint(
     except ValueError as e:
         error_message = str(e)
         if "path traversal" in error_message.lower():
-            raise OnyxError(OnyxErrorCode.UNAUTHORIZED, "Access denied")
+            raise HTTPException(status_code=403, detail="Access denied")
         elif "not found" in error_message.lower():
-            raise OnyxError(OnyxErrorCode.NOT_FOUND, error_message)
+            raise HTTPException(status_code=404, detail=error_message)
         elif "directory" in error_message.lower():
-            raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, "Cannot delete directory")
-        raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, error_message)
+            raise HTTPException(status_code=400, detail="Cannot delete directory")
+        raise HTTPException(status_code=400, detail=error_message)
 
     if not deleted:
-        raise OnyxError(OnyxErrorCode.NOT_FOUND, "File not found")
+        raise HTTPException(status_code=404, detail="File not found")
 
     return Response(status_code=204)

backend/onyx/server/features/build/api/user_library.py

@@ -32,6 +32,7 @@ from fastapi import APIRouter
 from fastapi import Depends
 from fastapi import File
 from fastapi import Form
+from fastapi import HTTPException
 from fastapi import Query
 from fastapi import UploadFile
 from pydantic import BaseModel
@@ -49,8 +50,6 @@ from onyx.db.engine.sql_engine import get_session
 from onyx.db.enums import ConnectorCredentialPairStatus
 from onyx.db.models import User
 from onyx.document_index.interfaces import DocumentMetadata
-from onyx.error_handling.error_codes import OnyxErrorCode
-from onyx.error_handling.exceptions import OnyxError
 from onyx.server.features.build.configs import USER_LIBRARY_MAX_FILE_SIZE_BYTES
 from onyx.server.features.build.configs import USER_LIBRARY_MAX_FILES_PER_UPLOAD
 from onyx.server.features.build.configs import USER_LIBRARY_MAX_TOTAL_SIZE_BYTES
@@ -185,12 +184,12 @@ def _validate_zip_contents(
     """Validate zip file contents before extraction.
 
     Checks file count limit and total decompressed size against storage quota.
-    Raises OnyxError on validation failure.
+    Raises HTTPException on validation failure.
     """
     if len(zip_file.namelist()) > USER_LIBRARY_MAX_FILES_PER_UPLOAD:
-        raise OnyxError(
-            OnyxErrorCode.VALIDATION_ERROR,
-            f"Zip contains too many files. Maximum is {USER_LIBRARY_MAX_FILES_PER_UPLOAD}.",
+        raise HTTPException(
+            status_code=400,
+            detail=f"Zip contains too many files. Maximum is {USER_LIBRARY_MAX_FILES_PER_UPLOAD}.",
         )
 
     # Zip bomb protection: check total decompressed size before extracting
@@ -198,9 +197,9 @@ def _validate_zip_contents(
         info.file_size for info in zip_file.infolist() if not info.is_dir()
     )
     if existing_usage + declared_total > USER_LIBRARY_MAX_TOTAL_SIZE_BYTES:
-        raise OnyxError(
-            OnyxErrorCode.VALIDATION_ERROR,
-            (
+        raise HTTPException(
+            status_code=400,
+            detail=(
                 f"Zip decompressed size ({declared_total // (1024*1024)}MB) "
                 f"would exceed storage limit."
             ),
@@ -214,19 +213,19 @@ def _verify_ownership_and_get_document(
 ) -> Any:
     """Verify the user owns the document and return it.
 
-    Raises OnyxError on authorization failure or if document not found.
+    Raises HTTPException on authorization failure or if document not found.
     """
     from onyx.db.document import get_document
 
     user_prefix = f"CRAFT_FILE__{user.id}__"
     if not document_id.startswith(user_prefix):
-        raise OnyxError(
-            OnyxErrorCode.UNAUTHORIZED, "Not authorized to modify this file"
+        raise HTTPException(
+            status_code=403, detail="Not authorized to modify this file"
         )
 
     doc = get_document(document_id, db_session)
     if doc is None:
-        raise OnyxError(OnyxErrorCode.NOT_FOUND, "File not found")
+        raise HTTPException(status_code=404, detail="File not found")
 
     return doc
 
@@ -334,13 +333,13 @@ async def upload_files(
     """
     tenant_id = get_current_tenant_id()
     if tenant_id is None:
-        raise OnyxError(OnyxErrorCode.INTERNAL_ERROR, "Tenant ID not found")
+        raise HTTPException(status_code=500, detail="Tenant ID not found")
 
     # Validate file count
     if len(files) > USER_LIBRARY_MAX_FILES_PER_UPLOAD:
-        raise OnyxError(
-            OnyxErrorCode.VALIDATION_ERROR,
-            f"Too many files. Maximum is {USER_LIBRARY_MAX_FILES_PER_UPLOAD} per upload.",
+        raise HTTPException(
+            status_code=400,
+            detail=f"Too many files. Maximum is {USER_LIBRARY_MAX_FILES_PER_UPLOAD} per upload.",
         )
 
     # Check cumulative storage usage
@@ -371,17 +370,17 @@ async def upload_files(
 
         # Validate individual file size
         if file_size > USER_LIBRARY_MAX_FILE_SIZE_BYTES:
-            raise OnyxError(
-                OnyxErrorCode.VALIDATION_ERROR,
-                f"File '{file.filename}' exceeds maximum size of {USER_LIBRARY_MAX_FILE_SIZE_BYTES // (1024*1024)}MB",
+            raise HTTPException(
+                status_code=400,
+                detail=f"File '{file.filename}' exceeds maximum size of {USER_LIBRARY_MAX_FILE_SIZE_BYTES // (1024*1024)}MB",
             )
 
         # Validate cumulative storage (existing + this upload batch)
         total_size += file_size
         if existing_usage + total_size > USER_LIBRARY_MAX_TOTAL_SIZE_BYTES:
-            raise OnyxError(
-                OnyxErrorCode.VALIDATION_ERROR,
-                f"Total storage would exceed maximum of {USER_LIBRARY_MAX_TOTAL_SIZE_BYTES // (1024*1024*1024)}GB",
+            raise HTTPException(
+                status_code=400,
+                detail=f"Total storage would exceed maximum of {USER_LIBRARY_MAX_TOTAL_SIZE_BYTES // (1024*1024*1024)}GB",
             )
 
         # Sanitize filename
@@ -450,14 +449,14 @@ async def upload_zip(
     """
     tenant_id = get_current_tenant_id()
     if tenant_id is None:
-        raise OnyxError(OnyxErrorCode.INTERNAL_ERROR, "Tenant ID not found")
+        raise HTTPException(status_code=500, detail="Tenant ID not found")
 
     # Read zip content
     content = await file.read()
     if len(content) > USER_LIBRARY_MAX_TOTAL_SIZE_BYTES:
-        raise OnyxError(
-            OnyxErrorCode.VALIDATION_ERROR,
-            f"Zip file exceeds maximum size of {USER_LIBRARY_MAX_TOTAL_SIZE_BYTES // (1024*1024*1024)}GB",
+        raise HTTPException(
+            status_code=400,
+            detail=f"Zip file exceeds maximum size of {USER_LIBRARY_MAX_TOTAL_SIZE_BYTES // (1024*1024*1024)}GB",
         )
 
     # Check cumulative storage usage
@@ -516,9 +515,9 @@ async def upload_zip(
 
                 # Validate cumulative storage
                 if existing_usage + total_size > USER_LIBRARY_MAX_TOTAL_SIZE_BYTES:
-                    raise OnyxError(
-                        OnyxErrorCode.VALIDATION_ERROR,
-                        f"Total storage would exceed maximum of {USER_LIBRARY_MAX_TOTAL_SIZE_BYTES // (1024*1024*1024)}GB",
+                    raise HTTPException(
+                        status_code=400,
+                        detail=f"Total storage would exceed maximum of {USER_LIBRARY_MAX_TOTAL_SIZE_BYTES // (1024*1024*1024)}GB",
                     )
 
                 # Build path preserving zip structure
@@ -561,7 +560,7 @@ async def upload_zip(
                 )
 
     except zipfile.BadZipFile:
-        raise OnyxError(OnyxErrorCode.VALIDATION_ERROR, "Invalid zip file")
+        raise HTTPException(status_code=400, detail="Invalid zip file")
 
     # Create directory document records so they appear in the tree view
     if directory_paths:
@@ -675,7 +674,7 @@ def toggle_file_sync(
 
     tenant_id = get_current_tenant_id()
     if tenant_id is None:
-        raise OnyxError(OnyxErrorCode.INTERNAL_ERROR, "Tenant ID not found")
+        raise HTTPException(status_code=500, detail="Tenant ID not found")
 
     doc = _verify_ownership_and_get_document(document_id, user, db_session)
 
@@ -720,7 +719,7 @@ def delete_file(
 
     tenant_id = get_current_tenant_id()
     if tenant_id is None:
-        raise OnyxError(OnyxErrorCode.INTERNAL_ERROR, "Tenant ID not found")
+        raise HTTPException(status_code=500, detail="Tenant ID not found")
 
     doc = _verify_ownership_and_get_document(document_id, user, db_session)
 

backend/onyx/server/features/build/sandbox/kubernetes/docker/templates/outputs/web/package-lock.json

@@ -961,9 +961,9 @@
       "license": "MIT"
     },
     "node_modules/@hono/node-server": {
-      "version": "1.19.9",
-      "resolved": "https://registry.npmjs.org/@hono/node-server/-/node-server-1.19.9.tgz",
-      "integrity": "sha512-vHL6w3ecZsky+8P5MD+eFfaGTyCeOHUIFYMGpQGbrBTSmNNoxv0if69rEZ5giu36weC5saFuznL411gRX7bJDw==",
+      "version": "1.19.10",
+      "resolved": "https://registry.npmjs.org/@hono/node-server/-/node-server-1.19.10.tgz",
+      "integrity": "sha512-hZ7nOssGqRgyV3FVVQdfi+U4q02uB23bpnYpdvNXkYTRRyWx84b7yf1ans+dnJ/7h41sGL3CeQTfO+ZGxuO+Iw==",
       "license": "MIT",
       "engines": {
         "node": ">=18.14.1"
@@ -1573,27 +1573,6 @@
         }
       }
     },
-    "node_modules/@isaacs/balanced-match": {
-      "version": "4.0.1",
-      "resolved": "https://registry.npmjs.org/@isaacs/balanced-match/-/balanced-match-4.0.1.tgz",
-      "integrity": "sha512-yzMTt9lEb8Gv7zRioUilSglI0c0smZ9k5D65677DLWLtWJaXIS3CqcGyUFByYKlnUj6TkjLVs54fBl6+TiGQDQ==",
-      "license": "MIT",
-      "engines": {
-        "node": "20 || >=22"
-      }
-    },
-    "node_modules/@isaacs/brace-expansion": {
-      "version": "5.0.1",
-      "resolved": "https://registry.npmjs.org/@isaacs/brace-expansion/-/brace-expansion-5.0.1.tgz",
-      "integrity": "sha512-WMz71T1JS624nWj2n2fnYAuPovhv7EUhk69R6i9dsVyzxt5eM3bjwvgk9L+APE1TRscGysAVMANkB0jh0LQZrQ==",
-      "license": "MIT",
-      "dependencies": {
-        "@isaacs/balanced-match": "^4.0.1"
-      },
-      "engines": {
-        "node": "20 || >=22"
-      }
-    },
     "node_modules/@jridgewell/gen-mapping": {
       "version": "0.3.13",
       "resolved": "https://registry.npmjs.org/@jridgewell/gen-mapping/-/gen-mapping-0.3.13.tgz",
@@ -1680,9 +1659,9 @@
       }
     },
     "node_modules/@modelcontextprotocol/sdk/node_modules/ajv": {
-      "version": "8.17.1",
-      "resolved": "https://registry.npmjs.org/ajv/-/ajv-8.17.1.tgz",
-      "integrity": "sha512-B/gBuNg5SiMTrPkC+A2+cW0RszwxYmn6VYxB/inlBStS5nx6xHIt/ehKRhIMhqusl7a8LjQoZnjCs5vhwxOQ1g==",
+      "version": "8.18.0",
+      "resolved": "https://registry.npmjs.org/ajv/-/ajv-8.18.0.tgz",
+      "integrity": "sha512-PlXPeEWMXMZ7sPYOHqmDyCJzcfNrUr3fGNKtezX14ykXOEIvyK81d+qydx89KY5O71FKMPaQ2vBfBFI5NHR63A==",
       "license": "MIT",
       "dependencies": {
         "fast-deep-equal": "^3.1.3",
@@ -3855,6 +3834,27 @@
         "path-browserify": "^1.0.1"
       }
     },
+    "node_modules/@ts-morph/common/node_modules/balanced-match": {
+      "version": "4.0.4",
+      "resolved": "https://registry.npmjs.org/balanced-match/-/balanced-match-4.0.4.tgz",
+      "integrity": "sha512-BLrgEcRTwX2o6gGxGOCNyMvGSp35YofuYzw9h1IMTRmKqttAZZVU67bdb9Pr2vUHA8+j3i2tJfjO6C6+4myGTA==",
+      "license": "MIT",
+      "engines": {
+        "node": "18 || 20 || >=22"
+      }
+    },
+    "node_modules/@ts-morph/common/node_modules/brace-expansion": {
+      "version": "5.0.3",
+      "resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-5.0.3.tgz",
+      "integrity": "sha512-fy6KJm2RawA5RcHkLa1z/ScpBeA762UF9KmZQxwIbDtRJrgLzM10depAiEQ+CXYcoiqW1/m96OAAoke2nE9EeA==",
+      "license": "MIT",
+      "dependencies": {
+        "balanced-match": "^4.0.2"
+      },
+      "engines": {
+        "node": "18 || 20 || >=22"
+      }
+    },
     "node_modules/@ts-morph/common/node_modules/fast-glob": {
       "version": "3.3.3",
       "resolved": "https://registry.npmjs.org/fast-glob/-/fast-glob-3.3.3.tgz",
@@ -3884,15 +3884,15 @@
       }
     },
     "node_modules/@ts-morph/common/node_modules/minimatch": {
-      "version": "10.1.1",
-      "resolved": "https://registry.npmjs.org/minimatch/-/minimatch-10.1.1.tgz",
-      "integrity": "sha512-enIvLvRAFZYXJzkCYG5RKmPfrFArdLv+R+lbQ53BmIMLIry74bjKzX6iHAm8WYamJkhSSEabrWN5D97XnKObjQ==",
+      "version": "10.2.4",
+      "resolved": "https://registry.npmjs.org/minimatch/-/minimatch-10.2.4.tgz",
+      "integrity": "sha512-oRjTw/97aTBN0RHbYCdtF1MQfvusSIBQM0IZEgzl6426+8jSC0nF1a/GmnVLpfB9yyr6g6FTqWqiZVbxrtaCIg==",
       "license": "BlueOak-1.0.0",
       "dependencies": {
-        "@isaacs/brace-expansion": "^5.0.0"
+        "brace-expansion": "^5.0.2"
       },
       "engines": {
-        "node": "20 || >=22"
+        "node": "18 || 20 || >=22"
       },
       "funding": {
         "url": "https://github.com/sponsors/isaacs"
@@ -4234,13 +4234,13 @@
       }
     },
     "node_modules/@typescript-eslint/typescript-estree/node_modules/minimatch": {
-      "version": "9.0.5",
-      "resolved": "https://registry.npmjs.org/minimatch/-/minimatch-9.0.5.tgz",
-      "integrity": "sha512-G6T0ZX48xgozx7587koeX9Ys2NYy6Gmv//P89sEte9V9whIapMNF4idKxnW2QtCcLiTWlb/wfCabAtAFWhhBow==",
+      "version": "9.0.9",
+      "resolved": "https://registry.npmjs.org/minimatch/-/minimatch-9.0.9.tgz",
+      "integrity": "sha512-OBwBN9AL4dqmETlpS2zasx+vTeWclWzkblfZk7KTA5j3jeOONz/tRCnZomUyvNg83wL5Zv9Ss6HMJXAgL8R2Yg==",
       "dev": true,
       "license": "ISC",
       "dependencies": {
-        "brace-expansion": "^2.0.1"
+        "brace-expansion": "^2.0.2"
       },
       "engines": {
         "node": ">=16 || 14 >=14.17"
@@ -4619,9 +4619,9 @@
       }
     },
     "node_modules/ajv": {
-      "version": "6.12.6",
-      "resolved": "https://registry.npmjs.org/ajv/-/ajv-6.12.6.tgz",
-      "integrity": "sha512-j3fVLgvTo527anyYyJOGTYJbG+vnnQYvE0m5mmkc1TK+nxAppkCLMIL0aZ4dblVCNoGShhm+kzE4ZUykBoMg4g==",
+      "version": "6.14.0",
+      "resolved": "https://registry.npmjs.org/ajv/-/ajv-6.14.0.tgz",
+      "integrity": "sha512-IWrosm/yrn43eiKqkfkHis7QioDleaXQHdDVPKg0FSwwd/DuvyX79TZnFOnYpB7dcsFAMmtFztZuXPDvSePkFw==",
       "dev": true,
       "license": "MIT",
       "dependencies": {
@@ -4653,9 +4653,9 @@
       }
     },
     "node_modules/ajv-formats/node_modules/ajv": {
-      "version": "8.17.1",
-      "resolved": "https://registry.npmjs.org/ajv/-/ajv-8.17.1.tgz",
-      "integrity": "sha512-B/gBuNg5SiMTrPkC+A2+cW0RszwxYmn6VYxB/inlBStS5nx6xHIt/ehKRhIMhqusl7a8LjQoZnjCs5vhwxOQ1g==",
+      "version": "8.18.0",
+      "resolved": "https://registry.npmjs.org/ajv/-/ajv-8.18.0.tgz",
+      "integrity": "sha512-PlXPeEWMXMZ7sPYOHqmDyCJzcfNrUr3fGNKtezX14ykXOEIvyK81d+qydx89KY5O71FKMPaQ2vBfBFI5NHR63A==",
       "license": "MIT",
       "dependencies": {
         "fast-deep-equal": "^3.1.3",
@@ -6758,12 +6758,12 @@
       }
     },
     "node_modules/express-rate-limit": {
-      "version": "8.2.1",
-      "resolved": "https://registry.npmjs.org/express-rate-limit/-/express-rate-limit-8.2.1.tgz",
-      "integrity": "sha512-PCZEIEIxqwhzw4KF0n7QF4QqruVTcF73O5kFKUnGOyjbCCgizBBiFaYpd/fnBLUMPw/BWw9OsiN7GgrNYr7j6g==",
+      "version": "8.3.0",
+      "resolved": "https://registry.npmjs.org/express-rate-limit/-/express-rate-limit-8.3.0.tgz",
+      "integrity": "sha512-KJzBawY6fB9FiZGdE/0aftepZ91YlaGIrV8vgblRM3J8X+dHx/aiowJWwkx6LIGyuqGiANsjSwwrbb8mifOJ4Q==",
       "license": "MIT",
       "dependencies": {
-        "ip-address": "10.0.1"
+        "ip-address": "10.1.0"
       },
       "engines": {
         "node": ">= 16"
@@ -7556,9 +7556,9 @@
       }
     },
     "node_modules/ip-address": {
-      "version": "10.0.1",
-      "resolved": "https://registry.npmjs.org/ip-address/-/ip-address-10.0.1.tgz",
-      "integrity": "sha512-NWv9YLW4PoW2B7xtzaS3NCot75m6nK7Icdv0o3lfMceJVRfSoQwqD4wEH5rLwoKJwUiZ/rfpiVBhnaF0FK4HoA==",
+      "version": "10.1.0",
+      "resolved": "https://registry.npmjs.org/ip-address/-/ip-address-10.1.0.tgz",
+      "integrity": "sha512-XXADHxXmvT9+CRxhXg56LJovE+bmWnEWB78LB83VZTprKTmaC5QfruXocxzTZ2Kl0DNwKuBdlIhjL8LeY8Sf8Q==",
       "license": "MIT",
       "engines": {
         "node": ">= 12"
@@ -8831,9 +8831,9 @@
       }
     },
     "node_modules/minimatch": {
-      "version": "3.1.2",
-      "resolved": "https://registry.npmjs.org/minimatch/-/minimatch-3.1.2.tgz",
-      "integrity": "sha512-J7p63hRiAjw1NDEww1W7i37+ByIrOWO5XQQAzZ3VOcL0PNybwpfmV/N05zFAzwQ9USyEcX6t3UO+K5aqBQOIHw==",
+      "version": "3.1.5",
+      "resolved": "https://registry.npmjs.org/minimatch/-/minimatch-3.1.5.tgz",
+      "integrity": "sha512-VgjWUsnnT6n+NUk6eZq77zeFdpW2LWDzP6zFGrCbHXiYNul5Dzqk2HHQ5uFH2DNW5Xbp8+jVzaeNt94ssEEl4w==",
       "dev": true,
       "license": "ISC",
       "dependencies": {
@@ -9699,9 +9699,9 @@
       }
     },
     "node_modules/qs": {
-      "version": "6.14.1",
-      "resolved": "https://registry.npmjs.org/qs/-/qs-6.14.1.tgz",
-      "integrity": "sha512-4EK3+xJl8Ts67nLYNwqw/dsFVnCf+qR7RgXSK9jEEm9unao3njwMDdmsdvoKBKHzxd7tCYz5e5M+SnMjdtXGQQ==",
+      "version": "6.14.2",
+      "resolved": "https://registry.npmjs.org/qs/-/qs-6.14.2.tgz",
+      "integrity": "sha512-V/yCWTTF7VJ9hIh18Ugr2zhJMP01MY7c5kh4J870L7imm6/DIzBsNLTXzMwUA3yZ5b/KBqLx8Kp3uRvd7xSe3Q==",
       "license": "BSD-3-Clause",
       "dependencies": {
         "side-channel": "^1.1.0"

backend/onyx/server/features/projects/projects_file_utils.py

@@ -10,6 +10,8 @@ from pydantic import Field
 from sqlalchemy.orm import Session
 
 from onyx.configs.app_configs import FILE_TOKEN_COUNT_THRESHOLD
+from onyx.configs.app_configs import USER_FILE_MAX_UPLOAD_SIZE_BYTES
+from onyx.configs.app_configs import USER_FILE_MAX_UPLOAD_SIZE_MB
 from onyx.db.llm import fetch_default_llm_model
 from onyx.file_processing.extract_file_text import extract_file_text
 from onyx.file_processing.extract_file_text import get_file_ext
@@ -35,6 +37,38 @@ def get_safe_filename(upload: UploadFile) -> str:
     return upload.filename
 
 
+def get_upload_size_bytes(upload: UploadFile) -> int | None:
+    """Best-effort file size in bytes without consuming the stream."""
+    if upload.size is not None:
+        return upload.size
+
+    try:
+        current_pos = upload.file.tell()
+        upload.file.seek(0, 2)
+        size = upload.file.tell()
+        upload.file.seek(current_pos)
+        return size
+    except Exception as e:
+        logger.warning(
+            "Could not determine upload size via stream seek "
+            f"(filename='{get_safe_filename(upload)}', "
+            f"error_type={type(e).__name__}, error={e})"
+        )
+        return None
+
+
+def is_upload_too_large(upload: UploadFile, max_bytes: int) -> bool:
+    """Return True when upload size is known and exceeds max_bytes."""
+    size_bytes = get_upload_size_bytes(upload)
+    if size_bytes is None:
+        logger.warning(
+            "Could not determine upload size; skipping size-limit check for "
+            f"'{get_safe_filename(upload)}'"
+        )
+        return False
+    return size_bytes > max_bytes
+
+
 # Guard against extremely large images
 Image.MAX_IMAGE_PIXELS = 12000 * 12000
 
@@ -159,6 +193,18 @@ def categorize_uploaded_files(
     for upload in files:
         try:
             filename = get_safe_filename(upload)
+
+            # Size limit is a hard safety cap and is enforced even when token
+            # threshold checks are skipped via SKIP_USERFILE_THRESHOLD settings.
+            if is_upload_too_large(upload, USER_FILE_MAX_UPLOAD_SIZE_BYTES):
+                results.rejected.append(
+                    RejectedFile(
+                        filename=filename,
+                        reason=f"Exceeds {USER_FILE_MAX_UPLOAD_SIZE_MB} MB file size limit",
+                    )
+                )
+                continue
+
             extension = get_file_ext(filename)
 
             # If image, estimate tokens via dedicated method first

backend/onyx/server/manage/llm/api.py

@@ -48,6 +48,7 @@ from onyx.llm.utils import test_llm
 from onyx.llm.well_known_providers.auto_update_service import (
     fetch_llm_recommendations_from_github,
 )
+from onyx.llm.well_known_providers.constants import LM_STUDIO_API_KEY_CONFIG_KEY
 from onyx.llm.well_known_providers.llm_provider_options import (
     fetch_available_well_known_llms,
 )
@@ -62,6 +63,8 @@ from onyx.server.manage.llm.models import LLMProviderDescriptor
 from onyx.server.manage.llm.models import LLMProviderResponse
 from onyx.server.manage.llm.models import LLMProviderUpsertRequest
 from onyx.server.manage.llm.models import LLMProviderView
+from onyx.server.manage.llm.models import LMStudioFinalModelResponse
+from onyx.server.manage.llm.models import LMStudioModelsRequest
 from onyx.server.manage.llm.models import OllamaFinalModelResponse
 from onyx.server.manage.llm.models import OllamaModelDetails
 from onyx.server.manage.llm.models import OllamaModelsRequest
@@ -73,6 +76,7 @@ from onyx.server.manage.llm.models import VisionProviderResponse
 from onyx.server.manage.llm.utils import generate_bedrock_display_name
 from onyx.server.manage.llm.utils import generate_ollama_display_name
 from onyx.server.manage.llm.utils import infer_vision_support
+from onyx.server.manage.llm.utils import is_reasoning_model
 from onyx.server.manage.llm.utils import is_valid_bedrock_model
 from onyx.server.manage.llm.utils import ModelMetadata
 from onyx.server.manage.llm.utils import strip_openrouter_vendor_prefix
@@ -441,6 +445,17 @@ def put_llm_provider(
         not existing_provider or not existing_provider.is_auto_mode
     )
 
+    # Before the upsert, check if this provider currently owns the global
+    # CHAT default. The upsert may cascade-delete model_configurations
+    # (and their flow mappings), so we need to remember this beforehand.
+    was_default_provider = False
+    if existing_provider and transitioning_to_auto_mode:
+        current_default = fetch_default_llm_model(db_session)
+        was_default_provider = (
+            current_default is not None
+            and current_default.llm_provider_id == existing_provider.id
+        )
+
     try:
         result = upsert_llm_provider(
             llm_provider_upsert_request=llm_provider_upsert_request,
@@ -463,6 +478,20 @@ def put_llm_provider(
                         updated_provider,
                         config,
                     )
+
+                    # If this provider was the default before the transition,
+                    # restore the default using the recommended model.
+                    if was_default_provider:
+                        recommended = config.get_default_model(
+                            llm_provider_upsert_request.provider
+                        )
+                        if recommended:
+                            update_default_provider(
+                                provider_id=updated_provider.id,
+                                model_name=recommended.name,
+                                db_session=db_session,
+                            )
+
                     # Refresh result with synced models
                     result = LLMProviderView.from_model(updated_provider)
 
@@ -1217,3 +1246,117 @@ def get_openrouter_available_models(
             logger.warning(f"Failed to sync OpenRouter models to DB: {e}")
 
     return sorted_results
+
+
+@admin_router.post("/lm-studio/available-models")
+def get_lm_studio_available_models(
+    request: LMStudioModelsRequest,
+    _: User = Depends(current_admin_user),
+    db_session: Session = Depends(get_session),
+) -> list[LMStudioFinalModelResponse]:
+    """Fetch available models from an LM Studio server.
+
+    Uses the LM Studio-native /api/v1/models endpoint which exposes
+    rich metadata including capabilities (vision, reasoning),
+    display names, and context lengths.
+    """
+    cleaned_api_base = request.api_base.strip().rstrip("/")
+    # Strip /v1 suffix that users may copy from OpenAI-compatible tool configs;
+    # the native metadata endpoint lives at /api/v1/models, not /v1/api/v1/models.
+    cleaned_api_base = cleaned_api_base.removesuffix("/v1")
+    if not cleaned_api_base:
+        raise OnyxError(
+            OnyxErrorCode.VALIDATION_ERROR,
+            "API base URL is required to fetch LM Studio models.",
+        )
+
+    # If provider_name is given and the api_key hasn't been changed by the user,
+    # fall back to the stored API key from the database (the form value is masked).
+    api_key = request.api_key
+    if request.provider_name and not request.api_key_changed:
+        existing_provider = fetch_existing_llm_provider(
+            name=request.provider_name, db_session=db_session
+        )
+        if existing_provider and existing_provider.custom_config:
+            api_key = existing_provider.custom_config.get(LM_STUDIO_API_KEY_CONFIG_KEY)
+
+    url = f"{cleaned_api_base}/api/v1/models"
+    headers: dict[str, str] = {}
+    if api_key:
+        headers["Authorization"] = f"Bearer {api_key}"
+
+    try:
+        response = httpx.get(url, headers=headers, timeout=10.0)
+        response.raise_for_status()
+        response_json = response.json()
+    except Exception as e:
+        raise OnyxError(
+            OnyxErrorCode.BAD_GATEWAY,
+            f"Failed to fetch LM Studio models: {e}",
+        )
+
+    models = response_json.get("models", [])
+    if not isinstance(models, list) or len(models) == 0:
+        raise OnyxError(
+            OnyxErrorCode.VALIDATION_ERROR,
+            "No models found from your LM Studio server.",
+        )
+
+    results: list[LMStudioFinalModelResponse] = []
+    for item in models:
+        # Filter to LLM-type models only (skip embeddings, etc.)
+        if item.get("type") != "llm":
+            continue
+
+        model_key = item.get("key")
+        if not model_key:
+            continue
+
+        display_name = item.get("display_name") or model_key
+        max_context_length = item.get("max_context_length")
+        capabilities = item.get("capabilities") or {}
+
+        results.append(
+            LMStudioFinalModelResponse(
+                name=model_key,
+                display_name=display_name,
+                max_input_tokens=max_context_length,
+                supports_image_input=capabilities.get("vision", False),
+                supports_reasoning=capabilities.get("reasoning", False)
+                or is_reasoning_model(model_key, display_name),
+            )
+        )
+
+    if not results:
+        raise OnyxError(
+            OnyxErrorCode.VALIDATION_ERROR,
+            "No compatible models found from LM Studio server.",
+        )
+
+    sorted_results = sorted(results, key=lambda m: m.name.lower())
+
+    # Sync new models to DB if provider_name is specified
+    if request.provider_name:
+        try:
+            models_to_sync = [
+                {
+                    "name": r.name,
+                    "display_name": r.display_name,
+                    "max_input_tokens": r.max_input_tokens,
+                    "supports_image_input": r.supports_image_input,
+                }
+                for r in sorted_results
+            ]
+            new_count = sync_model_configurations(
+                db_session=db_session,
+                provider_name=request.provider_name,
+                models=models_to_sync,
+            )
+            if new_count > 0:
+                logger.info(
+                    f"Added {new_count} new LM Studio models to provider '{request.provider_name}'"
+                )
+        except ValueError as e:
+            logger.warning(f"Failed to sync LM Studio models to DB: {e}")
+
+    return sorted_results

backend/onyx/server/manage/llm/models.py

@@ -371,6 +371,22 @@ class OpenRouterFinalModelResponse(BaseModel):
     supports_image_input: bool
 
 
+# LM Studio dynamic models fetch
+class LMStudioModelsRequest(BaseModel):
+    api_base: str
+    api_key: str | None = None
+    api_key_changed: bool = False
+    provider_name: str | None = None  # Optional: to save models to existing provider
+
+
+class LMStudioFinalModelResponse(BaseModel):
+    name: str  # Model ID from LM Studio (e.g., "lmstudio-community/Meta-Llama-3-8B")
+    display_name: str  # Human-readable name
+    max_input_tokens: int | None  # From LM Studio API or None if unavailable
+    supports_image_input: bool
+    supports_reasoning: bool
+
+
 class DefaultModel(BaseModel):
     provider_id: int
     model_name: str

backend/onyx/server/manage/llm/utils.py

@@ -12,6 +12,7 @@ from typing import TypedDict
 
 from onyx.llm.constants import BEDROCK_MODEL_NAME_MAPPINGS
 from onyx.llm.constants import LlmProviderNames
+from onyx.llm.constants import MODEL_PREFIX_TO_VENDOR
 from onyx.llm.constants import OLLAMA_MODEL_NAME_MAPPINGS
 from onyx.llm.constants import OLLAMA_MODEL_TO_VENDOR
 from onyx.llm.constants import PROVIDER_DISPLAY_NAMES
@@ -23,6 +24,7 @@ DYNAMIC_LLM_PROVIDERS = frozenset(
         LlmProviderNames.OPENROUTER,
         LlmProviderNames.BEDROCK,
         LlmProviderNames.OLLAMA_CHAT,
+        LlmProviderNames.LM_STUDIO,
     }
 )
 
@@ -348,4 +350,19 @@ def extract_vendor_from_model_name(model_name: str, provider: str) -> str | None
         # Fallback: capitalize the base name as vendor
         return base_name.split("-")[0].title()
 
+    elif provider == LlmProviderNames.LM_STUDIO:
+        # LM Studio model IDs can be paths like "publisher/model-name"
+        # or simple names. Use MODEL_PREFIX_TO_VENDOR for matching.
+
+        model_lower = model_name.lower()
+        # Check for slash-separated vendor prefix first
+        if "/" in model_lower:
+            vendor_key = model_lower.split("/")[0]
+            return PROVIDER_DISPLAY_NAMES.get(vendor_key, vendor_key.title())
+        # Fallback to model prefix matching
+        for prefix, vendor in MODEL_PREFIX_TO_VENDOR.items():
+            if model_lower.startswith(prefix):
+                return PROVIDER_DISPLAY_NAMES.get(vendor, vendor.title())
+        return None
+
     return None

backend/onyx/server/query_and_chat/streaming_models.py

@@ -41,6 +41,7 @@ class StreamingType(Enum):
     REASONING_DONE = "reasoning_done"
     CITATION_INFO = "citation_info"
     TOOL_CALL_DEBUG = "tool_call_debug"
+    TOOL_CALL_ARGUMENT_DELTA = "tool_call_argument_delta"
 
     MEMORY_TOOL_START = "memory_tool_start"
     MEMORY_TOOL_DELTA = "memory_tool_delta"
@@ -259,6 +260,15 @@ class CustomToolDelta(BaseObj):
     file_ids: list[str] | None = None
 
 
+class ToolCallArgumentDelta(BaseObj):
+    type: Literal["tool_call_argument_delta"] = (
+        StreamingType.TOOL_CALL_ARGUMENT_DELTA.value
+    )
+
+    tool_type: str
+    argument_deltas: dict[str, Any]
+
+
 ################################################
 # File Reader Packets
 ################################################
@@ -379,6 +389,7 @@ PacketObj = Union[
     # Citation Packets
     CitationInfo,
     ToolCallDebug,
+    ToolCallArgumentDelta,
     # Deep Research Packets
     DeepResearchPlanStart,
     DeepResearchPlanDelta,

backend/onyx/server/settings/models.py

@@ -78,6 +78,7 @@ class Settings(BaseModel):
 
     # User Knowledge settings
     user_knowledge_enabled: bool | None = True
+    user_file_max_upload_size_mb: int | None = None
 
     # Connector settings
     show_extra_connectors: bool | None = True

backend/onyx/server/settings/store.py

@@ -3,6 +3,7 @@ from onyx.configs.app_configs import DISABLE_USER_KNOWLEDGE
 from onyx.configs.app_configs import ENABLE_OPENSEARCH_INDEXING_FOR_ONYX
 from onyx.configs.app_configs import ONYX_QUERY_HISTORY_TYPE
 from onyx.configs.app_configs import SHOW_EXTRA_CONNECTORS
+from onyx.configs.app_configs import USER_FILE_MAX_UPLOAD_SIZE_MB
 from onyx.configs.constants import KV_SETTINGS_KEY
 from onyx.configs.constants import OnyxRedisLocks
 from onyx.key_value_store.factory import get_kv_store
@@ -50,6 +51,7 @@ def load_settings() -> Settings:
     if DISABLE_USER_KNOWLEDGE:
         settings.user_knowledge_enabled = False
 
+    settings.user_file_max_upload_size_mb = USER_FILE_MAX_UPLOAD_SIZE_MB
     settings.show_extra_connectors = SHOW_EXTRA_CONNECTORS
     settings.opensearch_indexing_enabled = ENABLE_OPENSEARCH_INDEXING_FOR_ONYX
     return settings

backend/onyx/tools/built_in_tools.py

@@ -56,3 +56,23 @@ def get_built_in_tool_ids() -> list[str]:
 
 def get_built_in_tool_by_id(in_code_tool_id: str) -> Type[BUILT_IN_TOOL_TYPES]:
     return BUILT_IN_TOOL_MAP[in_code_tool_id]
+
+
+def _build_tool_name_to_class() -> dict[str, Type[BUILT_IN_TOOL_TYPES]]:
+    """Build a mapping from LLM-facing tool name to tool class."""
+    result: dict[str, Type[BUILT_IN_TOOL_TYPES]] = {}
+    for cls in BUILT_IN_TOOL_MAP.values():
+        name_attr = cls.__dict__.get("name")
+        if isinstance(name_attr, property) and name_attr.fget is not None:
+            tool_name = name_attr.fget(cls)
+        elif isinstance(name_attr, str):
+            tool_name = name_attr
+        else:
+            raise ValueError(
+                f"Built-in tool {cls.__name__} must define a valid LLM-facing tool name"
+            )
+        result[tool_name] = cls
+    return result
+
+
+TOOL_NAME_TO_CLASS: dict[str, Type[BUILT_IN_TOOL_TYPES]] = _build_tool_name_to_class()

backend/onyx/tools/interface.py

@@ -92,3 +92,7 @@ class Tool(abc.ABC, Generic[TOverride]):
         **llm_kwargs: Any,
     ) -> ToolResponse:
         raise NotImplementedError
+
+    @classmethod
+    def should_emit_argument_deltas(cls) -> bool:
+        return False

backend/onyx/tools/tool_implementations/open_url/snippet_matcher.py

@@ -111,19 +111,26 @@ def _normalize_text_with_mapping(text: str) -> tuple[str, list[int]]:
     # Step 1: NFC normalization with position mapping
     nfc_text = unicodedata.normalize("NFC", text)
 
-    # Build mapping from NFC positions to original start positions
+    # Map NFD positions → original positions.
+    # NFD only decomposes, so each original char produces 1+ NFD chars.
+    nfd_to_orig: list[int] = []
+    for orig_idx, orig_char in enumerate(original_text):
+        nfd_of_char = unicodedata.normalize("NFD", orig_char)
+        for _ in nfd_of_char:
+            nfd_to_orig.append(orig_idx)
+
+    # Map NFC positions → NFD positions.
+    # Each NFC char, when decomposed, tells us exactly how many NFD
+    # chars it was composed from.
     nfc_to_orig: list[int] = []
-    orig_idx = 0
+    nfd_idx = 0
     for nfc_char in nfc_text:
-        nfc_to_orig.append(orig_idx)
-        # Find how many original chars contributed to this NFC char
-        for length in range(1, len(original_text) - orig_idx + 1):
-            substr = original_text[orig_idx : orig_idx + length]
-            if unicodedata.normalize("NFC", substr) == nfc_char:
-                orig_idx += length
-                break
+        if nfd_idx < len(nfd_to_orig):
+            nfc_to_orig.append(nfd_to_orig[nfd_idx])
         else:
-            orig_idx += 1  # Fallback
+            nfc_to_orig.append(len(original_text) - 1)
+        nfd_of_nfc = unicodedata.normalize("NFD", nfc_char)
+        nfd_idx += len(nfd_of_nfc)
 
     # Work with NFC text from here
     text = nfc_text

backend/onyx/tools/tool_implementations/python/python_tool.py

@@ -376,3 +376,8 @@ class PythonTool(Tool[PythonToolOverrideKwargs]):
                     rich_response=None,
                     llm_facing_response=llm_response,
                 )
+
+    @classmethod
+    @override
+    def should_emit_argument_deltas(cls) -> bool:
+        return True

backend/onyx/utils/encryption.py

@@ -11,16 +11,20 @@ logger = setup_logger()
 
 
 # IMPORTANT DO NOT DELETE, THIS IS USED BY fetch_versioned_implementation
-def _encrypt_string(input_str: str) -> bytes:
+def _encrypt_string(input_str: str, key: str | None = None) -> bytes:  # noqa: ARG001
     if ENCRYPTION_KEY_SECRET:
         logger.warning("MIT version of Onyx does not support encryption of secrets.")
+    elif key is not None:
+        logger.debug("MIT encrypt called with explicit key — key ignored.")
     return input_str.encode()
 
 
 # IMPORTANT DO NOT DELETE, THIS IS USED BY fetch_versioned_implementation
-def _decrypt_bytes(input_bytes: bytes) -> str:
-    # No need to double warn. If you wish to learn more about encryption features
-    # refer to the Onyx EE code
+def _decrypt_bytes(input_bytes: bytes, key: str | None = None) -> str:  # noqa: ARG001
+    if ENCRYPTION_KEY_SECRET:
+        logger.warning("MIT version of Onyx does not support decryption of secrets.")
+    elif key is not None:
+        logger.debug("MIT decrypt called with explicit key — key ignored.")
     return input_bytes.decode()
 
 
@@ -86,15 +90,15 @@ def _mask_list(items: list[Any]) -> list[Any]:
     return masked
 
 
-def encrypt_string_to_bytes(intput_str: str) -> bytes:
+def encrypt_string_to_bytes(intput_str: str, key: str | None = None) -> bytes:
     versioned_encryption_fn = fetch_versioned_implementation(
         "onyx.utils.encryption", "_encrypt_string"
     )
-    return versioned_encryption_fn(intput_str)
+    return versioned_encryption_fn(intput_str, key=key)
 
 
-def decrypt_bytes_to_string(intput_bytes: bytes) -> str:
+def decrypt_bytes_to_string(intput_bytes: bytes, key: str | None = None) -> str:
     versioned_decryption_fn = fetch_versioned_implementation(
         "onyx.utils.encryption", "_decrypt_bytes"
     )
-    return versioned_decryption_fn(intput_bytes)
+    return versioned_decryption_fn(intput_bytes, key=key)

backend/onyx/utils/jsonriver/__init__.py

@@ -0,0 +1,17 @@
+"""
+jsonriver - A streaming JSON parser for Python
+
+Parse JSON incrementally as it streams in, e.g. from a network request or a language model.
+Gives you a sequence of increasingly complete values.
+
+Copyright (c) 2023 Google LLC (original TypeScript implementation)
+Copyright (c) 2024 jsonriver-python contributors (Python port)
+SPDX-License-Identifier: BSD-3-Clause
+"""
+
+from .parse import _Parser as Parser
+from .parse import JsonObject
+from .parse import JsonValue
+
+__all__ = ["Parser", "JsonValue", "JsonObject"]
+__version__ = "0.0.1"

backend/onyx/utils/jsonriver/parse.py

@@ -0,0 +1,427 @@
+"""
+JSON parser for streaming incremental parsing
+
+Copyright (c) 2023 Google LLC (original TypeScript implementation)
+Copyright (c) 2024 jsonriver-python contributors (Python port)
+SPDX-License-Identifier: BSD-3-Clause
+"""
+
+from __future__ import annotations
+
+import copy
+from enum import IntEnum
+from typing import cast
+from typing import Union
+
+from .tokenize import _Input
+from .tokenize import json_token_type_to_string
+from .tokenize import JsonTokenType
+from .tokenize import Tokenizer
+
+
+# Type definitions for JSON values
+JsonValue = Union[None, bool, float, str, list["JsonValue"], dict[str, "JsonValue"]]
+JsonObject = dict[str, JsonValue]
+
+
+class _StateEnum(IntEnum):
+    """Parser state machine states"""
+
+    Initial = 0
+    InString = 1
+    InArray = 2
+    InObjectExpectingKey = 3
+    InObjectExpectingValue = 4
+
+
+class _State:
+    """Base class for parser states"""
+
+    type: _StateEnum
+    value: JsonValue | tuple[str, JsonObject] | None
+
+
+class _InitialState(_State):
+    """Initial state before any parsing"""
+
+    def __init__(self) -> None:
+        self.type = _StateEnum.Initial
+        self.value = None
+
+
+class _InStringState(_State):
+    """State while parsing a string"""
+
+    def __init__(self) -> None:
+        self.type = _StateEnum.InString
+        self.value = ""
+
+
+class _InArrayState(_State):
+    """State while parsing an array"""
+
+    def __init__(self) -> None:
+        self.type = _StateEnum.InArray
+        self.value: list[JsonValue] = []
+
+
+class _InObjectExpectingKeyState(_State):
+    """State while parsing an object, expecting a key"""
+
+    def __init__(self) -> None:
+        self.type = _StateEnum.InObjectExpectingKey
+        self.value: JsonObject = {}
+
+
+class _InObjectExpectingValueState(_State):
+    """State while parsing an object, expecting a value"""
+
+    def __init__(self, key: str, obj: JsonObject) -> None:
+        self.type = _StateEnum.InObjectExpectingValue
+        self.value = (key, obj)
+
+
+# Sentinel value to distinguish "not set" from "set to None/null"
+class _Unset:
+    pass
+
+
+_UNSET = _Unset()
+
+
+class _Parser:
+    """
+    Incremental JSON parser
+
+    Feed chunks of JSON text via feed() and get back progressively
+    more complete JSON values.
+    """
+
+    def __init__(self) -> None:
+        self._state_stack: list[_State] = [_InitialState()]
+        self._toplevel_value: JsonValue | _Unset = _UNSET
+        self._input = _Input()
+        self.tokenizer = Tokenizer(self._input, self)
+        self._finished = False
+        self._progressed = False
+        self._prev_snapshot: JsonValue | _Unset = _UNSET
+
+    def feed(self, chunk: str) -> list[JsonValue]:
+        """
+        Feed a chunk of JSON text and return deltas from the previous state.
+
+        Each element in the returned list represents what changed since the
+        last yielded value. For dicts, only changed/new keys are included,
+        with string values containing only the newly appended characters.
+        """
+        if self._finished:
+            return []
+
+        self._input.feed(chunk)
+        return self._collect_deltas()
+
+    @staticmethod
+    def _compute_delta(prev: JsonValue | None, current: JsonValue) -> JsonValue | None:
+        if prev is None:
+            return current
+
+        if isinstance(current, dict) and isinstance(prev, dict):
+            result: JsonObject = {}
+            for key in current:
+                cur_val = current[key]
+                prev_val = prev.get(key)
+                if key not in prev:
+                    result[key] = cur_val
+                elif isinstance(cur_val, str) and isinstance(prev_val, str):
+                    if cur_val != prev_val:
+                        result[key] = cur_val[len(prev_val) :]
+                elif isinstance(cur_val, list) and isinstance(prev_val, list):
+                    if cur_val != prev_val:
+                        new_items = cur_val[len(prev_val) :]
+                        # check if the last existing element was updated
+                        if (
+                            prev_val
+                            and len(cur_val) >= len(prev_val)
+                            and cur_val[len(prev_val) - 1] != prev_val[-1]
+                        ):
+                            result[key] = [cur_val[len(prev_val) - 1]] + new_items
+                        elif new_items:
+                            result[key] = new_items
+                elif cur_val != prev_val:
+                    result[key] = cur_val
+            return result if result else None
+
+        if isinstance(current, str) and isinstance(prev, str):
+            delta = current[len(prev) :]
+            return delta if delta else None
+
+        if isinstance(current, list) and isinstance(prev, list):
+            if current != prev:
+                new_items = current[len(prev) :]
+                if (
+                    prev
+                    and len(current) >= len(prev)
+                    and current[len(prev) - 1] != prev[-1]
+                ):
+                    return [current[len(prev) - 1]] + new_items
+                return new_items if new_items else None
+            return None
+
+        if current != prev:
+            return current
+        return None
+
+    def finish(self) -> list[JsonValue]:
+        """Signal that no more chunks will be fed. Validates trailing content.
+
+        Returns any final deltas produced by flushing pending tokens (e.g.
+        numbers, which have no terminator and wait for more input).
+        """
+        self._input.mark_complete()
+        # Pump once more so the tokenizer can emit tokens that were waiting
+        # for more input (e.g. numbers need buffer_complete to finalize).
+        results = self._collect_deltas()
+        self._input.expect_end_of_content()
+        return results
+
+    def _collect_deltas(self) -> list[JsonValue]:
+        """Run one pump cycle and return any deltas produced."""
+        results: list[JsonValue] = []
+        while True:
+            self._progressed = False
+            self.tokenizer.pump()
+
+            if self._progressed:
+                if self._toplevel_value is _UNSET:
+                    raise RuntimeError(
+                        "Internal error: toplevel_value should not be unset "
+                        "after progressing"
+                    )
+                current = copy.deepcopy(cast(JsonValue, self._toplevel_value))
+                if isinstance(self._prev_snapshot, _Unset):
+                    results.append(current)
+                else:
+                    delta = self._compute_delta(self._prev_snapshot, current)
+                    if delta is not None:
+                        results.append(delta)
+                self._prev_snapshot = current
+            else:
+                if not self._state_stack:
+                    self._finished = True
+                break
+        return results
+
+    # TokenHandler protocol implementation
+
+    def handle_null(self) -> None:
+        """Handle null token"""
+        self._handle_value_token(JsonTokenType.Null, None)
+
+    def handle_boolean(self, value: bool) -> None:
+        """Handle boolean token"""
+        self._handle_value_token(JsonTokenType.Boolean, value)
+
+    def handle_number(self, value: float) -> None:
+        """Handle number token"""
+        self._handle_value_token(JsonTokenType.Number, value)
+
+    def handle_string_start(self) -> None:
+        """Handle string start token"""
+        state = self._current_state()
+        if not self._progressed and state.type != _StateEnum.InObjectExpectingKey:
+            self._progressed = True
+
+        if state.type == _StateEnum.Initial:
+            self._state_stack.pop()
+            self._toplevel_value = self._progress_value(JsonTokenType.StringStart, None)
+
+        elif state.type == _StateEnum.InArray:
+            v = self._progress_value(JsonTokenType.StringStart, None)
+            arr = cast(list[JsonValue], state.value)
+            arr.append(v)
+
+        elif state.type == _StateEnum.InObjectExpectingKey:
+            self._state_stack.append(_InStringState())
+
+        elif state.type == _StateEnum.InObjectExpectingValue:
+            key, obj = cast(tuple[str, JsonObject], state.value)
+            sv = self._progress_value(JsonTokenType.StringStart, None)
+            obj[key] = sv
+
+        elif state.type == _StateEnum.InString:
+            raise ValueError(
+                f"Unexpected {json_token_type_to_string(JsonTokenType.StringStart)} "
+                f"token in the middle of string"
+            )
+
+    def handle_string_middle(self, value: str) -> None:
+        """Handle string middle token"""
+        state = self._current_state()
+
+        if not self._progressed:
+            if len(self._state_stack) >= 2:
+                prev = self._state_stack[-2]
+                if prev.type != _StateEnum.InObjectExpectingKey:
+                    self._progressed = True
+            else:
+                self._progressed = True
+
+        if state.type != _StateEnum.InString:
+            raise ValueError(
+                f"Unexpected {json_token_type_to_string(JsonTokenType.StringMiddle)} "
+                f"token when not in string"
+            )
+
+        assert isinstance(state.value, str)
+        state.value += value
+
+        parent_state = self._state_stack[-2] if len(self._state_stack) >= 2 else None
+        self._update_string_parent(state.value, parent_state)
+
+    def handle_string_end(self) -> None:
+        """Handle string end token"""
+        state = self._current_state()
+
+        if state.type != _StateEnum.InString:
+            raise ValueError(
+                f"Unexpected {json_token_type_to_string(JsonTokenType.StringEnd)} "
+                f"token when not in string"
+            )
+
+        self._state_stack.pop()
+        parent_state = self._state_stack[-1] if self._state_stack else None
+        assert isinstance(state.value, str)
+        self._update_string_parent(state.value, parent_state)
+
+    def handle_array_start(self) -> None:
+        """Handle array start token"""
+        self._handle_value_token(JsonTokenType.ArrayStart, None)
+
+    def handle_array_end(self) -> None:
+        """Handle array end token"""
+        state = self._current_state()
+        if state.type != _StateEnum.InArray:
+            raise ValueError(
+                f"Unexpected {json_token_type_to_string(JsonTokenType.ArrayEnd)} token"
+            )
+        self._state_stack.pop()
+
+    def handle_object_start(self) -> None:
+        """Handle object start token"""
+        self._handle_value_token(JsonTokenType.ObjectStart, None)
+
+    def handle_object_end(self) -> None:
+        """Handle object end token"""
+        state = self._current_state()
+
+        if state.type in (
+            _StateEnum.InObjectExpectingKey,
+            _StateEnum.InObjectExpectingValue,
+        ):
+            self._state_stack.pop()
+        else:
+            raise ValueError(
+                f"Unexpected {json_token_type_to_string(JsonTokenType.ObjectEnd)} token"
+            )
+
+    # Private helper methods
+
+    def _current_state(self) -> _State:
+        """Get current parser state"""
+        if not self._state_stack:
+            raise ValueError("Unexpected trailing input")
+        return self._state_stack[-1]
+
+    def _handle_value_token(self, token_type: JsonTokenType, value: JsonValue) -> None:
+        """Handle a complete value token"""
+        state = self._current_state()
+
+        if not self._progressed:
+            self._progressed = True
+
+        if state.type == _StateEnum.Initial:
+            self._state_stack.pop()
+            self._toplevel_value = self._progress_value(token_type, value)
+
+        elif state.type == _StateEnum.InArray:
+            v = self._progress_value(token_type, value)
+            arr = cast(list[JsonValue], state.value)
+            arr.append(v)
+
+        elif state.type == _StateEnum.InObjectExpectingValue:
+            key, obj = cast(tuple[str, JsonObject], state.value)
+            if token_type != JsonTokenType.StringStart:
+                self._state_stack.pop()
+                new_state = _InObjectExpectingKeyState()
+                new_state.value = obj
+                self._state_stack.append(new_state)
+
+            v = self._progress_value(token_type, value)
+            obj[key] = v
+
+        elif state.type == _StateEnum.InString:
+            raise ValueError(
+                f"Unexpected {json_token_type_to_string(token_type)} "
+                f"token in the middle of string"
+            )
+
+        elif state.type == _StateEnum.InObjectExpectingKey:
+            raise ValueError(
+                f"Unexpected {json_token_type_to_string(token_type)} "
+                f"token in the middle of object expecting key"
+            )
+
+    def _update_string_parent(self, updated: str, parent_state: _State | None) -> None:
+        """Update parent container with updated string value"""
+        if parent_state is None:
+            self._toplevel_value = updated
+
+        elif parent_state.type == _StateEnum.InArray:
+            arr = cast(list[JsonValue], parent_state.value)
+            arr[-1] = updated
+
+        elif parent_state.type == _StateEnum.InObjectExpectingValue:
+            key, obj = cast(tuple[str, JsonObject], parent_state.value)
+            obj[key] = updated
+            if self._state_stack and self._state_stack[-1] == parent_state:
+                self._state_stack.pop()
+                new_state = _InObjectExpectingKeyState()
+                new_state.value = obj
+                self._state_stack.append(new_state)
+
+        elif parent_state.type == _StateEnum.InObjectExpectingKey:
+            if self._state_stack and self._state_stack[-1] == parent_state:
+                self._state_stack.pop()
+                obj = cast(JsonObject, parent_state.value)
+                self._state_stack.append(_InObjectExpectingValueState(updated, obj))
+
+    def _progress_value(self, token_type: JsonTokenType, value: JsonValue) -> JsonValue:
+        """Create initial value for a token and push appropriate state"""
+        if token_type == JsonTokenType.Null:
+            return None
+
+        elif token_type == JsonTokenType.Boolean:
+            return value
+
+        elif token_type == JsonTokenType.Number:
+            return value
+
+        elif token_type == JsonTokenType.StringStart:
+            string_state = _InStringState()
+            self._state_stack.append(string_state)
+            return ""
+
+        elif token_type == JsonTokenType.ArrayStart:
+            array_state = _InArrayState()
+            self._state_stack.append(array_state)
+            return array_state.value
+
+        elif token_type == JsonTokenType.ObjectStart:
+            object_state = _InObjectExpectingKeyState()
+            self._state_stack.append(object_state)
+            return object_state.value
+
+        else:
+            raise ValueError(
+                f"Unexpected token type: {json_token_type_to_string(token_type)}"
+            )

backend/onyx/utils/jsonriver/tokenize.py

@@ -0,0 +1,514 @@
+"""
+JSON tokenizer for streaming incremental parsing
+
+Copyright (c) 2023 Google LLC (original TypeScript implementation)
+Copyright (c) 2024 jsonriver-python contributors (Python port)
+SPDX-License-Identifier: BSD-3-Clause
+"""
+
+from __future__ import annotations
+
+import re
+from enum import IntEnum
+from typing import Protocol
+
+
+class TokenHandler(Protocol):
+    """Protocol for handling JSON tokens"""
+
+    def handle_null(self) -> None: ...
+    def handle_boolean(self, value: bool) -> None: ...
+    def handle_number(self, value: float) -> None: ...
+    def handle_string_start(self) -> None: ...
+    def handle_string_middle(self, value: str) -> None: ...
+    def handle_string_end(self) -> None: ...
+    def handle_array_start(self) -> None: ...
+    def handle_array_end(self) -> None: ...
+    def handle_object_start(self) -> None: ...
+    def handle_object_end(self) -> None: ...
+
+
+class JsonTokenType(IntEnum):
+    """Types of JSON tokens"""
+
+    Null = 0
+    Boolean = 1
+    Number = 2
+    StringStart = 3
+    StringMiddle = 4
+    StringEnd = 5
+    ArrayStart = 6
+    ArrayEnd = 7
+    ObjectStart = 8
+    ObjectEnd = 9
+
+
+def json_token_type_to_string(token_type: JsonTokenType) -> str:
+    """Convert token type to readable string"""
+    names = {
+        JsonTokenType.Null: "null",
+        JsonTokenType.Boolean: "boolean",
+        JsonTokenType.Number: "number",
+        JsonTokenType.StringStart: "string start",
+        JsonTokenType.StringMiddle: "string middle",
+        JsonTokenType.StringEnd: "string end",
+        JsonTokenType.ArrayStart: "array start",
+        JsonTokenType.ArrayEnd: "array end",
+        JsonTokenType.ObjectStart: "object start",
+        JsonTokenType.ObjectEnd: "object end",
+    }
+    return names[token_type]
+
+
+class _State(IntEnum):
+    """Internal tokenizer states"""
+
+    ExpectingValue = 0
+    InString = 1
+    StartArray = 2
+    AfterArrayValue = 3
+    StartObject = 4
+    AfterObjectKey = 5
+    AfterObjectValue = 6
+    BeforeObjectKey = 7
+
+
+# Regex for validating JSON numbers
+_JSON_NUMBER_PATTERN = re.compile(r"^-?(0|[1-9]\d*)(\.\d+)?([eE][+-]?\d+)?$")
+
+
+def _parse_json_number(s: str) -> float:
+    """Parse a JSON number string, validating format"""
+    if not _JSON_NUMBER_PATTERN.match(s):
+        raise ValueError("Invalid number")
+    return float(s)
+
+
+class _Input:
+    """
+    Input buffer for chunk-based JSON parsing
+
+    Manages buffering of input chunks and provides methods for
+    consuming and inspecting the buffer.
+    """
+
+    def __init__(self) -> None:
+        self._buffer = ""
+        self._start_index = 0
+        self.buffer_complete = False
+
+    def feed(self, chunk: str) -> None:
+        """Add a chunk of data to the buffer"""
+        self._buffer += chunk
+
+    def mark_complete(self) -> None:
+        """Signal that no more chunks will be fed"""
+        self.buffer_complete = True
+
+    @property
+    def length(self) -> int:
+        """Number of characters remaining in buffer"""
+        return len(self._buffer) - self._start_index
+
+    def advance(self, length: int) -> None:
+        """Advance the start position by length characters"""
+        self._start_index += length
+
+    def peek(self, offset: int) -> str | None:
+        """Peek at character at offset, or None if not available"""
+        idx = self._start_index + offset
+        if idx < len(self._buffer):
+            return self._buffer[idx]
+        return None
+
+    def peek_char_code(self, offset: int) -> int:
+        """Get character code at offset"""
+        return ord(self._buffer[self._start_index + offset])
+
+    def slice(self, start: int, end: int) -> str:
+        """Slice buffer from start to end (relative to current position)"""
+        return self._buffer[self._start_index + start : self._start_index + end]
+
+    def commit(self) -> None:
+        """Commit consumed content, removing it from buffer"""
+        if self._start_index > 0:
+            self._buffer = self._buffer[self._start_index :]
+            self._start_index = 0
+
+    def remaining(self) -> str:
+        """Get all remaining content in buffer"""
+        return self._buffer[self._start_index :]
+
+    def expect_end_of_content(self) -> None:
+        """Verify no non-whitespace content remains"""
+        self.commit()
+        self.skip_past_whitespace()
+        if self.length != 0:
+            raise ValueError(f"Unexpected trailing content {self.remaining()!r}")
+
+    def skip_past_whitespace(self) -> None:
+        """Skip whitespace characters"""
+        i = self._start_index
+        while i < len(self._buffer):
+            c = ord(self._buffer[i])
+            if c in (32, 9, 10, 13):  # space, tab, \n, \r
+                i += 1
+            else:
+                break
+        self._start_index = i
+
+    def try_to_take_prefix(self, prefix: str) -> bool:
+        """Try to consume prefix from buffer, return True if successful"""
+        if self._buffer.startswith(prefix, self._start_index):
+            self._start_index += len(prefix)
+            return True
+        return False
+
+    def try_to_take(self, length: int) -> str | None:
+        """Try to take length characters, or None if not enough available"""
+        if self.length < length:
+            return None
+        result = self._buffer[self._start_index : self._start_index + length]
+        self._start_index += length
+        return result
+
+    def try_to_take_char_code(self) -> int | None:
+        """Try to take a single character as char code, or None if buffer empty"""
+        if self.length == 0:
+            return None
+        code = ord(self._buffer[self._start_index])
+        self._start_index += 1
+        return code
+
+    def take_until_quote_or_backslash(self) -> tuple[str, bool]:
+        """
+        Consume input up to first quote or backslash
+
+        Returns tuple of (consumed_content, pattern_found)
+        """
+        buf = self._buffer
+        i = self._start_index
+        while i < len(buf):
+            c = ord(buf[i])
+            if c <= 0x1F:
+                raise ValueError("Unescaped control character in string")
+            if c == 34 or c == 92:  # " or \
+                result = buf[self._start_index : i]
+                self._start_index = i
+                return (result, True)
+            i += 1
+
+        result = buf[self._start_index :]
+        self._start_index = len(buf)
+        return (result, False)
+
+
+class Tokenizer:
+    """
+    Tokenizer for chunk-based JSON parsing
+
+    Processes chunks fed into its input buffer and calls handler methods
+    as JSON tokens are recognized.
+    """
+
+    def __init__(self, input: _Input, handler: TokenHandler) -> None:
+        self.input = input
+        self._handler = handler
+        self._stack: list[_State] = [_State.ExpectingValue]
+        self._emitted_tokens = 0
+
+    def is_done(self) -> bool:
+        """Check if tokenization is complete"""
+        return len(self._stack) == 0 and self.input.length == 0
+
+    def pump(self) -> None:
+        """Process all available tokens in the buffer"""
+        while True:
+            before = self._emitted_tokens
+            self._tokenize_more()
+            if self._emitted_tokens == before:
+                self.input.commit()
+                return
+
+    def _tokenize_more(self) -> None:
+        """Process one step of tokenization based on current state"""
+        if not self._stack:
+            return
+
+        state = self._stack[-1]
+
+        if state == _State.ExpectingValue:
+            self._tokenize_value()
+        elif state == _State.InString:
+            self._tokenize_string()
+        elif state == _State.StartArray:
+            self._tokenize_array_start()
+        elif state == _State.AfterArrayValue:
+            self._tokenize_after_array_value()
+        elif state == _State.StartObject:
+            self._tokenize_object_start()
+        elif state == _State.AfterObjectKey:
+            self._tokenize_after_object_key()
+        elif state == _State.AfterObjectValue:
+            self._tokenize_after_object_value()
+        elif state == _State.BeforeObjectKey:
+            self._tokenize_before_object_key()
+
+    def _tokenize_value(self) -> None:
+        """Tokenize a JSON value"""
+        self.input.skip_past_whitespace()
+
+        if self.input.try_to_take_prefix("null"):
+            self._handler.handle_null()
+            self._emitted_tokens += 1
+            self._stack.pop()
+            return
+
+        if self.input.try_to_take_prefix("true"):
+            self._handler.handle_boolean(True)
+            self._emitted_tokens += 1
+            self._stack.pop()
+            return
+
+        if self.input.try_to_take_prefix("false"):
+            self._handler.handle_boolean(False)
+            self._emitted_tokens += 1
+            self._stack.pop()
+            return
+
+        if self.input.length > 0:
+            ch = self.input.peek_char_code(0)
+            if (48 <= ch <= 57) or ch == 45:  # 0-9 or -
+                # Scan for end of number
+                i = 0
+                while i < self.input.length:
+                    c = self.input.peek_char_code(i)
+                    if (48 <= c <= 57) or c in (45, 43, 46, 101, 69):  # 0-9 - + . e E
+                        i += 1
+                    else:
+                        break
+
+                if i == self.input.length and not self.input.buffer_complete:
+                    # Need more input (numbers have no terminator)
+                    return
+
+                number_chars = self.input.slice(0, i)
+                self.input.advance(i)
+                number = _parse_json_number(number_chars)
+                self._handler.handle_number(number)
+                self._emitted_tokens += 1
+                self._stack.pop()
+                return
+
+        if self.input.try_to_take_prefix('"'):
+            self._stack.pop()
+            self._stack.append(_State.InString)
+            self._handler.handle_string_start()
+            self._emitted_tokens += 1
+            self._tokenize_string()
+            return
+
+        if self.input.try_to_take_prefix("["):
+            self._stack.pop()
+            self._stack.append(_State.StartArray)
+            self._handler.handle_array_start()
+            self._emitted_tokens += 1
+            self._tokenize_array_start()
+            return
+
+        if self.input.try_to_take_prefix("{"):
+            self._stack.pop()
+            self._stack.append(_State.StartObject)
+            self._handler.handle_object_start()
+            self._emitted_tokens += 1
+            self._tokenize_object_start()
+            return
+
+    def _tokenize_string(self) -> None:
+        """Tokenize string content"""
+        while True:
+            chunk, interrupted = self.input.take_until_quote_or_backslash()
+            if chunk:
+                self._handler.handle_string_middle(chunk)
+                self._emitted_tokens += 1
+            elif not interrupted:
+                return
+
+            if interrupted:
+                if self.input.length == 0:
+                    return
+
+                next_char = self.input.peek(0)
+                if next_char == '"':
+                    self.input.advance(1)
+                    self._handler.handle_string_end()
+                    self._emitted_tokens += 1
+                    self._stack.pop()
+                    return
+
+                # Handle escape sequences
+                next_char2 = self.input.peek(1)
+                if next_char2 is None:
+                    return
+
+                value: str
+                if next_char2 == "u":
+                    # Unicode escape: need 4 hex digits
+                    if self.input.length < 6:
+                        return
+
+                    code = 0
+                    for j in range(2, 6):
+                        c = self.input.peek_char_code(j)
+                        if 48 <= c <= 57:  # 0-9
+                            digit = c - 48
+                        elif 65 <= c <= 70:  # A-F
+                            digit = c - 55
+                        elif 97 <= c <= 102:  # a-f
+                            digit = c - 87
+                        else:
+                            raise ValueError("Bad Unicode escape in JSON")
+                        code = (code << 4) | digit
+
+                    self.input.advance(6)
+                    self._handler.handle_string_middle(chr(code))
+                    self._emitted_tokens += 1
+                    continue
+
+                elif next_char2 == "n":
+                    value = "\n"
+                elif next_char2 == "r":
+                    value = "\r"
+                elif next_char2 == "t":
+                    value = "\t"
+                elif next_char2 == "b":
+                    value = "\b"
+                elif next_char2 == "f":
+                    value = "\f"
+                elif next_char2 == "\\":
+                    value = "\\"
+                elif next_char2 == "/":
+                    value = "/"
+                elif next_char2 == '"':
+                    value = '"'
+                else:
+                    raise ValueError("Bad escape in string")
+
+                self.input.advance(2)
+                self._handler.handle_string_middle(value)
+                self._emitted_tokens += 1
+
+    def _tokenize_array_start(self) -> None:
+        """Tokenize start of array (check for empty or first element)"""
+        self.input.skip_past_whitespace()
+        if self.input.length == 0:
+            return
+
+        if self.input.try_to_take_prefix("]"):
+            self._handler.handle_array_end()
+            self._emitted_tokens += 1
+            self._stack.pop()
+            return
+
+        self._stack.pop()
+        self._stack.append(_State.AfterArrayValue)
+        self._stack.append(_State.ExpectingValue)
+        self._tokenize_value()
+
+    def _tokenize_after_array_value(self) -> None:
+        """Tokenize after an array value (expect , or ])"""
+        self.input.skip_past_whitespace()
+        next_char = self.input.try_to_take_char_code()
+
+        if next_char is None:
+            return
+        elif next_char == 0x5D:  # ]
+            self._handler.handle_array_end()
+            self._emitted_tokens += 1
+            self._stack.pop()
+            return
+        elif next_char == 0x2C:  # ,
+            self._stack.append(_State.ExpectingValue)
+            self._tokenize_value()
+            return
+        else:
+            raise ValueError(f"Expected , or ], got {chr(next_char)!r}")
+
+    def _tokenize_object_start(self) -> None:
+        """Tokenize start of object (check for empty or first key)"""
+        self.input.skip_past_whitespace()
+        next_char = self.input.try_to_take_char_code()
+
+        if next_char is None:
+            return
+        elif next_char == 0x7D:  # }
+            self._handler.handle_object_end()
+            self._emitted_tokens += 1
+            self._stack.pop()
+            return
+        elif next_char == 0x22:  # "
+            self._stack.pop()
+            self._stack.append(_State.AfterObjectKey)
+            self._stack.append(_State.InString)
+            self._handler.handle_string_start()
+            self._emitted_tokens += 1
+            self._tokenize_string()
+            return
+        else:
+            raise ValueError(f"Expected start of object key, got {chr(next_char)!r}")
+
+    def _tokenize_after_object_key(self) -> None:
+        """Tokenize after object key (expect :)"""
+        self.input.skip_past_whitespace()
+        next_char = self.input.try_to_take_char_code()
+
+        if next_char is None:
+            return
+        elif next_char == 0x3A:  # :
+            self._stack.pop()
+            self._stack.append(_State.AfterObjectValue)
+            self._stack.append(_State.ExpectingValue)
+            self._tokenize_value()
+            return
+        else:
+            raise ValueError(f"Expected colon after object key, got {chr(next_char)!r}")
+
+    def _tokenize_after_object_value(self) -> None:
+        """Tokenize after object value (expect , or })"""
+        self.input.skip_past_whitespace()
+        next_char = self.input.try_to_take_char_code()
+
+        if next_char is None:
+            return
+        elif next_char == 0x7D:  # }
+            self._handler.handle_object_end()
+            self._emitted_tokens += 1
+            self._stack.pop()
+            return
+        elif next_char == 0x2C:  # ,
+            self._stack.pop()
+            self._stack.append(_State.BeforeObjectKey)
+            self._tokenize_before_object_key()
+            return
+        else:
+            raise ValueError(
+                f"Expected , or }} after object value, got {chr(next_char)!r}"
+            )
+
+    def _tokenize_before_object_key(self) -> None:
+        """Tokenize before object key (after comma)"""
+        self.input.skip_past_whitespace()
+        next_char = self.input.try_to_take_char_code()
+
+        if next_char is None:
+            return
+        elif next_char == 0x22:  # "
+            self._stack.pop()
+            self._stack.append(_State.AfterObjectKey)
+            self._stack.append(_State.InString)
+            self._handler.handle_string_start()
+            self._emitted_tokens += 1
+            self._tokenize_string()
+            return
+        else:
+            raise ValueError(f"Expected start of object key, got {chr(next_char)!r}")

backend/onyx/utils/postgres_sanitization.py

@@ -1,30 +1,49 @@
+import re
 from typing import Any
 
 from onyx.access.models import ExternalAccess
 from onyx.connectors.models import BasicExpertInfo
 from onyx.connectors.models import Document
 from onyx.connectors.models import HierarchyNode
+from onyx.utils.logger import setup_logger
+
+logger = setup_logger()
+
+_SURROGATE_RE = re.compile(r"[\ud800-\udfff]")
 
 
-def _sanitize_string(value: str) -> str:
-    return value.replace("\x00", "")
+def sanitize_string(value: str) -> str:
+    """Strip characters that PostgreSQL text/JSONB columns cannot store.
+
+    Removes:
+    - NUL bytes (\\x00)
+    - UTF-16 surrogates (\\ud800-\\udfff), which are invalid in UTF-8
+    """
+    sanitized = value.replace("\x00", "")
+    sanitized = _SURROGATE_RE.sub("", sanitized)
+    if value and not sanitized:
+        logger.warning(
+            "sanitize_string: all characters were removed from a non-empty string"
+        )
+    return sanitized
 
 
-def _sanitize_json_like(value: Any) -> Any:
+def sanitize_json_like(value: Any) -> Any:
+    """Recursively sanitize all strings in a JSON-like structure (dict/list/tuple)."""
     if isinstance(value, str):
-        return _sanitize_string(value)
+        return sanitize_string(value)
 
     if isinstance(value, list):
-        return [_sanitize_json_like(item) for item in value]
+        return [sanitize_json_like(item) for item in value]
 
     if isinstance(value, tuple):
-        return tuple(_sanitize_json_like(item) for item in value)
+        return tuple(sanitize_json_like(item) for item in value)
 
     if isinstance(value, dict):
         sanitized: dict[Any, Any] = {}
         for key, nested_value in value.items():
-            cleaned_key = _sanitize_string(key) if isinstance(key, str) else key
-            sanitized[cleaned_key] = _sanitize_json_like(nested_value)
+            cleaned_key = sanitize_string(key) if isinstance(key, str) else key
+            sanitized[cleaned_key] = sanitize_json_like(nested_value)
         return sanitized
 
     return value
@@ -34,27 +53,27 @@ def _sanitize_expert_info(expert: BasicExpertInfo) -> BasicExpertInfo:
     return expert.model_copy(
         update={
             "display_name": (
-                _sanitize_string(expert.display_name)
+                sanitize_string(expert.display_name)
                 if expert.display_name is not None
                 else None
             ),
             "first_name": (
-                _sanitize_string(expert.first_name)
+                sanitize_string(expert.first_name)
                 if expert.first_name is not None
                 else None
             ),
             "middle_initial": (
-                _sanitize_string(expert.middle_initial)
+                sanitize_string(expert.middle_initial)
                 if expert.middle_initial is not None
                 else None
             ),
             "last_name": (
-                _sanitize_string(expert.last_name)
+                sanitize_string(expert.last_name)
                 if expert.last_name is not None
                 else None
             ),
             "email": (
-                _sanitize_string(expert.email) if expert.email is not None else None
+                sanitize_string(expert.email) if expert.email is not None else None
             ),
         }
     )
@@ -63,10 +82,10 @@ def _sanitize_expert_info(expert: BasicExpertInfo) -> BasicExpertInfo:
 def _sanitize_external_access(external_access: ExternalAccess) -> ExternalAccess:
     return ExternalAccess(
         external_user_emails={
-            _sanitize_string(email) for email in external_access.external_user_emails
+            sanitize_string(email) for email in external_access.external_user_emails
         },
         external_user_group_ids={
-            _sanitize_string(group_id)
+            sanitize_string(group_id)
             for group_id in external_access.external_user_group_ids
         },
         is_public=external_access.is_public,
@@ -76,26 +95,26 @@ def _sanitize_external_access(external_access: ExternalAccess) -> ExternalAccess
 def sanitize_document_for_postgres(document: Document) -> Document:
     cleaned_doc = document.model_copy(deep=True)
 
-    cleaned_doc.id = _sanitize_string(cleaned_doc.id)
-    cleaned_doc.semantic_identifier = _sanitize_string(cleaned_doc.semantic_identifier)
+    cleaned_doc.id = sanitize_string(cleaned_doc.id)
+    cleaned_doc.semantic_identifier = sanitize_string(cleaned_doc.semantic_identifier)
     if cleaned_doc.title is not None:
-        cleaned_doc.title = _sanitize_string(cleaned_doc.title)
+        cleaned_doc.title = sanitize_string(cleaned_doc.title)
     if cleaned_doc.parent_hierarchy_raw_node_id is not None:
-        cleaned_doc.parent_hierarchy_raw_node_id = _sanitize_string(
+        cleaned_doc.parent_hierarchy_raw_node_id = sanitize_string(
             cleaned_doc.parent_hierarchy_raw_node_id
         )
 
     cleaned_doc.metadata = {
-        _sanitize_string(key): (
-            [_sanitize_string(item) for item in value]
+        sanitize_string(key): (
+            [sanitize_string(item) for item in value]
             if isinstance(value, list)
-            else _sanitize_string(value)
+            else sanitize_string(value)
         )
         for key, value in cleaned_doc.metadata.items()
     }
 
     if cleaned_doc.doc_metadata is not None:
-        cleaned_doc.doc_metadata = _sanitize_json_like(cleaned_doc.doc_metadata)
+        cleaned_doc.doc_metadata = sanitize_json_like(cleaned_doc.doc_metadata)
 
     if cleaned_doc.primary_owners is not None:
         cleaned_doc.primary_owners = [
@@ -113,11 +132,11 @@ def sanitize_document_for_postgres(document: Document) -> Document:
 
     for section in cleaned_doc.sections:
         if section.link is not None:
-            section.link = _sanitize_string(section.link)
+            section.link = sanitize_string(section.link)
         if section.text is not None:
-            section.text = _sanitize_string(section.text)
+            section.text = sanitize_string(section.text)
         if section.image_file_id is not None:
-            section.image_file_id = _sanitize_string(section.image_file_id)
+            section.image_file_id = sanitize_string(section.image_file_id)
 
     return cleaned_doc
 
@@ -129,12 +148,12 @@ def sanitize_documents_for_postgres(documents: list[Document]) -> list[Document]
 def sanitize_hierarchy_node_for_postgres(node: HierarchyNode) -> HierarchyNode:
     cleaned_node = node.model_copy(deep=True)
 
-    cleaned_node.raw_node_id = _sanitize_string(cleaned_node.raw_node_id)
-    cleaned_node.display_name = _sanitize_string(cleaned_node.display_name)
+    cleaned_node.raw_node_id = sanitize_string(cleaned_node.raw_node_id)
+    cleaned_node.display_name = sanitize_string(cleaned_node.display_name)
     if cleaned_node.raw_parent_id is not None:
-        cleaned_node.raw_parent_id = _sanitize_string(cleaned_node.raw_parent_id)
+        cleaned_node.raw_parent_id = sanitize_string(cleaned_node.raw_parent_id)
     if cleaned_node.link is not None:
-        cleaned_node.link = _sanitize_string(cleaned_node.link)
+        cleaned_node.link = sanitize_string(cleaned_node.link)
 
     if cleaned_node.external_access is not None:
         cleaned_node.external_access = _sanitize_external_access(

backend/onyx/utils/sensitive.py

@@ -128,6 +128,8 @@ class SensitiveValue(Generic[T]):
         value = self._decrypt()
 
         if not apply_mask:
+            # Callers must not mutate the returned dict — doing so would
+            # desync the cache from the encrypted bytes and the DB.
             return value
 
         # Apply masking
@@ -174,18 +176,20 @@ class SensitiveValue(Generic[T]):
         )
 
     def __eq__(self, other: Any) -> bool:
-        """Prevent direct comparison which might expose value."""
-        if isinstance(other, SensitiveValue):
-            # Compare encrypted bytes for equality check
-            return self._encrypted_bytes == other._encrypted_bytes
-        raise SensitiveAccessError(
-            "Cannot compare SensitiveValue with non-SensitiveValue. "
-            "Use .get_value(apply_mask=True/False) to access the value for comparison."
-        )
+        """Compare SensitiveValues by their decrypted content."""
+        # NOTE: if you attempt to compare a string/dict to a SensitiveValue,
+        # this comparison will return NotImplemented, which then evaluates to False.
+        # This is the convention and required for SQLAlchemy's attribute tracking.
+        if not isinstance(other, SensitiveValue):
+            return NotImplemented
+        return self._decrypt() == other._decrypt()
 
     def __hash__(self) -> int:
-        """Allow hashing based on encrypted bytes."""
-        return hash(self._encrypted_bytes)
+        """Hash based on decrypted content."""
+        value = self._decrypt()
+        if isinstance(value, dict):
+            return hash(json.dumps(value, sort_keys=True))
+        return hash(value)
 
     # Prevent JSON serialization
     def __json__(self) -> Any:

backend/onyx/utils/variable_functionality.py

@@ -24,6 +24,9 @@ class OnyxVersion:
     def set_ee(self) -> None:
         self._is_ee = True
 
+    def unset_ee(self) -> None:
+        self._is_ee = False
+
     def is_ee_version(self) -> bool:
         return self._is_ee
 

backend/requirements/default.txt

@@ -65,7 +65,7 @@ attrs==25.4.0
     #   jsonschema
     #   referencing
     #   zeep
-authlib==1.6.6
+authlib==1.6.7
     # via fastmcp
 babel==2.17.0
     # via courlan
@@ -109,9 +109,7 @@ brotli==1.2.0
 bytecode==0.17.0
     # via ddtrace
 cachetools==6.2.2
-    # via
-    #   google-auth
-    #   py-key-value-aio
+    # via py-key-value-aio
 caio==0.9.25
     # via aiofile
 celery==5.5.1
@@ -190,6 +188,7 @@ courlan==1.3.2
 cryptography==46.0.5
     # via
     #   authlib
+    #   google-auth
     #   msal
     #   msoffcrypto-tool
     #   pdfminer-six
@@ -230,9 +229,7 @@ distro==1.9.0
 dnspython==2.8.0
     # via email-validator
 docstring-parser==0.17.0
-    # via
-    #   cyclopts
-    #   google-cloud-aiplatform
+    # via cyclopts
 docutils==0.22.3
     # via rich-rst
 dropbox==12.0.2
@@ -297,26 +294,15 @@ gitdb==4.0.12
 gitpython==3.1.45
     # via braintrust
 google-api-core==2.28.1
-    # via
-    #   google-api-python-client
-    #   google-cloud-aiplatform
-    #   google-cloud-bigquery
-    #   google-cloud-core
-    #   google-cloud-resource-manager
-    #   google-cloud-storage
+    # via google-api-python-client
 google-api-python-client==2.86.0
     # via onyx
-google-auth==2.43.0
+google-auth==2.48.0
     # via
     #   google-api-core
     #   google-api-python-client
     #   google-auth-httplib2
     #   google-auth-oauthlib
-    #   google-cloud-aiplatform
-    #   google-cloud-bigquery
-    #   google-cloud-core
-    #   google-cloud-resource-manager
-    #   google-cloud-storage
     #   google-genai
     #   kubernetes
 google-auth-httplib2==0.1.0
@@ -325,51 +311,16 @@ google-auth-httplib2==0.1.0
     #   onyx
 google-auth-oauthlib==1.0.0
     # via onyx
-google-cloud-aiplatform==1.121.0
-    # via onyx
-google-cloud-bigquery==3.38.0
-    # via google-cloud-aiplatform
-google-cloud-core==2.5.0
-    # via
-    #   google-cloud-bigquery
-    #   google-cloud-storage
-google-cloud-resource-manager==1.15.0
-    # via google-cloud-aiplatform
-google-cloud-storage==2.19.0
-    # via google-cloud-aiplatform
-google-crc32c==1.7.1
-    # via
-    #   google-cloud-storage
-    #   google-resumable-media
 google-genai==1.52.0
-    # via
-    #   google-cloud-aiplatform
-    #   onyx
-google-resumable-media==2.7.2
-    # via
-    #   google-cloud-bigquery
-    #   google-cloud-storage
+    # via onyx
 googleapis-common-protos==1.72.0
     # via
     #   google-api-core
-    #   grpc-google-iam-v1
-    #   grpcio-status
     #   opentelemetry-exporter-otlp-proto-http
 greenlet==3.2.4
     # via
     #   playwright
     #   sqlalchemy
-grpc-google-iam-v1==0.14.3
-    # via google-cloud-resource-manager
-grpcio==1.76.0
-    # via
-    #   google-api-core
-    #   google-cloud-resource-manager
-    #   googleapis-common-protos
-    #   grpc-google-iam-v1
-    #   grpcio-status
-grpcio-status==1.76.0
-    # via google-api-core
 h11==0.16.0
     # via
     #   httpcore
@@ -670,8 +621,6 @@ packaging==24.2
     #   dask
     #   distributed
     #   fastmcp
-    #   google-cloud-aiplatform
-    #   google-cloud-bigquery
     #   huggingface-hub
     #   jira
     #   kombu
@@ -721,19 +670,12 @@ propcache==0.4.1
     #   aiohttp
     #   yarl
 proto-plus==1.26.1
-    # via
-    #   google-api-core
-    #   google-cloud-aiplatform
-    #   google-cloud-resource-manager
+    # via google-api-core
 protobuf==6.33.5
     # via
     #   ddtrace
     #   google-api-core
-    #   google-cloud-aiplatform
-    #   google-cloud-resource-manager
     #   googleapis-common-protos
-    #   grpc-google-iam-v1
-    #   grpcio-status
     #   onnxruntime
     #   opentelemetry-proto
     #   proto-plus
@@ -771,7 +713,6 @@ pydantic==2.11.7
     #   exa-py
     #   fastapi
     #   fastmcp
-    #   google-cloud-aiplatform
     #   google-genai
     #   langchain-core
     #   langfuse
@@ -835,7 +776,6 @@ python-dateutil==2.8.2
     #   botocore
     #   celery
     #   dateparser
-    #   google-cloud-bigquery
     #   htmldate
     #   hubspot-api-client
     #   kubernetes
@@ -927,8 +867,6 @@ requests==2.32.5
     #   dropbox
     #   exa-py
     #   google-api-core
-    #   google-cloud-bigquery
-    #   google-cloud-storage
     #   google-genai
     #   hubspot-api-client
     #   huggingface-hub
@@ -1002,9 +940,7 @@ sendgrid==6.12.5
 sentry-sdk==2.14.0
     # via onyx
 shapely==2.0.6
-    # via
-    #   google-cloud-aiplatform
-    #   onyx
+    # via onyx
 shellingham==1.5.4
     # via typer
 simple-salesforce==1.12.6
@@ -1118,9 +1054,7 @@ typing-extensions==4.15.0
     #   exa-py
     #   exceptiongroup
     #   fastapi
-    #   google-cloud-aiplatform
     #   google-genai
-    #   grpcio
     #   huggingface-hub
     #   jira
     #   langchain-core

backend/requirements/dev.txt

@@ -59,8 +59,6 @@ botocore==1.39.11
     #   s3transfer
 brotli==1.2.0
     # via onyx
-cachetools==6.2.2
-    # via google-auth
 celery-types==0.19.0
     # via onyx
 certifi==2025.11.12
@@ -100,7 +98,9 @@ comm==0.2.3
 contourpy==1.3.3
     # via matplotlib
 cryptography==46.0.5
-    # via pyjwt
+    # via
+    #   google-auth
+    #   pyjwt
 cycler==0.12.1
     # via matplotlib
 debugpy==1.8.17
@@ -115,8 +115,6 @@ distlib==0.4.0
     # via virtualenv
 distro==1.9.0
     # via openai
-docstring-parser==0.17.0
-    # via google-cloud-aiplatform
 durationpy==0.10
     # via kubernetes
 execnet==2.1.2
@@ -145,65 +143,14 @@ frozenlist==1.8.0
     #   aiosignal
 fsspec==2025.10.0
     # via huggingface-hub
-google-api-core==2.28.1
+google-auth==2.48.0
     # via
-    #   google-cloud-aiplatform
-    #   google-cloud-bigquery
-    #   google-cloud-core
-    #   google-cloud-resource-manager
-    #   google-cloud-storage
-google-auth==2.43.0
-    # via
-    #   google-api-core
-    #   google-cloud-aiplatform
-    #   google-cloud-bigquery
-    #   google-cloud-core
-    #   google-cloud-resource-manager
-    #   google-cloud-storage
     #   google-genai
     #   kubernetes
-google-cloud-aiplatform==1.121.0
-    # via onyx
-google-cloud-bigquery==3.38.0
-    # via google-cloud-aiplatform
-google-cloud-core==2.5.0
-    # via
-    #   google-cloud-bigquery
-    #   google-cloud-storage
-google-cloud-resource-manager==1.15.0
-    # via google-cloud-aiplatform
-google-cloud-storage==2.19.0
-    # via google-cloud-aiplatform
-google-crc32c==1.7.1
-    # via
-    #   google-cloud-storage
-    #   google-resumable-media
 google-genai==1.52.0
-    # via
-    #   google-cloud-aiplatform
-    #   onyx
-google-resumable-media==2.7.2
-    # via
-    #   google-cloud-bigquery
-    #   google-cloud-storage
-googleapis-common-protos==1.72.0
-    # via
-    #   google-api-core
-    #   grpc-google-iam-v1
-    #   grpcio-status
+    # via onyx
 greenlet==3.2.4 ; platform_machine == 'AMD64' or platform_machine == 'WIN32' or platform_machine == 'aarch64' or platform_machine == 'amd64' or platform_machine == 'ppc64le' or platform_machine == 'win32' or platform_machine == 'x86_64'
     # via sqlalchemy
-grpc-google-iam-v1==0.14.3
-    # via google-cloud-resource-manager
-grpcio==1.76.0
-    # via
-    #   google-api-core
-    #   google-cloud-resource-manager
-    #   googleapis-common-protos
-    #   grpc-google-iam-v1
-    #   grpcio-status
-grpcio-status==1.76.0
-    # via google-api-core
 h11==0.16.0
     # via
     #   httpcore
@@ -311,13 +258,12 @@ numpy==2.4.1
     #   contourpy
     #   matplotlib
     #   pandas-stubs
-    #   shapely
     #   voyageai
 oauthlib==3.2.2
     # via
     #   kubernetes
     #   requests-oauthlib
-onyx-devtools==0.6.2
+onyx-devtools==0.6.3
     # via onyx
 openai==2.14.0
     # via
@@ -330,8 +276,6 @@ openapi-generator-cli==7.17.0
 packaging==24.2
     # via
     #   black
-    #   google-cloud-aiplatform
-    #   google-cloud-bigquery
     #   hatchling
     #   huggingface-hub
     #   ipykernel
@@ -374,20 +318,6 @@ propcache==0.4.1
     # via
     #   aiohttp
     #   yarl
-proto-plus==1.26.1
-    # via
-    #   google-api-core
-    #   google-cloud-aiplatform
-    #   google-cloud-resource-manager
-protobuf==6.33.5
-    # via
-    #   google-api-core
-    #   google-cloud-aiplatform
-    #   google-cloud-resource-manager
-    #   googleapis-common-protos
-    #   grpc-google-iam-v1
-    #   grpcio-status
-    #   proto-plus
 psutil==7.1.3
     # via ipykernel
 ptyprocess==0.7.0 ; sys_platform != 'emscripten' and sys_platform != 'win32'
@@ -409,7 +339,6 @@ pydantic==2.11.7
     #   agent-client-protocol
     #   cohere
     #   fastapi
-    #   google-cloud-aiplatform
     #   google-genai
     #   litellm
     #   mcp
@@ -450,7 +379,6 @@ python-dateutil==2.8.2
     # via
     #   aiobotocore
     #   botocore
-    #   google-cloud-bigquery
     #   jupyter-client
     #   kubernetes
     #   matplotlib
@@ -485,9 +413,6 @@ reorder-python-imports-black==3.14.0
 requests==2.32.5
     # via
     #   cohere
-    #   google-api-core
-    #   google-cloud-bigquery
-    #   google-cloud-storage
     #   google-genai
     #   huggingface-hub
     #   kubernetes
@@ -510,8 +435,6 @@ s3transfer==0.13.1
     # via boto3
 sentry-sdk==2.14.0
     # via onyx
-shapely==2.0.6
-    # via google-cloud-aiplatform
 six==1.17.0
     # via
     #   kubernetes
@@ -602,9 +525,7 @@ typing-extensions==4.15.0
     #   celery-types
     #   cohere
     #   fastapi
-    #   google-cloud-aiplatform
     #   google-genai
-    #   grpcio
     #   huggingface-hub
     #   ipython
     #   mcp

backend/requirements/ee.txt

@@ -53,8 +53,6 @@ botocore==1.39.11
     #   s3transfer
 brotli==1.2.0
     # via onyx
-cachetools==6.2.2
-    # via google-auth
 certifi==2025.11.12
     # via
     #   httpcore
@@ -79,15 +77,15 @@ colorama==0.4.6 ; sys_platform == 'win32'
     #   click
     #   tqdm
 cryptography==46.0.5
-    # via pyjwt
+    # via
+    #   google-auth
+    #   pyjwt
 decorator==5.2.1
     # via retry
 discord-py==2.4.0
     # via onyx
 distro==1.9.0
     # via openai
-docstring-parser==0.17.0
-    # via google-cloud-aiplatform
 durationpy==0.10
     # via kubernetes
 fastapi==0.133.1
@@ -104,63 +102,12 @@ frozenlist==1.8.0
     #   aiosignal
 fsspec==2025.10.0
     # via huggingface-hub
-google-api-core==2.28.1
+google-auth==2.48.0
     # via
-    #   google-cloud-aiplatform
-    #   google-cloud-bigquery
-    #   google-cloud-core
-    #   google-cloud-resource-manager
-    #   google-cloud-storage
-google-auth==2.43.0
-    # via
-    #   google-api-core
-    #   google-cloud-aiplatform
-    #   google-cloud-bigquery
-    #   google-cloud-core
-    #   google-cloud-resource-manager
-    #   google-cloud-storage
     #   google-genai
     #   kubernetes
-google-cloud-aiplatform==1.121.0
-    # via onyx
-google-cloud-bigquery==3.38.0
-    # via google-cloud-aiplatform
-google-cloud-core==2.5.0
-    # via
-    #   google-cloud-bigquery
-    #   google-cloud-storage
-google-cloud-resource-manager==1.15.0
-    # via google-cloud-aiplatform
-google-cloud-storage==2.19.0
-    # via google-cloud-aiplatform
-google-crc32c==1.7.1
-    # via
-    #   google-cloud-storage
-    #   google-resumable-media
 google-genai==1.52.0
-    # via
-    #   google-cloud-aiplatform
-    #   onyx
-google-resumable-media==2.7.2
-    # via
-    #   google-cloud-bigquery
-    #   google-cloud-storage
-googleapis-common-protos==1.72.0
-    # via
-    #   google-api-core
-    #   grpc-google-iam-v1
-    #   grpcio-status
-grpc-google-iam-v1==0.14.3
-    # via google-cloud-resource-manager
-grpcio==1.76.0
-    # via
-    #   google-api-core
-    #   google-cloud-resource-manager
-    #   googleapis-common-protos
-    #   grpc-google-iam-v1
-    #   grpcio-status
-grpcio-status==1.76.0
-    # via google-api-core
+    # via onyx
 h11==0.16.0
     # via
     #   httpcore
@@ -221,9 +168,7 @@ multidict==6.7.0
     #   aiohttp
     #   yarl
 numpy==2.4.1
-    # via
-    #   shapely
-    #   voyageai
+    # via voyageai
 oauthlib==3.2.2
     # via
     #   kubernetes
@@ -233,10 +178,7 @@ openai==2.14.0
     #   litellm
     #   onyx
 packaging==24.2
-    # via
-    #   google-cloud-aiplatform
-    #   google-cloud-bigquery
-    #   huggingface-hub
+    # via huggingface-hub
 parameterized==0.9.0
     # via cohere
 posthog==3.7.4
@@ -251,20 +193,6 @@ propcache==0.4.1
     # via
     #   aiohttp
     #   yarl
-proto-plus==1.26.1
-    # via
-    #   google-api-core
-    #   google-cloud-aiplatform
-    #   google-cloud-resource-manager
-protobuf==6.33.5
-    # via
-    #   google-api-core
-    #   google-cloud-aiplatform
-    #   google-cloud-resource-manager
-    #   googleapis-common-protos
-    #   grpc-google-iam-v1
-    #   grpcio-status
-    #   proto-plus
 py==1.11.0
     # via retry
 pyasn1==0.6.2
@@ -280,7 +208,6 @@ pydantic==2.11.7
     #   agent-client-protocol
     #   cohere
     #   fastapi
-    #   google-cloud-aiplatform
     #   google-genai
     #   litellm
     #   mcp
@@ -297,7 +224,6 @@ python-dateutil==2.8.2
     # via
     #   aiobotocore
     #   botocore
-    #   google-cloud-bigquery
     #   kubernetes
     #   posthog
 python-dotenv==1.1.1
@@ -321,9 +247,6 @@ regex==2025.11.3
 requests==2.32.5
     # via
     #   cohere
-    #   google-api-core
-    #   google-cloud-bigquery
-    #   google-cloud-storage
     #   google-genai
     #   huggingface-hub
     #   kubernetes
@@ -345,8 +268,6 @@ s3transfer==0.13.1
     # via boto3
 sentry-sdk==2.14.0
     # via onyx
-shapely==2.0.6
-    # via google-cloud-aiplatform
 six==1.17.0
     # via
     #   kubernetes
@@ -385,9 +306,7 @@ typing-extensions==4.15.0
     #   anyio
     #   cohere
     #   fastapi
-    #   google-cloud-aiplatform
     #   google-genai
-    #   grpcio
     #   huggingface-hub
     #   mcp
     #   openai

backend/requirements/model_server.txt

@@ -57,8 +57,6 @@ botocore==1.39.11
     #   s3transfer
 brotli==1.2.0
     # via onyx
-cachetools==6.2.2
-    # via google-auth
 celery==5.5.1
     # via sentry-sdk
 certifi==2025.11.12
@@ -95,15 +93,15 @@ colorama==0.4.6 ; sys_platform == 'win32'
     #   click
     #   tqdm
 cryptography==46.0.5
-    # via pyjwt
+    # via
+    #   google-auth
+    #   pyjwt
 decorator==5.2.1
     # via retry
 discord-py==2.4.0
     # via onyx
 distro==1.9.0
     # via openai
-docstring-parser==0.17.0
-    # via google-cloud-aiplatform
 durationpy==0.10
     # via kubernetes
 einops==0.8.1
@@ -129,63 +127,12 @@ fsspec==2025.10.0
     # via
     #   huggingface-hub
     #   torch
-google-api-core==2.28.1
+google-auth==2.48.0
     # via
-    #   google-cloud-aiplatform
-    #   google-cloud-bigquery
-    #   google-cloud-core
-    #   google-cloud-resource-manager
-    #   google-cloud-storage
-google-auth==2.43.0
-    # via
-    #   google-api-core
-    #   google-cloud-aiplatform
-    #   google-cloud-bigquery
-    #   google-cloud-core
-    #   google-cloud-resource-manager
-    #   google-cloud-storage
     #   google-genai
     #   kubernetes
-google-cloud-aiplatform==1.121.0
-    # via onyx
-google-cloud-bigquery==3.38.0
-    # via google-cloud-aiplatform
-google-cloud-core==2.5.0
-    # via
-    #   google-cloud-bigquery
-    #   google-cloud-storage
-google-cloud-resource-manager==1.15.0
-    # via google-cloud-aiplatform
-google-cloud-storage==2.19.0
-    # via google-cloud-aiplatform
-google-crc32c==1.7.1
-    # via
-    #   google-cloud-storage
-    #   google-resumable-media
 google-genai==1.52.0
-    # via
-    #   google-cloud-aiplatform
-    #   onyx
-google-resumable-media==2.7.2
-    # via
-    #   google-cloud-bigquery
-    #   google-cloud-storage
-googleapis-common-protos==1.72.0
-    # via
-    #   google-api-core
-    #   grpc-google-iam-v1
-    #   grpcio-status
-grpc-google-iam-v1==0.14.3
-    # via google-cloud-resource-manager
-grpcio==1.76.0
-    # via
-    #   google-api-core
-    #   google-cloud-resource-manager
-    #   googleapis-common-protos
-    #   grpc-google-iam-v1
-    #   grpcio-status
-grpcio-status==1.76.0
-    # via google-api-core
+    # via onyx
 h11==0.16.0
     # via
     #   httpcore
@@ -263,7 +210,6 @@ numpy==2.4.1
     #   onyx
     #   scikit-learn
     #   scipy
-    #   shapely
     #   transformers
     #   voyageai
 nvidia-cublas-cu12==12.8.4.1 ; platform_machine == 'x86_64' and sys_platform == 'linux'
@@ -316,8 +262,6 @@ openai==2.14.0
 packaging==24.2
     # via
     #   accelerate
-    #   google-cloud-aiplatform
-    #   google-cloud-bigquery
     #   huggingface-hub
     #   kombu
     #   transformers
@@ -337,20 +281,6 @@ propcache==0.4.1
     # via
     #   aiohttp
     #   yarl
-proto-plus==1.26.1
-    # via
-    #   google-api-core
-    #   google-cloud-aiplatform
-    #   google-cloud-resource-manager
-protobuf==6.33.5
-    # via
-    #   google-api-core
-    #   google-cloud-aiplatform
-    #   google-cloud-resource-manager
-    #   googleapis-common-protos
-    #   grpc-google-iam-v1
-    #   grpcio-status
-    #   proto-plus
 psutil==7.1.3
     # via accelerate
 py==1.11.0
@@ -368,7 +298,6 @@ pydantic==2.11.7
     #   agent-client-protocol
     #   cohere
     #   fastapi
-    #   google-cloud-aiplatform
     #   google-genai
     #   litellm
     #   mcp
@@ -386,7 +315,6 @@ python-dateutil==2.8.2
     #   aiobotocore
     #   botocore
     #   celery
-    #   google-cloud-bigquery
     #   kubernetes
 python-dotenv==1.1.1
     # via
@@ -413,9 +341,6 @@ regex==2025.11.3
 requests==2.32.5
     # via
     #   cohere
-    #   google-api-core
-    #   google-cloud-bigquery
-    #   google-cloud-storage
     #   google-genai
     #   huggingface-hub
     #   kubernetes
@@ -452,8 +377,6 @@ sentry-sdk==2.14.0
     # via onyx
 setuptools==80.9.0 ; python_full_version >= '3.12'
     # via torch
-shapely==2.0.6
-    # via google-cloud-aiplatform
 six==1.17.0
     # via
     #   kubernetes
@@ -510,9 +433,7 @@ typing-extensions==4.15.0
     #   anyio
     #   cohere
     #   fastapi
-    #   google-cloud-aiplatform
     #   google-genai
-    #   grpcio
     #   huggingface-hub
     #   mcp
     #   openai

backend/scripts/debugging/opensearch/opensearch_debug.py

@@ -0,0 +1,171 @@
+#!/usr/bin/env python3
+"""A utility to interact with OpenSearch.
+
+Usage:
+    python3 opensearch_debug.py --help
+    python3 opensearch_debug.py list
+    python3 opensearch_debug.py delete <index_name>
+
+Environment Variables:
+    OPENSEARCH_HOST: OpenSearch host
+    OPENSEARCH_REST_API_PORT: OpenSearch port
+    OPENSEARCH_ADMIN_USERNAME: Admin username
+    OPENSEARCH_ADMIN_PASSWORD: Admin password
+
+Dependencies:
+    backend/shared_configs/configs.py
+    backend/onyx/document_index/opensearch/client.py
+"""
+
+import argparse
+import os
+import sys
+
+from onyx.document_index.opensearch.client import OpenSearchClient
+from onyx.document_index.opensearch.client import OpenSearchIndexClient
+from shared_configs.configs import MULTI_TENANT
+
+
+def list_indices(client: OpenSearchClient) -> None:
+    indices = client.list_indices_with_info()
+    print(f"Found {len(indices)} indices.")
+    print("-" * 80)
+    for index in sorted(indices, key=lambda x: x.name):
+        print(f"Index: {index.name}")
+        print(f"Health: {index.health}")
+        print(f"Status: {index.status}")
+        print(f"Num Primary Shards: {index.num_primary_shards}")
+        print(f"Num Replica Shards: {index.num_replica_shards}")
+        print(f"Docs Count: {index.docs_count}")
+        print(f"Docs Deleted: {index.docs_deleted}")
+        print(f"Created At: {index.created_at}")
+        print(f"Total Size: {index.total_size}")
+        print(f"Primary Shards Size: {index.primary_shards_size}")
+        print("-" * 80)
+
+
+def delete_index(client: OpenSearchIndexClient) -> None:
+    if not client.index_exists():
+        print(f"Index '{client._index_name}' does not exist.")
+        return
+
+    confirm = input(f"Delete index '{client._index_name}'? (yes/no): ")
+    if confirm.lower() != "yes":
+        print("Aborted.")
+        return
+
+    if client.delete_index():
+        print(f"Deleted index '{client._index_name}'.")
+    else:
+        print(f"Failed to delete index '{client._index_name}' for an unknown reason.")
+
+
+def main() -> None:
+    def add_standard_arguments(parser: argparse.ArgumentParser) -> None:
+        parser.add_argument(
+            "--host",
+            help="OpenSearch host. If not provided, will fall back to OPENSEARCH_HOST, then prompt "
+            "for input.",
+            type=str,
+            default=os.environ.get("OPENSEARCH_HOST", ""),
+        )
+        parser.add_argument(
+            "--port",
+            help="OpenSearch port. If not provided, will fall back to OPENSEARCH_REST_API_PORT, "
+            "then prompt for input.",
+            type=int,
+            default=int(os.environ.get("OPENSEARCH_REST_API_PORT", 0)),
+        )
+        parser.add_argument(
+            "--username",
+            help="OpenSearch username. If not provided, will fall back to OPENSEARCH_ADMIN_USERNAME, "
+            "then prompt for input.",
+            type=str,
+            default=os.environ.get("OPENSEARCH_ADMIN_USERNAME", ""),
+        )
+        parser.add_argument(
+            "--password",
+            help="OpenSearch password. If not provided, will fall back to OPENSEARCH_ADMIN_PASSWORD, "
+            "then prompt for input.",
+            type=str,
+            default=os.environ.get("OPENSEARCH_ADMIN_PASSWORD", ""),
+        )
+        parser.add_argument(
+            "--no-ssl", help="Disable SSL.", action="store_true", default=False
+        )
+        parser.add_argument(
+            "--no-verify-certs",
+            help="Disable certificate verification (for self-signed certs).",
+            action="store_true",
+            default=False,
+        )
+        parser.add_argument(
+            "--use-aws-managed-opensearch",
+            help="Whether to use AWS-managed OpenSearch. If not provided, will fall back to checking "
+            "USING_AWS_MANAGED_OPENSEARCH=='true', then default to False.",
+            action=argparse.BooleanOptionalAction,
+            default=os.environ.get("USING_AWS_MANAGED_OPENSEARCH", "").lower()
+            == "true",
+        )
+
+    parser = argparse.ArgumentParser(
+        description="A utility to interact with OpenSearch."
+    )
+    subparsers = parser.add_subparsers(
+        dest="command", help="Command to execute.", required=True
+    )
+
+    list_parser = subparsers.add_parser("list", help="List all indices with info.")
+    add_standard_arguments(list_parser)
+
+    delete_parser = subparsers.add_parser("delete", help="Delete an index.")
+    delete_parser.add_argument("index", help="Index name.", type=str)
+    add_standard_arguments(delete_parser)
+
+    args = parser.parse_args()
+
+    if not (host := args.host or input("Enter the OpenSearch host: ")):
+        print("Error: OpenSearch host is required.")
+        sys.exit(1)
+    if not (port := args.port or int(input("Enter the OpenSearch port: "))):
+        print("Error: OpenSearch port is required.")
+        sys.exit(1)
+    if not (username := args.username or input("Enter the OpenSearch username: ")):
+        print("Error: OpenSearch username is required.")
+        sys.exit(1)
+    if not (password := args.password or input("Enter the OpenSearch password: ")):
+        print("Error: OpenSearch password is required.")
+        sys.exit(1)
+    print("Using AWS-managed OpenSearch: ", args.use_aws_managed_opensearch)
+    print(f"MULTI_TENANT: {MULTI_TENANT}")
+
+    with (
+        OpenSearchIndexClient(
+            index_name=args.index,
+            host=host,
+            port=port,
+            auth=(username, password),
+            use_ssl=not args.no_ssl,
+            verify_certs=not args.no_verify_certs,
+        )
+        if args.command == "delete"
+        else OpenSearchClient(
+            host=host,
+            port=port,
+            auth=(username, password),
+            use_ssl=not args.no_ssl,
+            verify_certs=not args.no_verify_certs,
+        )
+    ) as client:
+        if not client.ping():
+            print("Error: Could not connect to OpenSearch.")
+            sys.exit(1)
+
+        if args.command == "list":
+            list_indices(client)
+        elif args.command == "delete":
+            delete_index(client)
+
+
+if __name__ == "__main__":
+    main()

backend/scripts/decrypt.py

@@ -1,48 +1,93 @@
+"""Decrypt a raw hex-encoded credential value.
+
+Usage:
+    python -m scripts.decrypt <hex_value>
+    python -m scripts.decrypt <hex_value> --key "my-encryption-key"
+    python -m scripts.decrypt <hex_value> --key ""
+
+Pass --key "" to skip decryption and just decode the raw bytes as UTF-8.
+Omit --key to use the current ENCRYPTION_KEY_SECRET from the environment.
+"""
+
+import argparse
 import binascii
 import json
+import os
 import sys
 
-from onyx.utils.encryption import decrypt_bytes_to_string
+parent_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+sys.path.append(parent_dir)
+
+from onyx.utils.encryption import decrypt_bytes_to_string  # noqa: E402
+from onyx.utils.variable_functionality import global_version  # noqa: E402
 
 
-def decrypt_raw_credential(encrypted_value: str) -> None:
-    """Decrypt and display a raw encrypted credential value
+def decrypt_raw_credential(encrypted_value: str, key: str | None = None) -> None:
+    """Decrypt and display a raw encrypted credential value.
 
     Args:
-        encrypted_value: The hex encoded encrypted credential value
+        encrypted_value: The hex-encoded encrypted credential value.
+        key: Encryption key to use. None means use ENCRYPTION_KEY_SECRET,
+             empty string means just decode as UTF-8.
     """
+    # Strip common hex prefixes
+    if encrypted_value.startswith("\\x"):
+        encrypted_value = encrypted_value[2:]
+    elif encrypted_value.startswith("x"):
+        encrypted_value = encrypted_value[1:]
+    print(encrypted_value)
+
     try:
-        # If string starts with 'x', remove it as it's just a prefix indicating hex
-        if encrypted_value.startswith("x"):
-            encrypted_value = encrypted_value[1:]
-        elif encrypted_value.startswith("\\x"):
-            encrypted_value = encrypted_value[2:]
-
-        # Convert hex string to bytes
-        encrypted_bytes = binascii.unhexlify(encrypted_value)
-
-        # Decrypt the bytes
-        decrypted_str = decrypt_bytes_to_string(encrypted_bytes)
-
-        # Parse and pretty print the decrypted JSON
-        decrypted_json = json.loads(decrypted_str)
-        print("Decrypted credential value:")
-        print(json.dumps(decrypted_json, indent=2))
-
+        raw_bytes = binascii.unhexlify(encrypted_value)
     except binascii.Error:
-        print("Error: Invalid hex encoded string")
+        print("Error: Invalid hex-encoded string")
+        sys.exit(1)
 
-    except json.JSONDecodeError as e:
-        print(f"Decrypted raw value (not JSON): {e}")
+    if key == "":
+        # Empty key → just decode as UTF-8, no decryption
+        try:
+            decrypted_str = raw_bytes.decode("utf-8")
+        except UnicodeDecodeError as e:
+            print(f"Error decoding bytes as UTF-8: {e}")
+            sys.exit(1)
+    else:
+        print(key)
+        try:
+            decrypted_str = decrypt_bytes_to_string(raw_bytes, key=key)
+        except Exception as e:
+            print(f"Error decrypting value: {e}")
+            sys.exit(1)
 
-    except Exception as e:
-        print(f"Error decrypting value: {e}")
+    # Try to pretty-print as JSON, otherwise print raw
+    try:
+        parsed = json.loads(decrypted_str)
+        print(json.dumps(parsed, indent=2))
+    except json.JSONDecodeError:
+        print(decrypted_str)
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(
+        description="Decrypt a hex-encoded credential value."
+    )
+    parser.add_argument(
+        "value",
+        help="Hex-encoded encrypted value to decrypt.",
+    )
+    parser.add_argument(
+        "--key",
+        default=None,
+        help=(
+            "Encryption key. Omit to use ENCRYPTION_KEY_SECRET from env. "
+            'Pass "" (empty) to just decode as UTF-8 without decryption.'
+        ),
+    )
+    args = parser.parse_args()
+
+    global_version.set_ee()
+    decrypt_raw_credential(args.value, key=args.key)
+    global_version.unset_ee()
 
 
 if __name__ == "__main__":
-    if len(sys.argv) != 2:
-        print("Usage: python decrypt.py <hex_encoded_encrypted_value>")
-        sys.exit(1)
-
-    encrypted_value = sys.argv[1]
-    decrypt_raw_credential(encrypted_value)
+    main()

backend/scripts/reencrypt_secrets.py

@@ -0,0 +1,107 @@
+"""Re-encrypt secrets under the current ENCRYPTION_KEY_SECRET.
+
+Decrypts all encrypted columns using the old key (or raw decode if the old key
+is empty), then re-encrypts them with the current ENCRYPTION_KEY_SECRET.
+
+Usage (docker):
+    docker exec -it onyx-api_server-1 \
+        python -m scripts.reencrypt_secrets --old-key "previous-key"
+
+Usage (kubernetes):
+    kubectl exec -it <pod> -- \
+        python -m scripts.reencrypt_secrets --old-key "previous-key"
+
+Omit --old-key (or pass "") if secrets were not previously encrypted.
+
+For multi-tenant deployments, pass --tenant-id to target a specific tenant,
+or --all-tenants to iterate every tenant.
+"""
+
+import argparse
+import os
+import sys
+
+parent_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+sys.path.append(parent_dir)
+
+from onyx.db.rotate_encryption_key import rotate_encryption_key  # noqa: E402
+from onyx.db.engine.sql_engine import get_session_with_tenant  # noqa: E402
+from onyx.db.engine.sql_engine import SqlEngine  # noqa: E402
+from onyx.db.engine.tenant_utils import get_all_tenant_ids  # noqa: E402
+from onyx.utils.variable_functionality import global_version  # noqa: E402
+from shared_configs.configs import POSTGRES_DEFAULT_SCHEMA  # noqa: E402
+
+
+def _run_for_tenant(tenant_id: str, old_key: str | None, dry_run: bool = False) -> None:
+    print(f"Re-encrypting secrets for tenant: {tenant_id}")
+    with get_session_with_tenant(tenant_id=tenant_id) as db_session:
+        results = rotate_encryption_key(db_session, old_key=old_key, dry_run=dry_run)
+
+    if results:
+        for col, count in results.items():
+            print(
+                f"  {col}: {count} row(s) {'would be ' if dry_run else ''}re-encrypted"
+            )
+    else:
+        print("No rows needed re-encryption.")
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(
+        description="Re-encrypt secrets under the current encryption key."
+    )
+    parser.add_argument(
+        "--old-key",
+        default=None,
+        help="Previous encryption key. Omit or pass empty string if not applicable.",
+    )
+    parser.add_argument(
+        "--dry-run",
+        action="store_true",
+        help="Show what would be re-encrypted without making changes.",
+    )
+
+    tenant_group = parser.add_mutually_exclusive_group()
+    tenant_group.add_argument(
+        "--tenant-id",
+        default=None,
+        help="Target a specific tenant schema.",
+    )
+    tenant_group.add_argument(
+        "--all-tenants",
+        action="store_true",
+        help="Iterate all tenants.",
+    )
+
+    args = parser.parse_args()
+
+    old_key = args.old_key if args.old_key else None
+
+    global_version.set_ee()
+    SqlEngine.init_engine(pool_size=5, max_overflow=2)
+
+    if args.dry_run:
+        print("DRY RUN — no changes will be made")
+
+    if args.all_tenants:
+        tenant_ids = get_all_tenant_ids()
+        print(f"Found {len(tenant_ids)} tenant(s)")
+        failed_tenants: list[str] = []
+        for tid in tenant_ids:
+            try:
+                _run_for_tenant(tid, old_key, dry_run=args.dry_run)
+            except Exception as e:
+                print(f"  ERROR for tenant {tid}: {e}")
+                failed_tenants.append(tid)
+        if failed_tenants:
+            print(f"FAILED tenants ({len(failed_tenants)}): {failed_tenants}")
+            sys.exit(1)
+    else:
+        tenant_id = args.tenant_id or POSTGRES_DEFAULT_SCHEMA
+        _run_for_tenant(tenant_id, old_key, dry_run=args.dry_run)
+
+    print("Done.")
+
+
+if __name__ == "__main__":
+    main()

backend/tests/README.md

@@ -0,0 +1,71 @@
+# Backend Tests
+
+## Test Types
+
+There are four test categories, ordered by increasing scope:
+
+### Unit Tests (`tests/unit/`)
+
+No external services. Mock all I/O with `unittest.mock`. Use for complex, isolated
+logic (e.g. citation processing, encryption).
+
+```bash
+pytest -xv backend/tests/unit
+```
+
+### External Dependency Unit Tests (`tests/external_dependency_unit/`)
+
+External services (Postgres, Redis, Vespa, OpenAI, etc.) are running, but Onyx
+application containers are not. Tests call functions directly and can mock selectively.
+
+Use when you need a real database or real API calls but want control over setup.
+
+```bash
+python -m dotenv -f .vscode/.env run -- pytest backend/tests/external_dependency_unit
+```
+
+### Integration Tests (`tests/integration/`)
+
+Full Onyx deployment running. No mocking. Prefer this over other test types when possible.
+
+```bash
+python -m dotenv -f .vscode/.env run -- pytest backend/tests/integration
+```
+
+### Playwright / E2E Tests (`web/tests/e2e/`)
+
+Full stack including web server. Use for frontend-backend coordination.
+
+```bash
+npx playwright test <TEST_NAME>
+```
+
+## Shared Fixtures
+
+Shared fixtures live in `backend/tests/conftest.py`. Test subdirectories can define
+their own `conftest.py` for directory-scoped fixtures.
+
+## Best Practices
+
+### Use `enable_ee` fixture instead of inlining
+
+Enables EE mode for a test, with proper teardown and cache clearing.
+
+```python
+# Whole file (in a test module, NOT in conftest.py)
+pytestmark = pytest.mark.usefixtures("enable_ee")
+
+# Whole directory — add an autouse wrapper to the directory's conftest.py
+@pytest.fixture(autouse=True)
+def _enable_ee_for_directory(enable_ee: None) -> None:  # noqa: ARG001
+    """Wraps the shared enable_ee fixture with autouse for this directory."""
+
+# Single test
+def test_something(enable_ee: None) -> None: ...
+```
+
+**Note:** `pytestmark` in a `conftest.py` does NOT apply markers to tests in that
+directory — it only affects tests defined in the conftest itself (which is none).
+Use the autouse fixture wrapper pattern shown above instead.
+
+Do NOT inline `global_version.set_ee()` — always use the fixture.

backend/tests/conftest.py

@@ -0,0 +1,24 @@
+"""Root conftest — shared fixtures available to all test directories."""
+
+from collections.abc import Generator
+
+import pytest
+
+from onyx.utils.variable_functionality import fetch_versioned_implementation
+from onyx.utils.variable_functionality import global_version
+
+
+@pytest.fixture()
+def enable_ee() -> Generator[None, None, None]:
+    """Temporarily enable EE mode for a single test.
+
+    Restores the previous EE state and clears the versioned-implementation
+    cache on teardown so state doesn't leak between tests.
+    """
+    was_ee = global_version.is_ee_version()
+    global_version.set_ee()
+    fetch_versioned_implementation.cache_clear()
+    yield
+    if not was_ee:
+        global_version.unset_ee()
+    fetch_versioned_implementation.cache_clear()

backend/tests/daily/connectors/confluence/test_confluence_permissions_basic.py

@@ -45,7 +45,7 @@ def confluence_connector() -> ConfluenceConnector:
 def test_confluence_connector_permissions(
     mock_get_api_key: MagicMock,  # noqa: ARG001
     confluence_connector: ConfluenceConnector,
-    set_ee_on: None,  # noqa: ARG001
+    enable_ee: None,  # noqa: ARG001
 ) -> None:
     # Get all doc IDs from the full connector
     all_full_doc_ids = set()
@@ -93,7 +93,7 @@ def test_confluence_connector_permissions(
 def test_confluence_connector_restriction_handling(
     mock_get_api_key: MagicMock,  # noqa: ARG001
     mock_db_provider_class: MagicMock,
-    set_ee_on: None,  # noqa: ARG001
+    enable_ee: None,  # noqa: ARG001
 ) -> None:
     # Test space key
     test_space_key = "DailyPermS"

backend/tests/daily/connectors/conftest.py

@@ -4,8 +4,6 @@ from unittest.mock import patch
 
 import pytest
 
-from onyx.utils.variable_functionality import global_version
-
 
 @pytest.fixture
 def mock_get_unstructured_api_key() -> Generator[MagicMock, None, None]:
@@ -14,14 +12,3 @@ def mock_get_unstructured_api_key() -> Generator[MagicMock, None, None]:
         return_value=None,
     ) as mock:
         yield mock
-
-
-@pytest.fixture
-def set_ee_on() -> Generator[None, None, None]:
-    """Need EE to be enabled for these tests to work since
-    perm syncing is a an EE-only feature."""
-    global_version.set_ee()
-
-    yield
-
-    global_version._is_ee = False

backend/tests/daily/connectors/google_drive/test_drive_perm_sync.py

@@ -98,7 +98,7 @@ def _build_connector(
 
 def test_gdrive_perm_sync_with_real_data(
     google_drive_service_acct_connector_factory: Callable[..., GoogleDriveConnector],
-    set_ee_on: None,  # noqa: ARG001
+    enable_ee: None,  # noqa: ARG001
 ) -> None:
     """
     Test gdrive_doc_sync and gdrive_group_sync with real data from the test drive.

backend/tests/daily/connectors/slack/test_slack_perm_sync.py

@@ -1,12 +1,10 @@
 import time
-from collections.abc import Generator
 
 import pytest
 
 from onyx.connectors.models import HierarchyNode
 from onyx.connectors.models import SlimDocument
 from onyx.connectors.slack.connector import SlackConnector
-from onyx.utils.variable_functionality import global_version
 from tests.daily.connectors.utils import load_all_from_connector
 
 
@@ -19,16 +17,7 @@ PRIVATE_CHANNEL_USERS = [
     "test_user_2@onyx-test.com",
 ]
 
-
-@pytest.fixture(autouse=True)
-def set_ee_on() -> Generator[None, None, None]:
-    """Need EE to be enabled for these tests to work since
-    perm syncing is a an EE-only feature."""
-    global_version.set_ee()
-
-    yield
-
-    global_version._is_ee = False
+pytestmark = pytest.mark.usefixtures("enable_ee")
 
 
 @pytest.mark.parametrize(

backend/tests/daily/connectors/teams/test_teams_connector.py

@@ -1,13 +1,11 @@
 import os
 import time
-from collections.abc import Generator
 
 import pytest
 
 from onyx.access.models import ExternalAccess
 from onyx.connectors.models import HierarchyNode
 from onyx.connectors.teams.connector import TeamsConnector
-from onyx.utils.variable_functionality import global_version
 from tests.daily.connectors.teams.models import TeamsThread
 from tests.daily.connectors.utils import load_all_from_connector
 
@@ -168,18 +166,9 @@ def test_slim_docs_retrieval_from_teams_connector(
         _assert_is_valid_external_access(external_access=slim_doc.external_access)
 
 
-@pytest.fixture(autouse=False)
-def set_ee_on() -> Generator[None, None, None]:
-    """Need EE to be enabled for perm sync tests to work since
-    perm syncing is an EE-only feature."""
-    global_version.set_ee()
-    yield
-    global_version._is_ee = False
-
-
 def test_load_from_checkpoint_with_perm_sync(
     teams_connector: TeamsConnector,
-    set_ee_on: None,  # noqa: ARG001
+    enable_ee: None,  # noqa: ARG001
 ) -> None:
     """Test that load_from_checkpoint_with_perm_sync returns documents with external_access.
 

backend/tests/external_dependency_unit/celery/test_docprocessing_priority.py

@@ -145,6 +145,10 @@ class TestDocprocessingPriorityInDocumentExtraction:
     @patch("onyx.background.indexing.run_docfetching.get_document_batch_storage")
     @patch("onyx.background.indexing.run_docfetching.MemoryTracer")
     @patch("onyx.background.indexing.run_docfetching._get_connector_runner")
+    @patch(
+        "onyx.background.indexing.run_docfetching.strip_null_characters",
+        side_effect=lambda batch: batch,
+    )
     @patch(
         "onyx.background.indexing.run_docfetching.get_recent_completed_attempts_for_cc_pair"
     )
@@ -169,6 +173,7 @@ class TestDocprocessingPriorityInDocumentExtraction:
         mock_save_checkpoint: MagicMock,  # noqa: ARG002
         mock_get_last_successful_attempt_poll_range_end: MagicMock,
         mock_get_recent_completed_attempts: MagicMock,
+        mock_strip_null_characters: MagicMock,  # noqa: ARG002
         mock_get_connector_runner: MagicMock,
         mock_memory_tracer_class: MagicMock,
         mock_get_batch_storage: MagicMock,

backend/tests/external_dependency_unit/connectors/jira/test_jira_doc_sync.py

@@ -1,5 +1,6 @@
 from typing import Any
 
+import pytest
 from pydantic import BaseModel
 from sqlalchemy.orm import Session
 
@@ -14,13 +15,14 @@ from onyx.db.models import ConnectorCredentialPair
 from onyx.db.models import Credential
 from onyx.db.utils import DocumentRow
 from onyx.db.utils import SortOrder
-from onyx.utils.variable_functionality import global_version
 
 
 # In order to get these tests to run, use the credentials from Bitwarden.
 # Search up "ENV vars for local and Github tests", and find the Jira relevant key-value pairs.
 # Required env vars: JIRA_USER_EMAIL, JIRA_API_TOKEN
 
+pytestmark = pytest.mark.usefixtures("enable_ee")
+
 
 class DocExternalAccessSet(BaseModel):
     """A version of DocExternalAccess that uses sets for comparison."""
@@ -52,9 +54,6 @@ def test_jira_doc_sync(
     This test uses the AS project which has applicationRole permission,
     meaning all documents should be marked as public.
     """
-    # NOTE: must set EE on or else the connector will skip the perm syncing
-    global_version.set_ee()
-
     try:
         # Use AS project specifically for this test
         connector_config = {
@@ -150,9 +149,6 @@ def test_jira_doc_sync_with_specific_permissions(
     This test uses a project that has specific user permissions to verify
     that specific users are correctly extracted.
     """
-    # NOTE: must set EE on or else the connector will skip the perm syncing
-    global_version.set_ee()
-
     try:
         # Use SUP project which has specific user permissions
         connector_config = {

backend/tests/external_dependency_unit/connectors/jira/test_jira_group_sync.py

@@ -1,5 +1,6 @@
 from typing import Any
 
+import pytest
 from sqlalchemy.orm import Session
 
 from ee.onyx.external_permissions.jira.group_sync import jira_group_sync
@@ -18,6 +19,8 @@ from tests.daily.connectors.confluence.models import ExternalUserGroupSet
 # Search up "ENV vars for local and Github tests", and find the Jira relevant key-value pairs.
 # Required env vars: JIRA_USER_EMAIL, JIRA_API_TOKEN
 
+pytestmark = pytest.mark.usefixtures("enable_ee")
+
 # Expected groups from the danswerai.atlassian.net Jira instance
 # Note: These groups are shared with Confluence since they're both Atlassian products
 # App accounts (bots, integrations) are filtered out

backend/tests/external_dependency_unit/db/test_credential_sensitive_value.py

@@ -0,0 +1,90 @@
+"""Test that Credential with nested JSON round-trips through SensitiveValue correctly.
+
+Exercises the full encrypt → store → read → decrypt → SensitiveValue path
+with realistic nested OAuth credential data, and verifies SQLAlchemy dirty
+tracking works with nested dict comparison.
+
+Requires a running Postgres instance.
+"""
+
+from sqlalchemy.orm import Session
+
+from onyx.configs.constants import DocumentSource
+from onyx.db.models import Credential
+from onyx.utils.sensitive import SensitiveValue
+
+# NOTE: this is not the real shape of a Drive credential,
+# but it is intended to test nested JSON credential handling
+
+_NESTED_CRED_JSON = {
+    "oauth_tokens": {
+        "access_token": "ya29.abc123",
+        "refresh_token": "1//xEg-def456",
+    },
+    "scopes": ["read", "write", "admin"],
+    "client_config": {
+        "client_id": "123.apps.googleusercontent.com",
+        "client_secret": "GOCSPX-secret",
+    },
+}
+
+
+def test_nested_credential_json_round_trip(db_session: Session) -> None:
+    """Nested OAuth credential survives encrypt → store → read → decrypt."""
+    credential = Credential(
+        source=DocumentSource.GOOGLE_DRIVE,
+        credential_json=_NESTED_CRED_JSON,
+    )
+    db_session.add(credential)
+    db_session.flush()
+
+    # Immediate read (no DB round-trip) — tests the set event wrapping
+    assert isinstance(credential.credential_json, SensitiveValue)
+    assert credential.credential_json.get_value(apply_mask=False) == _NESTED_CRED_JSON
+
+    # DB round-trip — tests process_result_value
+    db_session.expire(credential)
+    reloaded = credential.credential_json
+    assert isinstance(reloaded, SensitiveValue)
+    assert reloaded.get_value(apply_mask=False) == _NESTED_CRED_JSON
+
+    db_session.rollback()
+
+
+def test_reassign_same_nested_json_not_dirty(db_session: Session) -> None:
+    """Re-assigning the same nested dict should not mark the session dirty."""
+    credential = Credential(
+        source=DocumentSource.GOOGLE_DRIVE,
+        credential_json=_NESTED_CRED_JSON,
+    )
+    db_session.add(credential)
+    db_session.flush()
+
+    # Clear dirty state from the insert
+    db_session.expire(credential)
+    _ = credential.credential_json  # force reload
+
+    # Re-assign identical value
+    credential.credential_json = _NESTED_CRED_JSON  # type: ignore[assignment]
+    assert not db_session.is_modified(credential)
+
+    db_session.rollback()
+
+
+def test_assign_different_nested_json_is_dirty(db_session: Session) -> None:
+    """Assigning a different nested dict should mark the session dirty."""
+    credential = Credential(
+        source=DocumentSource.GOOGLE_DRIVE,
+        credential_json=_NESTED_CRED_JSON,
+    )
+    db_session.add(credential)
+    db_session.flush()
+
+    db_session.expire(credential)
+    _ = credential.credential_json  # force reload
+
+    modified_cred = {**_NESTED_CRED_JSON, "scopes": ["read"]}
+    credential.credential_json = modified_cred  # type: ignore[assignment]
+    assert db_session.is_modified(credential)
+
+    db_session.rollback()

backend/tests/external_dependency_unit/db/test_rotate_encryption_key.py

@@ -0,0 +1,305 @@
+"""Tests for rotate_encryption_key against real Postgres.
+
+Uses real ORM models (Credential, InternetSearchProvider) and the actual
+Postgres database. Discovery is mocked in rotation tests to scope mutations
+to only the test rows — the real _discover_encrypted_columns walk is tested
+separately in TestDiscoverEncryptedColumns.
+
+Requires a running Postgres instance. Run with::
+
+    python -m dotenv -f .vscode/.env run -- pytest tests/external_dependency_unit/db/test_rotate_encryption_key.py
+"""
+
+import json
+from collections.abc import Generator
+from unittest.mock import patch
+
+import pytest
+from sqlalchemy import LargeBinary
+from sqlalchemy import select
+from sqlalchemy import text
+from sqlalchemy.orm import Session
+
+from ee.onyx.utils.encryption import _decrypt_bytes
+from ee.onyx.utils.encryption import _encrypt_string
+from ee.onyx.utils.encryption import _get_trimmed_key
+from onyx.configs.constants import DocumentSource
+from onyx.db.models import Credential
+from onyx.db.models import EncryptedJson
+from onyx.db.models import EncryptedString
+from onyx.db.models import InternetSearchProvider
+from onyx.db.rotate_encryption_key import _discover_encrypted_columns
+from onyx.db.rotate_encryption_key import rotate_encryption_key
+from onyx.utils.variable_functionality import fetch_versioned_implementation
+from onyx.utils.variable_functionality import global_version
+
+EE_MODULE = "ee.onyx.utils.encryption"
+ROTATE_MODULE = "onyx.db.rotate_encryption_key"
+
+OLD_KEY = "o" * 16
+NEW_KEY = "n" * 16
+
+
+@pytest.fixture(autouse=True)
+def _enable_ee() -> Generator[None, None, None]:
+    prev = global_version._is_ee
+    global_version.set_ee()
+    fetch_versioned_implementation.cache_clear()
+    yield
+    global_version._is_ee = prev
+    fetch_versioned_implementation.cache_clear()
+
+
+@pytest.fixture(autouse=True)
+def _clear_key_cache() -> None:
+    _get_trimmed_key.cache_clear()
+
+
+def _raw_credential_bytes(db_session: Session, credential_id: int) -> bytes | None:
+    """Read raw bytes from credential_json, bypassing the TypeDecorator."""
+    col = Credential.__table__.c.credential_json
+    stmt = select(col.cast(LargeBinary)).where(
+        Credential.__table__.c.id == credential_id
+    )
+    return db_session.execute(stmt).scalar()
+
+
+def _raw_isp_bytes(db_session: Session, isp_id: int) -> bytes | None:
+    """Read raw bytes from InternetSearchProvider.api_key."""
+    col = InternetSearchProvider.__table__.c.api_key
+    stmt = select(col.cast(LargeBinary)).where(
+        InternetSearchProvider.__table__.c.id == isp_id
+    )
+    return db_session.execute(stmt).scalar()
+
+
+class TestDiscoverEncryptedColumns:
+    """Verify _discover_encrypted_columns finds real production models."""
+
+    def test_discovers_credential_json(self) -> None:
+        results = _discover_encrypted_columns()
+        found = {
+            (model_cls.__tablename__, col_name, is_json)  # type: ignore[attr-defined]
+            for model_cls, col_name, _, is_json in results
+        }
+        assert ("credential", "credential_json", True) in found
+
+    def test_discovers_internet_search_provider_api_key(self) -> None:
+        results = _discover_encrypted_columns()
+        found = {
+            (model_cls.__tablename__, col_name, is_json)  # type: ignore[attr-defined]
+            for model_cls, col_name, _, is_json in results
+        }
+        assert ("internet_search_provider", "api_key", False) in found
+
+    def test_all_encrypted_string_columns_are_not_json(self) -> None:
+        results = _discover_encrypted_columns()
+        for model_cls, col_name, _, is_json in results:
+            col = getattr(model_cls, col_name).property.columns[0]
+            if isinstance(col.type, EncryptedString):
+                assert not is_json, (
+                    f"{model_cls.__tablename__}.{col_name} is EncryptedString "  # type: ignore[attr-defined]
+                    f"but is_json={is_json}"
+                )
+
+    def test_all_encrypted_json_columns_are_json(self) -> None:
+        results = _discover_encrypted_columns()
+        for model_cls, col_name, _, is_json in results:
+            col = getattr(model_cls, col_name).property.columns[0]
+            if isinstance(col.type, EncryptedJson):
+                assert is_json, (
+                    f"{model_cls.__tablename__}.{col_name} is EncryptedJson "  # type: ignore[attr-defined]
+                    f"but is_json={is_json}"
+                )
+
+
+class TestRotateCredential:
+    """Test rotation against the real Credential table (EncryptedJson).
+
+    Discovery is scoped to only the Credential model to avoid mutating
+    other tables in the test database.
+    """
+
+    @pytest.fixture(autouse=True)
+    def _limit_discovery(self) -> Generator[None, None, None]:
+        with patch(
+            f"{ROTATE_MODULE}._discover_encrypted_columns",
+            return_value=[(Credential, "credential_json", ["id"], True)],
+        ):
+            yield
+
+    @pytest.fixture()
+    def credential_id(
+        self, db_session: Session, tenant_context: None  # noqa: ARG002
+    ) -> Generator[int, None, None]:
+        """Insert a Credential row with raw encrypted bytes, clean up after."""
+        config = {"api_key": "sk-test-1234", "endpoint": "https://example.com"}
+        encrypted = _encrypt_string(json.dumps(config), key=OLD_KEY)
+
+        result = db_session.execute(
+            text(
+                "INSERT INTO credential "
+                "(source, credential_json, admin_public, curator_public) "
+                "VALUES (:source, :cred_json, true, false) "
+                "RETURNING id"
+            ),
+            {"source": DocumentSource.INGESTION_API.value, "cred_json": encrypted},
+        )
+        cred_id = result.scalar_one()
+        db_session.commit()
+
+        yield cred_id
+
+        db_session.execute(
+            text("DELETE FROM credential WHERE id = :id"), {"id": cred_id}
+        )
+        db_session.commit()
+
+    def test_rotates_credential_json(
+        self, db_session: Session, credential_id: int
+    ) -> None:
+        with (
+            patch(f"{ROTATE_MODULE}.ENCRYPTION_KEY_SECRET", NEW_KEY),
+            patch(f"{EE_MODULE}.ENCRYPTION_KEY_SECRET", NEW_KEY),
+        ):
+            totals = rotate_encryption_key(db_session, old_key=OLD_KEY)
+
+        assert totals.get("credential.credential_json", 0) >= 1
+
+        raw = _raw_credential_bytes(db_session, credential_id)
+        assert raw is not None
+        decrypted = json.loads(_decrypt_bytes(raw, key=NEW_KEY))
+        assert decrypted["api_key"] == "sk-test-1234"
+        assert decrypted["endpoint"] == "https://example.com"
+
+    def test_skips_already_rotated(
+        self, db_session: Session, credential_id: int
+    ) -> None:
+        with (
+            patch(f"{ROTATE_MODULE}.ENCRYPTION_KEY_SECRET", NEW_KEY),
+            patch(f"{EE_MODULE}.ENCRYPTION_KEY_SECRET", NEW_KEY),
+        ):
+            rotate_encryption_key(db_session, old_key=OLD_KEY)
+            _ = rotate_encryption_key(db_session, old_key=OLD_KEY)
+
+        raw = _raw_credential_bytes(db_session, credential_id)
+        assert raw is not None
+        decrypted = json.loads(_decrypt_bytes(raw, key=NEW_KEY))
+        assert decrypted["api_key"] == "sk-test-1234"
+
+    def test_dry_run_does_not_modify(
+        self, db_session: Session, credential_id: int
+    ) -> None:
+        original = _raw_credential_bytes(db_session, credential_id)
+
+        with (
+            patch(f"{ROTATE_MODULE}.ENCRYPTION_KEY_SECRET", NEW_KEY),
+            patch(f"{EE_MODULE}.ENCRYPTION_KEY_SECRET", NEW_KEY),
+        ):
+            totals = rotate_encryption_key(db_session, old_key=OLD_KEY, dry_run=True)
+
+        assert totals.get("credential.credential_json", 0) >= 1
+
+        raw_after = _raw_credential_bytes(db_session, credential_id)
+        assert raw_after == original
+
+
+class TestRotateInternetSearchProvider:
+    """Test rotation against the real InternetSearchProvider table (EncryptedString).
+
+    Discovery is scoped to only the InternetSearchProvider model to avoid
+    mutating other tables in the test database.
+    """
+
+    @pytest.fixture(autouse=True)
+    def _limit_discovery(self) -> Generator[None, None, None]:
+        with patch(
+            f"{ROTATE_MODULE}._discover_encrypted_columns",
+            return_value=[
+                (InternetSearchProvider, "api_key", ["id"], False),
+            ],
+        ):
+            yield
+
+    @pytest.fixture()
+    def isp_id(
+        self, db_session: Session, tenant_context: None  # noqa: ARG002
+    ) -> Generator[int, None, None]:
+        """Insert an InternetSearchProvider row with raw encrypted bytes."""
+        encrypted = _encrypt_string("sk-secret-api-key", key=OLD_KEY)
+
+        result = db_session.execute(
+            text(
+                "INSERT INTO internet_search_provider "
+                "(name, provider_type, api_key, is_active) "
+                "VALUES (:name, :ptype, :api_key, false) "
+                "RETURNING id"
+            ),
+            {
+                "name": f"test-rotation-{id(self)}",
+                "ptype": "test",
+                "api_key": encrypted,
+            },
+        )
+        isp_id = result.scalar_one()
+        db_session.commit()
+
+        yield isp_id
+
+        db_session.execute(
+            text("DELETE FROM internet_search_provider WHERE id = :id"),
+            {"id": isp_id},
+        )
+        db_session.commit()
+
+    def test_rotates_api_key(self, db_session: Session, isp_id: int) -> None:
+        with (
+            patch(f"{ROTATE_MODULE}.ENCRYPTION_KEY_SECRET", NEW_KEY),
+            patch(f"{EE_MODULE}.ENCRYPTION_KEY_SECRET", NEW_KEY),
+        ):
+            totals = rotate_encryption_key(db_session, old_key=OLD_KEY)
+
+        assert totals.get("internet_search_provider.api_key", 0) >= 1
+
+        raw = _raw_isp_bytes(db_session, isp_id)
+        assert raw is not None
+        assert _decrypt_bytes(raw, key=NEW_KEY) == "sk-secret-api-key"
+
+    def test_rotates_from_unencrypted(
+        self, db_session: Session, tenant_context: None  # noqa: ARG002
+    ) -> None:
+        """Test rotating data that was stored without any encryption key."""
+        result = db_session.execute(
+            text(
+                "INSERT INTO internet_search_provider "
+                "(name, provider_type, api_key, is_active) "
+                "VALUES (:name, :ptype, :api_key, false) "
+                "RETURNING id"
+            ),
+            {
+                "name": f"test-raw-{id(self)}",
+                "ptype": "test",
+                "api_key": b"raw-api-key",
+            },
+        )
+        isp_id = result.scalar_one()
+        db_session.commit()
+
+        try:
+            with (
+                patch(f"{ROTATE_MODULE}.ENCRYPTION_KEY_SECRET", NEW_KEY),
+                patch(f"{EE_MODULE}.ENCRYPTION_KEY_SECRET", NEW_KEY),
+            ):
+                totals = rotate_encryption_key(db_session, old_key=None)
+
+            assert totals.get("internet_search_provider.api_key", 0) >= 1
+
+            raw = _raw_isp_bytes(db_session, isp_id)
+            assert raw is not None
+            assert _decrypt_bytes(raw, key=NEW_KEY) == "raw-api-key"
+        finally:
+            db_session.execute(
+                text("DELETE FROM internet_search_provider WHERE id = :id"),
+                {"id": isp_id},
+            )
+            db_session.commit()

backend/tests/external_dependency_unit/llm/test_llm_provider_auto_mode.py

@@ -698,6 +698,99 @@ class TestAutoModeMissingFlows:
 class TestAutoModeTransitionsAndResync:
     """Tests for auto/manual transitions, config evolution, and sync idempotency."""
 
+    def test_transition_to_auto_mode_preserves_default(
+        self,
+        db_session: Session,
+        provider_name: str,
+    ) -> None:
+        """When the default provider transitions from manual to auto mode,
+        the global default should be preserved (set to the recommended model).
+
+        Steps:
+        1. Create a manual-mode provider with models, set it as global default.
+        2. Transition to auto mode (model_configurations=[] triggers cascade
+           delete of old ModelConfigurations and their LLMModelFlow rows).
+        3. Verify the provider is still the global default, now using the
+           recommended default model from the GitHub config.
+        """
+        initial_models = [
+            ModelConfigurationUpsertRequest(name="gpt-4o", is_visible=True),
+            ModelConfigurationUpsertRequest(name="gpt-4o-mini", is_visible=True),
+        ]
+
+        auto_config = _create_mock_llm_recommendations(
+            provider=LlmProviderNames.OPENAI,
+            default_model_name="gpt-4o-mini",
+            additional_models=["gpt-4o"],
+        )
+
+        try:
+            # Step 1: Create manual-mode provider and set as default
+            put_llm_provider(
+                llm_provider_upsert_request=LLMProviderUpsertRequest(
+                    name=provider_name,
+                    provider=LlmProviderNames.OPENAI,
+                    api_key="sk-test-key-00000000000000000000000000000000000",
+                    api_key_changed=True,
+                    is_auto_mode=False,
+                    model_configurations=initial_models,
+                ),
+                is_creation=True,
+                _=_create_mock_admin(),
+                db_session=db_session,
+            )
+
+            db_session.expire_all()
+            provider = fetch_existing_llm_provider(
+                name=provider_name, db_session=db_session
+            )
+            assert provider is not None
+            update_default_provider(provider.id, "gpt-4o", db_session)
+
+            default_before = fetch_default_llm_model(db_session)
+            assert default_before is not None
+            assert default_before.name == "gpt-4o"
+            assert default_before.llm_provider_id == provider.id
+
+            # Step 2: Transition to auto mode
+            with patch(
+                "onyx.server.manage.llm.api.fetch_llm_recommendations_from_github",
+                return_value=auto_config,
+            ):
+                put_llm_provider(
+                    llm_provider_upsert_request=LLMProviderUpsertRequest(
+                        id=provider.id,
+                        name=provider_name,
+                        provider=LlmProviderNames.OPENAI,
+                        api_key=None,
+                        api_key_changed=False,
+                        is_auto_mode=True,
+                        model_configurations=[],
+                    ),
+                    is_creation=False,
+                    _=_create_mock_admin(),
+                    db_session=db_session,
+                )
+
+            # Step 3: Default should be preserved on this provider
+            db_session.expire_all()
+            default_after = fetch_default_llm_model(db_session)
+            assert default_after is not None, (
+                "Default model should not be None after transitioning to auto mode — "
+                "the provider was the default before and should remain so"
+            )
+            assert (
+                default_after.llm_provider_id == provider.id
+            ), "Default should still belong to the same provider after transition"
+            assert default_after.name == "gpt-4o-mini", (
+                f"Default should be updated to the recommended model 'gpt-4o-mini', "
+                f"got '{default_after.name}'"
+            )
+
+        finally:
+            db_session.rollback()
+            _cleanup_provider(db_session, provider_name)
+
     def test_auto_to_manual_mode_preserves_models_and_stops_syncing(
         self,
         db_session: Session,
@@ -1042,14 +1135,19 @@ class TestAutoModeTransitionsAndResync:
             assert visibility["gpt-4o"] is False, "Removed default should be hidden"
             assert visibility["gpt-4o-mini"] is True, "New default should be visible"
 
-            # The LLMModelFlow row for gpt-4o still exists (is_default=True),
-            # but the model is hidden. fetch_default_llm_model filters on
-            # is_visible=True, so it should NOT return gpt-4o.
+            # The old default (gpt-4o) is now hidden. sync_auto_mode_models
+            # should update the global default to the new recommended default
+            # (gpt-4o-mini) so that it is not silently lost.
             db_session.expire_all()
             default_after = fetch_default_llm_model(db_session)
-            assert (
-                default_after is None or default_after.name != "gpt-4o"
-            ), "Hidden model should not be returned as the default"
+            assert default_after is not None, (
+                "Default model should not be None — sync should set the new "
+                "recommended default when the old one is hidden"
+            )
+            assert default_after.name == "gpt-4o-mini", (
+                f"Default should be updated to the new recommended model "
+                f"'gpt-4o-mini', but got '{default_after.name}'"
+            )
 
         finally:
             db_session.rollback()

backend/tests/external_dependency_unit/slack_bot/test_slack_bot_crud.py

@@ -0,0 +1,85 @@
+"""Tests that SlackBot CRUD operations return properly typed SensitiveValue fields.
+
+Regression test for the bug where insert_slack_bot/update_slack_bot returned
+objects with raw string tokens instead of SensitiveValue wrappers, causing
+'str object has no attribute get_value' errors in SlackBot.from_model().
+"""
+
+from uuid import uuid4
+
+from sqlalchemy.orm import Session
+
+from onyx.db.slack_bot import insert_slack_bot
+from onyx.db.slack_bot import update_slack_bot
+from onyx.server.manage.models import SlackBot
+from onyx.utils.sensitive import SensitiveValue
+
+
+def _unique(prefix: str) -> str:
+    return f"{prefix}-{uuid4().hex[:8]}"
+
+
+def test_insert_slack_bot_returns_sensitive_values(db_session: Session) -> None:
+    bot_token = _unique("xoxb-insert")
+    app_token = _unique("xapp-insert")
+    user_token = _unique("xoxp-insert")
+
+    slack_bot = insert_slack_bot(
+        db_session=db_session,
+        name=_unique("test-bot-insert"),
+        enabled=True,
+        bot_token=bot_token,
+        app_token=app_token,
+        user_token=user_token,
+    )
+
+    assert isinstance(slack_bot.bot_token, SensitiveValue)
+    assert isinstance(slack_bot.app_token, SensitiveValue)
+    assert isinstance(slack_bot.user_token, SensitiveValue)
+
+    assert slack_bot.bot_token.get_value(apply_mask=False) == bot_token
+    assert slack_bot.app_token.get_value(apply_mask=False) == app_token
+    assert slack_bot.user_token.get_value(apply_mask=False) == user_token
+
+    # Verify from_model works without error
+    pydantic_bot = SlackBot.from_model(slack_bot)
+    assert pydantic_bot.bot_token  # masked, but not empty
+    assert pydantic_bot.app_token
+
+
+def test_update_slack_bot_returns_sensitive_values(db_session: Session) -> None:
+    slack_bot = insert_slack_bot(
+        db_session=db_session,
+        name=_unique("test-bot-update"),
+        enabled=True,
+        bot_token=_unique("xoxb-update"),
+        app_token=_unique("xapp-update"),
+    )
+
+    new_bot_token = _unique("xoxb-update-new")
+    new_app_token = _unique("xapp-update-new")
+    new_user_token = _unique("xoxp-update-new")
+
+    updated = update_slack_bot(
+        db_session=db_session,
+        slack_bot_id=slack_bot.id,
+        name=_unique("test-bot-updated"),
+        enabled=False,
+        bot_token=new_bot_token,
+        app_token=new_app_token,
+        user_token=new_user_token,
+    )
+
+    assert isinstance(updated.bot_token, SensitiveValue)
+    assert isinstance(updated.app_token, SensitiveValue)
+    assert isinstance(updated.user_token, SensitiveValue)
+
+    assert updated.bot_token.get_value(apply_mask=False) == new_bot_token
+    assert updated.app_token.get_value(apply_mask=False) == new_app_token
+    assert updated.user_token.get_value(apply_mask=False) == new_user_token
+
+    # Verify from_model works without error
+    pydantic_bot = SlackBot.from_model(updated)
+    assert pydantic_bot.bot_token
+    assert pydantic_bot.app_token
+    assert pydantic_bot.user_token is not None

backend/tests/external_dependency_unit/tools/test_oauth_config_crud.py

@@ -21,6 +21,8 @@ from onyx.db.oauth_config import get_tools_by_oauth_config
 from onyx.db.oauth_config import get_user_oauth_token
 from onyx.db.oauth_config import update_oauth_config
 from onyx.db.oauth_config import upsert_user_oauth_token
+from onyx.db.tools import delete_tool__no_commit
+from onyx.db.tools import update_tool
 from tests.external_dependency_unit.conftest import create_test_user
 
 
@@ -146,8 +148,16 @@ class TestOAuthConfigCRUD:
         )
 
         # Secrets should be preserved
-        assert updated_config.client_id == original_client_id
-        assert updated_config.client_secret == original_client_secret
+        assert updated_config.client_id is not None
+        assert original_client_id is not None
+        assert updated_config.client_id.get_value(
+            apply_mask=False
+        ) == original_client_id.get_value(apply_mask=False)
+        assert updated_config.client_secret is not None
+        assert original_client_secret is not None
+        assert updated_config.client_secret.get_value(
+            apply_mask=False
+        ) == original_client_secret.get_value(apply_mask=False)
         # But name should be updated
         assert updated_config.name == new_name
 
@@ -171,9 +181,14 @@ class TestOAuthConfigCRUD:
         )
 
         # client_id should be cleared (empty string)
-        assert updated_config.client_id == ""
+        assert updated_config.client_id is not None
+        assert updated_config.client_id.get_value(apply_mask=False) == ""
         # client_secret should be preserved
-        assert updated_config.client_secret == original_client_secret
+        assert updated_config.client_secret is not None
+        assert original_client_secret is not None
+        assert updated_config.client_secret.get_value(
+            apply_mask=False
+        ) == original_client_secret.get_value(apply_mask=False)
 
     def test_update_oauth_config_clear_client_secret(self, db_session: Session) -> None:
         """Test clearing client_secret while preserving client_id"""
@@ -188,9 +203,14 @@ class TestOAuthConfigCRUD:
         )
 
         # client_secret should be cleared (empty string)
-        assert updated_config.client_secret == ""
+        assert updated_config.client_secret is not None
+        assert updated_config.client_secret.get_value(apply_mask=False) == ""
         # client_id should be preserved
-        assert updated_config.client_id == original_client_id
+        assert updated_config.client_id is not None
+        assert original_client_id is not None
+        assert updated_config.client_id.get_value(
+            apply_mask=False
+        ) == original_client_id.get_value(apply_mask=False)
 
     def test_update_oauth_config_clear_both_secrets(self, db_session: Session) -> None:
         """Test clearing both client_id and client_secret"""
@@ -205,8 +225,10 @@ class TestOAuthConfigCRUD:
         )
 
         # Both should be cleared (empty strings)
-        assert updated_config.client_id == ""
-        assert updated_config.client_secret == ""
+        assert updated_config.client_id is not None
+        assert updated_config.client_id.get_value(apply_mask=False) == ""
+        assert updated_config.client_secret is not None
+        assert updated_config.client_secret.get_value(apply_mask=False) == ""
 
     def test_update_oauth_config_authorization_url(self, db_session: Session) -> None:
         """Test updating authorization_url"""
@@ -273,7 +295,8 @@ class TestOAuthConfigCRUD:
         assert updated_config.token_url == new_token_url
         assert updated_config.scopes == new_scopes
         assert updated_config.additional_params == new_params
-        assert updated_config.client_id == new_client_id
+        assert updated_config.client_id is not None
+        assert updated_config.client_id.get_value(apply_mask=False) == new_client_id
 
     def test_delete_oauth_config(self, db_session: Session) -> None:
         """Test deleting an OAuth configuration"""
@@ -312,6 +335,85 @@ class TestOAuthConfigCRUD:
         # Tool should still exist but oauth_config_id should be NULL
         assert tool.oauth_config_id is None
 
+    def test_update_tool_cleans_up_orphaned_oauth_config(
+        self, db_session: Session
+    ) -> None:
+        """Test that changing a tool's oauth_config_id deletes the old config if no other tool uses it."""
+        old_config = _create_test_oauth_config(db_session)
+        new_config = _create_test_oauth_config(db_session)
+        tool = _create_test_tool_with_oauth(db_session, old_config)
+        old_config_id = old_config.id
+
+        update_tool(
+            tool_id=tool.id,
+            name=None,
+            description=None,
+            openapi_schema=None,
+            custom_headers=None,
+            user_id=None,
+            db_session=db_session,
+            passthrough_auth=None,
+            oauth_config_id=new_config.id,
+        )
+
+        assert tool.oauth_config_id == new_config.id
+        assert get_oauth_config(old_config_id, db_session) is None
+
+    def test_delete_tool_cleans_up_orphaned_oauth_config(
+        self, db_session: Session
+    ) -> None:
+        """Test that deleting the last tool referencing an OAuthConfig also deletes the config."""
+        config = _create_test_oauth_config(db_session)
+        tool = _create_test_tool_with_oauth(db_session, config)
+        config_id = config.id
+
+        delete_tool__no_commit(tool.id, db_session)
+        db_session.commit()
+
+        assert get_oauth_config(config_id, db_session) is None
+
+    def test_update_tool_preserves_shared_oauth_config(
+        self, db_session: Session
+    ) -> None:
+        """Test that updating one tool's oauth_config_id preserves the config when another tool still uses it."""
+        shared_config = _create_test_oauth_config(db_session)
+        new_config = _create_test_oauth_config(db_session)
+        tool_a = _create_test_tool_with_oauth(db_session, shared_config)
+        tool_b = _create_test_tool_with_oauth(db_session, shared_config)
+        shared_config_id = shared_config.id
+
+        # Move tool_a to a new config; tool_b still references shared_config
+        update_tool(
+            tool_id=tool_a.id,
+            name=None,
+            description=None,
+            openapi_schema=None,
+            custom_headers=None,
+            user_id=None,
+            db_session=db_session,
+            passthrough_auth=None,
+            oauth_config_id=new_config.id,
+        )
+
+        assert tool_a.oauth_config_id == new_config.id
+        assert tool_b.oauth_config_id == shared_config_id
+        assert get_oauth_config(shared_config_id, db_session) is not None
+
+    def test_delete_tool_preserves_shared_oauth_config(
+        self, db_session: Session
+    ) -> None:
+        """Test that deleting one tool preserves the config when another tool still uses it."""
+        shared_config = _create_test_oauth_config(db_session)
+        tool_a = _create_test_tool_with_oauth(db_session, shared_config)
+        tool_b = _create_test_tool_with_oauth(db_session, shared_config)
+        shared_config_id = shared_config.id
+
+        delete_tool__no_commit(tool_a.id, db_session)
+        db_session.commit()
+
+        assert tool_b.oauth_config_id == shared_config_id
+        assert get_oauth_config(shared_config_id, db_session) is not None
+
 
 class TestOAuthUserTokenCRUD:
     """Tests for OAuth user token CRUD operations"""
@@ -335,7 +437,8 @@ class TestOAuthUserTokenCRUD:
         assert user_token.id is not None
         assert user_token.oauth_config_id == oauth_config.id
         assert user_token.user_id == user.id
-        assert user_token.token_data == token_data
+        assert user_token.token_data is not None
+        assert user_token.token_data.get_value(apply_mask=False) == token_data
         assert user_token.created_at is not None
         assert user_token.updated_at is not None
 
@@ -365,8 +468,13 @@ class TestOAuthUserTokenCRUD:
 
         # Should be the same token record (updated, not inserted)
         assert updated_token.id == initial_token_id
-        assert updated_token.token_data == updated_token_data
-        assert updated_token.token_data != initial_token_data
+        assert updated_token.token_data is not None
+        assert (
+            updated_token.token_data.get_value(apply_mask=False) == updated_token_data
+        )
+        assert (
+            updated_token.token_data.get_value(apply_mask=False) != initial_token_data
+        )
 
     def test_get_user_oauth_token(self, db_session: Session) -> None:
         """Test retrieving a user's OAuth token"""
@@ -382,7 +490,8 @@ class TestOAuthUserTokenCRUD:
 
         assert retrieved_token is not None
         assert retrieved_token.id == created_token.id
-        assert retrieved_token.token_data == token_data
+        assert retrieved_token.token_data is not None
+        assert retrieved_token.token_data.get_value(apply_mask=False) == token_data
 
     def test_get_user_oauth_token_not_found(self, db_session: Session) -> None:
         """Test retrieving a non-existent user token returns None"""
@@ -438,7 +547,8 @@ class TestOAuthUserTokenCRUD:
         retrieved_token = get_user_oauth_token(oauth_config.id, user.id, db_session)
         assert retrieved_token is not None
         assert retrieved_token.id == updated_token.id
-        assert retrieved_token.token_data == token_data2
+        assert retrieved_token.token_data is not None
+        assert retrieved_token.token_data.get_value(apply_mask=False) == token_data2
 
     def test_cascade_delete_user_tokens_on_config_deletion(
         self, db_session: Session

backend/tests/external_dependency_unit/tools/test_oauth_token_manager.py

@@ -374,8 +374,14 @@ class TestOAuthTokenManagerCodeExchange:
         assert call_args[0][0] == oauth_config.token_url
         assert call_args[1]["data"]["grant_type"] == "authorization_code"
         assert call_args[1]["data"]["code"] == "auth_code_123"
-        assert call_args[1]["data"]["client_id"] == oauth_config.client_id
-        assert call_args[1]["data"]["client_secret"] == oauth_config.client_secret
+        assert oauth_config.client_id is not None
+        assert oauth_config.client_secret is not None
+        assert call_args[1]["data"]["client_id"] == oauth_config.client_id.get_value(
+            apply_mask=False
+        )
+        assert call_args[1]["data"][
+            "client_secret"
+        ] == oauth_config.client_secret.get_value(apply_mask=False)
         assert call_args[1]["data"]["redirect_uri"] == "https://example.com/callback"
 
     @patch("onyx.auth.oauth_token_manager.requests.post")

backend/tests/external_dependency_unit/tools/test_python_tool.py

@@ -950,6 +950,7 @@ from onyx.server.query_and_chat.streaming_models import Packet
 from onyx.server.query_and_chat.streaming_models import PythonToolDelta
 from onyx.server.query_and_chat.streaming_models import PythonToolStart
 from onyx.server.query_and_chat.streaming_models import SectionEnd
+from onyx.server.query_and_chat.streaming_models import ToolCallArgumentDelta
 from onyx.tools.tool_implementations.python.python_tool import PythonTool
 from tests.external_dependency_unit.answer.stream_test_builder import StreamTestBuilder
 from tests.external_dependency_unit.answer.stream_test_utils import create_chat_session
@@ -1294,9 +1295,18 @@ def test_code_interpreter_replay_packets_include_code_and_output(
             ).expect(
                 Packet(
                     placement=create_placement(0),
-                    obj=PythonToolStart(code=code),
+                    obj=ToolCallArgumentDelta(
+                        tool_type="python",
+                        argument_deltas={"code": code},
+                    ),
                 ),
                 forward=2,
+            ).expect(
+                Packet(
+                    placement=create_placement(0),
+                    obj=PythonToolStart(code=code),
+                ),
+                forward=False,
             ).expect(
                 Packet(
                     placement=create_placement(0),

backend/tests/integration/tests/discord_bot/test_discord_bot_db.py

@@ -64,7 +64,8 @@ class TestBotConfigAPI:
         db_session.commit()
 
         assert config is not None
-        assert config.bot_token == "test_token_123"
+        assert config.bot_token is not None
+        assert config.bot_token.get_value(apply_mask=False) == "test_token_123"
 
         # Cleanup
         delete_discord_bot_config(db_session)

backend/tests/unit/ee/conftest.py

@@ -0,0 +1,8 @@
+"""Auto-enable EE mode for all tests under tests/unit/ee/."""
+
+import pytest
+
+
+@pytest.fixture(autouse=True)
+def _enable_ee_for_directory(enable_ee: None) -> None:  # noqa: ARG001
+    """Wraps the shared enable_ee fixture with autouse for this directory."""

backend/tests/unit/ee/onyx/utils/test_encryption.py

@@ -0,0 +1,165 @@
+"""Tests for EE AES-CBC encryption/decryption with explicit key support.
+
+With EE mode enabled (via conftest), fetch_versioned_implementation resolves
+to the EE implementations, so no patching of the MIT layer is needed.
+"""
+
+from unittest.mock import patch
+
+import pytest
+
+from ee.onyx.utils.encryption import _decrypt_bytes
+from ee.onyx.utils.encryption import _encrypt_string
+from ee.onyx.utils.encryption import _get_trimmed_key
+from ee.onyx.utils.encryption import decrypt_bytes_to_string
+from ee.onyx.utils.encryption import encrypt_string_to_bytes
+
+EE_MODULE = "ee.onyx.utils.encryption"
+
+# Keys must be exactly 16, 24, or 32 bytes for AES
+KEY_16 = "a" * 16
+KEY_16_ALT = "b" * 16
+KEY_24 = "d" * 24
+KEY_32 = "c" * 32
+
+
+@pytest.fixture(autouse=True)
+def _clear_key_cache() -> None:
+    _get_trimmed_key.cache_clear()
+
+
+class TestEncryptDecryptRoundTrip:
+    def test_roundtrip_with_env_key(self) -> None:
+        with patch(f"{EE_MODULE}.ENCRYPTION_KEY_SECRET", KEY_16):
+            encrypted = _encrypt_string("hello world")
+            assert encrypted != b"hello world"
+            assert _decrypt_bytes(encrypted) == "hello world"
+
+    def test_roundtrip_with_explicit_key(self) -> None:
+        encrypted = _encrypt_string("secret data", key=KEY_32)
+        assert encrypted != b"secret data"
+        assert _decrypt_bytes(encrypted, key=KEY_32) == "secret data"
+
+    def test_roundtrip_no_key(self) -> None:
+        """Without any key, data is raw-encoded (no encryption)."""
+        with patch(f"{EE_MODULE}.ENCRYPTION_KEY_SECRET", ""):
+            encrypted = _encrypt_string("plain text")
+            assert encrypted == b"plain text"
+            assert _decrypt_bytes(encrypted) == "plain text"
+
+    def test_explicit_key_overrides_env(self) -> None:
+        with patch(f"{EE_MODULE}.ENCRYPTION_KEY_SECRET", KEY_16):
+            encrypted = _encrypt_string("data", key=KEY_16_ALT)
+            with pytest.raises(ValueError):
+                _decrypt_bytes(encrypted, key=KEY_16)
+            assert _decrypt_bytes(encrypted, key=KEY_16_ALT) == "data"
+
+    def test_different_encryptions_produce_different_bytes(self) -> None:
+        """Each encryption uses a random IV, so results differ."""
+        a = _encrypt_string("same", key=KEY_16)
+        b = _encrypt_string("same", key=KEY_16)
+        assert a != b
+
+    def test_roundtrip_empty_string(self) -> None:
+        encrypted = _encrypt_string("", key=KEY_16)
+        assert encrypted != b""
+        assert _decrypt_bytes(encrypted, key=KEY_16) == ""
+
+    def test_roundtrip_unicode(self) -> None:
+        text = "日本語テスト 🔐 émojis"
+        encrypted = _encrypt_string(text, key=KEY_16)
+        assert _decrypt_bytes(encrypted, key=KEY_16) == text
+
+
+class TestDecryptFallbackBehavior:
+    def test_wrong_env_key_falls_back_to_raw_decode(self) -> None:
+        """Default key path: AES fails on non-AES data → fallback to raw decode."""
+        raw = "readable text".encode()
+        with patch(f"{EE_MODULE}.ENCRYPTION_KEY_SECRET", KEY_16):
+            assert _decrypt_bytes(raw) == "readable text"
+
+    def test_explicit_wrong_key_raises(self) -> None:
+        """Explicit key path: AES fails → raises, no fallback."""
+        encrypted = _encrypt_string("secret", key=KEY_16)
+        with pytest.raises(ValueError):
+            _decrypt_bytes(encrypted, key=KEY_16_ALT)
+
+    def test_explicit_none_key_with_no_env(self) -> None:
+        """key=None with empty env → raw decode."""
+        with patch(f"{EE_MODULE}.ENCRYPTION_KEY_SECRET", ""):
+            assert _decrypt_bytes(b"hello", key=None) == "hello"
+
+    def test_explicit_empty_string_key(self) -> None:
+        """key='' means no encryption."""
+        encrypted = _encrypt_string("test", key="")
+        assert encrypted == b"test"
+        assert _decrypt_bytes(encrypted, key="") == "test"
+
+
+class TestKeyValidation:
+    def test_key_too_short_raises(self) -> None:
+        with pytest.raises(RuntimeError, match="too short"):
+            _encrypt_string("data", key="short")
+
+    def test_16_byte_key(self) -> None:
+        encrypted = _encrypt_string("data", key=KEY_16)
+        assert _decrypt_bytes(encrypted, key=KEY_16) == "data"
+
+    def test_24_byte_key(self) -> None:
+        encrypted = _encrypt_string("data", key=KEY_24)
+        assert _decrypt_bytes(encrypted, key=KEY_24) == "data"
+
+    def test_32_byte_key(self) -> None:
+        encrypted = _encrypt_string("data", key=KEY_32)
+        assert _decrypt_bytes(encrypted, key=KEY_32) == "data"
+
+    def test_long_key_truncated_to_32(self) -> None:
+        """Keys longer than 32 bytes are truncated to 32."""
+        long_key = "e" * 64
+        encrypted = _encrypt_string("data", key=long_key)
+        assert _decrypt_bytes(encrypted, key=long_key) == "data"
+
+    def test_20_byte_key_trimmed_to_16(self) -> None:
+        """A 20-byte key is trimmed to the largest valid AES size that fits (16)."""
+        key_20 = "f" * 20
+        encrypted = _encrypt_string("data", key=key_20)
+        assert _decrypt_bytes(encrypted, key=key_20) == "data"
+
+        # Verify it was trimmed to 16 by checking that the first 16 bytes
+        # of the key can also decrypt it
+        key_16_same_prefix = "f" * 16
+        assert _decrypt_bytes(encrypted, key=key_16_same_prefix) == "data"
+
+    def test_25_byte_key_trimmed_to_24(self) -> None:
+        """A 25-byte key is trimmed to the largest valid AES size that fits (24)."""
+        key_25 = "g" * 25
+        encrypted = _encrypt_string("data", key=key_25)
+        assert _decrypt_bytes(encrypted, key=key_25) == "data"
+
+        key_24_same_prefix = "g" * 24
+        assert _decrypt_bytes(encrypted, key=key_24_same_prefix) == "data"
+
+    def test_30_byte_key_trimmed_to_24(self) -> None:
+        """A 30-byte key is trimmed to the largest valid AES size that fits (24)."""
+        key_30 = "h" * 30
+        encrypted = _encrypt_string("data", key=key_30)
+        assert _decrypt_bytes(encrypted, key=key_30) == "data"
+
+        key_24_same_prefix = "h" * 24
+        assert _decrypt_bytes(encrypted, key=key_24_same_prefix) == "data"
+
+
+class TestWrapperFunctions:
+    """Test encrypt_string_to_bytes / decrypt_bytes_to_string pass key through.
+
+    With EE mode enabled, the wrappers resolve to EE implementations automatically.
+    """
+
+    def test_wrapper_passes_key(self) -> None:
+        encrypted = encrypt_string_to_bytes("payload", key=KEY_16)
+        assert decrypt_bytes_to_string(encrypted, key=KEY_16) == "payload"
+
+    def test_wrapper_no_key_uses_env(self) -> None:
+        with patch(f"{EE_MODULE}.ENCRYPTION_KEY_SECRET", KEY_32):
+            encrypted = encrypt_string_to_bytes("payload")
+            assert decrypt_bytes_to_string(encrypted) == "payload"

backend/tests/unit/onyx/chat/test_argument_delta_streaming.py

@@ -0,0 +1,630 @@
+from typing import Any
+from unittest.mock import MagicMock
+from unittest.mock import patch
+
+from onyx.chat.tool_call_args_streaming import maybe_emit_argument_delta
+from onyx.server.query_and_chat.placement import Placement
+from onyx.server.query_and_chat.streaming_models import ToolCallArgumentDelta
+from onyx.utils.jsonriver import Parser
+
+
+def _make_tool_call_delta(
+    index: int = 0,
+    name: str | None = None,
+    arguments: str | None = None,
+    function_is_none: bool = False,
+) -> MagicMock:
+    """Create a mock tool_call_delta matching the LiteLLM streaming shape."""
+    delta = MagicMock()
+    delta.index = index
+    if function_is_none:
+        delta.function = None
+    else:
+        delta.function = MagicMock()
+        delta.function.name = name
+        delta.function.arguments = arguments
+    return delta
+
+
+def _make_placement() -> Placement:
+    return Placement(turn_index=0, tab_index=0)
+
+
+def _mock_tool_class(emit: bool = True) -> MagicMock:
+    cls = MagicMock()
+    cls.should_emit_argument_deltas.return_value = emit
+    return cls
+
+
+def _collect(
+    tc_map: dict[int, dict[str, Any]],
+    delta: MagicMock,
+    placement: Placement | None = None,
+    parsers: dict[int, Parser] | None = None,
+) -> list[Any]:
+    """Run maybe_emit_argument_delta and return the yielded packets."""
+    return list(
+        maybe_emit_argument_delta(
+            tc_map,
+            delta,
+            placement or _make_placement(),
+            parsers if parsers is not None else {},
+        )
+    )
+
+
+def _stream_fragments(
+    fragments: list[str],
+    tc_map: dict[int, dict[str, Any]],
+    placement: Placement | None = None,
+) -> list[str]:
+    """Feed fragments into maybe_emit_argument_delta one by one, returning
+    all emitted content values concatenated per-key as a flat list."""
+    pl = placement or _make_placement()
+    parsers: dict[int, Parser] = {}
+    emitted: list[str] = []
+    for frag in fragments:
+        tc_map[0]["arguments"] += frag
+        delta = _make_tool_call_delta(arguments=frag)
+        for packet in maybe_emit_argument_delta(tc_map, delta, pl, parsers=parsers):
+            obj = packet.obj
+            assert isinstance(obj, ToolCallArgumentDelta)
+            for value in obj.argument_deltas.values():
+                emitted.append(value)
+    return emitted
+
+
+class TestMaybeEmitArgumentDeltaGuards:
+    """Tests for conditions that cause no packet to be emitted."""
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_no_emission_when_tool_does_not_opt_in(
+        self, mock_get_tool: MagicMock
+    ) -> None:
+        """Tools that return False from should_emit_argument_deltas emit nothing."""
+        mock_get_tool.return_value = _mock_tool_class(emit=False)
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": '{"code": "x'}
+        }
+        assert _collect(tc_map, _make_tool_call_delta(arguments="x")) == []
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_no_emission_when_tool_class_unknown(
+        self, mock_get_tool: MagicMock
+    ) -> None:
+        mock_get_tool.return_value = None
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "unknown", "arguments": '{"code": "x'}
+        }
+        assert _collect(tc_map, _make_tool_call_delta(arguments="x")) == []
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_no_emission_when_no_argument_fragment(
+        self, mock_get_tool: MagicMock
+    ) -> None:
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": '{"code": "x'}
+        }
+        assert _collect(tc_map, _make_tool_call_delta(arguments=None)) == []
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_no_emission_when_key_value_not_started(
+        self, mock_get_tool: MagicMock
+    ) -> None:
+        """Key exists in JSON but its string value hasn't begun yet."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": '{"code":'}
+        }
+        assert _collect(tc_map, _make_tool_call_delta(arguments=":")) == []
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_no_emission_before_any_key(self, mock_get_tool: MagicMock) -> None:
+        """Only the opening brace has arrived — no key to stream yet."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": "{"}
+        }
+        assert _collect(tc_map, _make_tool_call_delta(arguments="{")) == []
+
+
+class TestMaybeEmitArgumentDeltaBasic:
+    """Tests for correct packet content and incremental emission."""
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_emits_packet_with_correct_fields(self, mock_get_tool: MagicMock) -> None:
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = ['{"code": "', "print(1)", '"}']
+
+        pl = _make_placement()
+        parsers: dict[int, Parser] = {}
+        all_packets = []
+        for frag in fragments:
+            tc_map[0]["arguments"] += frag
+            packets = _collect(
+                tc_map, _make_tool_call_delta(arguments=frag), pl, parsers
+            )
+            all_packets.extend(packets)
+
+        assert len(all_packets) >= 1
+        # Verify packet structure
+        obj = all_packets[0].obj
+        assert isinstance(obj, ToolCallArgumentDelta)
+        assert obj.tool_type == "python"
+        # All emitted content should reconstruct the value
+        full_code = ""
+        for p in all_packets:
+            assert isinstance(p.obj, ToolCallArgumentDelta)
+            if "code" in p.obj.argument_deltas:
+                full_code += p.obj.argument_deltas["code"]
+        assert full_code == "print(1)"
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_emits_only_new_content_on_subsequent_call(
+        self, mock_get_tool: MagicMock
+    ) -> None:
+        """After a first emission, subsequent calls emit only the diff."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        parsers: dict[int, Parser] = {}
+        pl = _make_placement()
+
+        # First fragment opens the string
+        tc_map[0]["arguments"] = '{"code": "abc'
+        packets_1 = _collect(
+            tc_map, _make_tool_call_delta(arguments='{"code": "abc'), pl, parsers
+        )
+        code_1 = ""
+        for p in packets_1:
+            assert isinstance(p.obj, ToolCallArgumentDelta)
+            code_1 += p.obj.argument_deltas.get("code", "")
+        assert code_1 == "abc"
+
+        # Second fragment appends more
+        tc_map[0]["arguments"] = '{"code": "abcdef'
+        packets_2 = _collect(
+            tc_map, _make_tool_call_delta(arguments="def"), pl, parsers
+        )
+        code_2 = ""
+        for p in packets_2:
+            assert isinstance(p.obj, ToolCallArgumentDelta)
+            code_2 += p.obj.argument_deltas.get("code", "")
+        assert code_2 == "def"
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_handles_multiple_keys_sequentially(self, mock_get_tool: MagicMock) -> None:
+        """When a second key starts, emissions switch to that key."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = [
+            '{"code": "x',
+            '", "output": "hello',
+            '"}',
+        ]
+
+        emitted = _stream_fragments(fragments, tc_map)
+        full = "".join(emitted)
+        assert "x" in full
+        assert "hello" in full
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_delta_spans_key_boundary(self, mock_get_tool: MagicMock) -> None:
+        """A single delta contains the end of one value and the start of the next key."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = [
+            '{"code": "x',
+            'y", "lang": "py',
+            '"}',
+        ]
+
+        emitted = _stream_fragments(fragments, tc_map)
+        full = "".join(emitted)
+        assert "xy" in full
+        assert "py" in full
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_empty_value_emits_nothing(self, mock_get_tool: MagicMock) -> None:
+        """An empty string value has nothing to emit."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        # Opening quote just arrived, value is empty
+        tc_map[0]["arguments"] = '{"code": "'
+        packets = _collect(tc_map, _make_tool_call_delta(arguments='{"code": "'))
+        # No string content yet, so either no packet or empty deltas
+        for p in packets:
+            assert isinstance(p.obj, ToolCallArgumentDelta)
+            assert p.obj.argument_deltas.get("code", "") == ""
+
+
+class TestMaybeEmitArgumentDeltaDecoding:
+    """Tests verifying that JSON escape sequences are properly decoded."""
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_decodes_newlines(self, mock_get_tool: MagicMock) -> None:
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = ['{"code": "line1\\nline2"}']
+
+        emitted = _stream_fragments(fragments, tc_map)
+        assert "".join(emitted) == "line1\nline2"
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_decodes_tabs(self, mock_get_tool: MagicMock) -> None:
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = ['{"code": "\\tindented"}']
+
+        emitted = _stream_fragments(fragments, tc_map)
+        assert "".join(emitted) == "\tindented"
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_decodes_escaped_quotes(self, mock_get_tool: MagicMock) -> None:
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = ['{"code": "say \\"hi\\""}']
+
+        emitted = _stream_fragments(fragments, tc_map)
+        assert "".join(emitted) == 'say "hi"'
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_decodes_escaped_backslashes(self, mock_get_tool: MagicMock) -> None:
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = ['{"code": "path\\\\dir"}']
+
+        emitted = _stream_fragments(fragments, tc_map)
+        assert "".join(emitted) == "path\\dir"
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_decodes_unicode_escape(self, mock_get_tool: MagicMock) -> None:
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = ['{"code": "\\u0041"}']
+
+        emitted = _stream_fragments(fragments, tc_map)
+        assert "".join(emitted) == "A"
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_incomplete_escape_at_end_decoded_on_next_chunk(
+        self, mock_get_tool: MagicMock
+    ) -> None:
+        """A trailing backslash (incomplete escape) is completed in the next chunk."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = ['{"code": "hello\\', 'n"}']
+
+        emitted = _stream_fragments(fragments, tc_map)
+        assert "".join(emitted) == "hello\n"
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_incomplete_unicode_escape_completed_on_next_chunk(
+        self, mock_get_tool: MagicMock
+    ) -> None:
+        """A partial \\uXX sequence is completed in the next chunk."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = ['{"code": "hello\\u00', '41"}']
+
+        emitted = _stream_fragments(fragments, tc_map)
+        assert "".join(emitted) == "helloA"
+
+
+class TestArgumentDeltaStreamingE2E:
+    """Simulates realistic sequences of LLM argument deltas to verify
+    the full pipeline produces correct decoded output."""
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_realistic_python_code_streaming(self, mock_get_tool: MagicMock) -> None:
+        """Streams: {"code": "print('hello')\\nprint('world')"}"""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = [
+            '{"',
+            "code",
+            '": "',
+            "print(",
+            "'hello')",
+            "\\n",
+            "print(",
+            "'world')",
+            '"}',
+        ]
+
+        full = "".join(_stream_fragments(fragments, tc_map))
+        assert full == "print('hello')\nprint('world')"
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_streaming_with_tabs_and_newlines(self, mock_get_tool: MagicMock) -> None:
+        """Streams code with tabs and newlines."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = [
+            '{"code": "',
+            "if True:",
+            "\\n",
+            "\\t",
+            "pass",
+            '"}',
+        ]
+
+        full = "".join(_stream_fragments(fragments, tc_map))
+        assert full == "if True:\n\tpass"
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_split_escape_sequence(self, mock_get_tool: MagicMock) -> None:
+        """An escape sequence split across two fragments (backslash in one,
+        'n' in the next) should still decode correctly."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = [
+            '{"code": "hello',
+            "\\",
+            "n",
+            'world"}',
+        ]
+
+        full = "".join(_stream_fragments(fragments, tc_map))
+        assert full == "hello\nworld"
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_multiple_newlines_and_indentation(self, mock_get_tool: MagicMock) -> None:
+        """Streams a multi-line function with multiple escape sequences."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = [
+            '{"code": "',
+            "def foo():",
+            "\\n",
+            "\\t",
+            "x = 1",
+            "\\n",
+            "\\t",
+            "return x",
+            '"}',
+        ]
+
+        full = "".join(_stream_fragments(fragments, tc_map))
+        assert full == "def foo():\n\tx = 1\n\treturn x"
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_two_keys_streamed_sequentially(self, mock_get_tool: MagicMock) -> None:
+        """Streams code first, then a second key (language) — both decoded."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = [
+            '{"code": "',
+            "x = 1",
+            '", "language": "',
+            "python",
+            '"}',
+        ]
+
+        emitted = _stream_fragments(fragments, tc_map)
+        # Should have emissions for both keys
+        full = "".join(emitted)
+        assert "x = 1" in full
+        assert "python" in full
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_code_containing_dict_literal(self, mock_get_tool: MagicMock) -> None:
+        """Python code like `x = {"key": "val"}` contains JSON-like patterns.
+        The escaped quotes inside the *outer* JSON value should prevent the
+        inner `"key":` from being mistaken for a top-level JSON key."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        # The LLM sends: {"code": "x = {\"key\": \"val\"}"}
+        # The inner quotes are escaped as \" in the JSON value.
+        fragments = [
+            '{"code": "',
+            "x = {",
+            '\\"key\\"',
+            ": ",
+            '\\"val\\"',
+            "}",
+            '"}',
+        ]
+
+        full = "".join(_stream_fragments(fragments, tc_map))
+        assert full == 'x = {"key": "val"}'
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_code_with_colon_in_value(self, mock_get_tool: MagicMock) -> None:
+        """Colons inside the string value should not confuse key detection."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = [
+            '{"code": "',
+            "url = ",
+            '\\"https://example.com\\"',
+            '"}',
+        ]
+
+        full = "".join(_stream_fragments(fragments, tc_map))
+        assert full == 'url = "https://example.com"'
+
+
+class TestMaybeEmitArgumentDeltaEdgeCases:
+    """Edge cases not covered by the standard test classes."""
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_no_emission_when_function_is_none(self, mock_get_tool: MagicMock) -> None:
+        """Some delta chunks have function=None (e.g. role-only deltas)."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": '{"code": "x'}
+        }
+        delta = _make_tool_call_delta(arguments=None, function_is_none=True)
+        assert _collect(tc_map, delta) == []
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_multiple_concurrent_tool_calls(self, mock_get_tool: MagicMock) -> None:
+        """Two tool calls streaming at different indices in parallel."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""},
+            1: {"id": "tc_2", "name": "python", "arguments": ""},
+        }
+
+        parsers: dict[int, Parser] = {}
+        pl = _make_placement()
+
+        # Feed full JSON to index 0
+        tc_map[0]["arguments"] = '{"code": "aaa"}'
+        packets_0 = _collect(
+            tc_map,
+            _make_tool_call_delta(index=0, arguments='{"code": "aaa"}'),
+            pl,
+            parsers,
+        )
+        code_0 = ""
+        for p in packets_0:
+            assert isinstance(p.obj, ToolCallArgumentDelta)
+            code_0 += p.obj.argument_deltas.get("code", "")
+        assert code_0 == "aaa"
+
+        # Feed full JSON to index 1
+        tc_map[1]["arguments"] = '{"code": "bbb"}'
+        packets_1 = _collect(
+            tc_map,
+            _make_tool_call_delta(index=1, arguments='{"code": "bbb"}'),
+            pl,
+            parsers,
+        )
+        code_1 = ""
+        for p in packets_1:
+            assert isinstance(p.obj, ToolCallArgumentDelta)
+            code_1 += p.obj.argument_deltas.get("code", "")
+        assert code_1 == "bbb"
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_delta_with_four_arguments(self, mock_get_tool: MagicMock) -> None:
+        """A single delta contains four complete key-value pairs."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        full = '{"a": "one", "b": "two", "c": "three", "d": "four"}'
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        tc_map[0]["arguments"] = full
+        parsers: dict[int, Parser] = {}
+        packets = _collect(
+            tc_map, _make_tool_call_delta(arguments=full), parsers=parsers
+        )
+
+        # Collect all argument deltas across packets
+        all_deltas: dict[str, str] = {}
+        for p in packets:
+            assert isinstance(p.obj, ToolCallArgumentDelta)
+            for k, v in p.obj.argument_deltas.items():
+                all_deltas[k] = all_deltas.get(k, "") + v
+
+        assert all_deltas == {
+            "a": "one",
+            "b": "two",
+            "c": "three",
+            "d": "four",
+        }
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_delta_on_second_arg_after_first_complete(
+        self, mock_get_tool: MagicMock
+    ) -> None:
+        """First argument is fully complete; delta only adds to the second."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+
+        fragments = [
+            '{"code": "print(1)", "lang": "py',
+            '"}',
+        ]
+
+        emitted = _stream_fragments(fragments, tc_map)
+        full = "".join(emitted)
+        assert "print(1)" in full
+        assert "py" in full
+
+    @patch("onyx.chat.tool_call_args_streaming._get_tool_class")
+    def test_non_string_values_skipped(self, mock_get_tool: MagicMock) -> None:
+        """Non-string values (numbers, booleans, null) are skipped — they are
+        available in the final tool-call kickoff packet. String arguments
+        following them are still emitted."""
+        mock_get_tool.return_value = _mock_tool_class()
+
+        tc_map: dict[int, dict[str, Any]] = {
+            0: {"id": "tc_1", "name": "python", "arguments": ""}
+        }
+        fragments = ['{"timeout": 30, "code": "hello"}']
+
+        emitted = _stream_fragments(fragments, tc_map)
+        full = "".join(emitted)
+        assert full == "hello"

backend/tests/unit/onyx/chat/test_llm_loop.py

@@ -2,7 +2,6 @@
 
 import pytest
 
-from onyx.chat.llm_loop import _should_keep_bedrock_tool_definitions
 from onyx.chat.llm_loop import _try_fallback_tool_extraction
 from onyx.chat.llm_loop import construct_message_history
 from onyx.chat.models import ChatLoadedFile
@@ -14,22 +13,11 @@ from onyx.chat.models import LlmStepResult
 from onyx.chat.models import ToolCallSimple
 from onyx.configs.constants import MessageType
 from onyx.file_store.models import ChatFileType
-from onyx.llm.constants import LlmProviderNames
 from onyx.llm.interfaces import ToolChoiceOptions
 from onyx.server.query_and_chat.placement import Placement
 from onyx.tools.models import ToolCallKickoff
 
 
-class _StubConfig:
-    def __init__(self, model_provider: str) -> None:
-        self.model_provider = model_provider
-
-
-class _StubLLM:
-    def __init__(self, model_provider: str) -> None:
-        self.config = _StubConfig(model_provider=model_provider)
-
-
 def create_message(
     content: str, message_type: MessageType, token_count: int | None = None
 ) -> ChatMessageSimple:
@@ -946,37 +934,6 @@ class TestForgottenFileMetadata:
             assert "moby_dick.txt" in forgotten.message
 
 
-class TestBedrockToolConfigGuard:
-    def test_bedrock_with_tool_history_keeps_tool_definitions(self) -> None:
-        llm = _StubLLM(LlmProviderNames.BEDROCK)
-        history = [
-            create_message("Question", MessageType.USER, 5),
-            create_assistant_with_tool_call("tc_1", "search", 5),
-            create_tool_response("tc_1", "Tool output", 5),
-        ]
-
-        assert _should_keep_bedrock_tool_definitions(llm, history) is True
-
-    def test_bedrock_without_tool_history_does_not_keep_tool_definitions(self) -> None:
-        llm = _StubLLM(LlmProviderNames.BEDROCK)
-        history = [
-            create_message("Question", MessageType.USER, 5),
-            create_message("Answer", MessageType.ASSISTANT, 5),
-        ]
-
-        assert _should_keep_bedrock_tool_definitions(llm, history) is False
-
-    def test_non_bedrock_with_tool_history_does_not_keep_tool_definitions(self) -> None:
-        llm = _StubLLM(LlmProviderNames.OPENAI)
-        history = [
-            create_message("Question", MessageType.USER, 5),
-            create_assistant_with_tool_call("tc_1", "search", 5),
-            create_tool_response("tc_1", "Tool output", 5),
-        ]
-
-        assert _should_keep_bedrock_tool_definitions(llm, history) is False
-
-
 class TestFallbackToolExtraction:
     def _tool_defs(self) -> list[dict]:
         return [