fix(license): exclude service account users from seat count (#10053)

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

.devcontainer/Dockerfile

@@ -0,0 +1,64 @@
+FROM ubuntu:26.04@sha256:cc925e589b7543b910fea57a240468940003fbfc0515245a495dd0ad8fe7cef1
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+  curl \
+  fd-find \
+  fzf \
+  git \
+  jq \
+  less \
+  make \
+  neovim \
+  openssh-client \
+  python3-venv \
+  ripgrep \
+  sudo \
+  ca-certificates \
+  iptables \
+  ipset \
+  iproute2 \
+  dnsutils \
+  unzip \
+  wget \
+  zsh \
+  && curl -fsSL https://deb.nodesource.com/setup_20.x | bash - \
+  && apt-get install -y nodejs \
+  && install -m 0755 -d /etc/apt/keyrings \
+  && curl -fsSL https://download.docker.com/linux/ubuntu/gpg -o /etc/apt/keyrings/docker.asc \
+  && echo "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.asc] https://download.docker.com/linux/ubuntu $(. /etc/os-release && echo "$VERSION_CODENAME") stable" > /etc/apt/sources.list.d/docker.list \
+  && curl -fsSL https://cli.github.com/packages/githubcli-archive-keyring.gpg -o /etc/apt/keyrings/githubcli-archive-keyring.gpg \
+  && chmod go+r /etc/apt/keyrings/githubcli-archive-keyring.gpg \
+  && echo "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/githubcli-archive-keyring.gpg] https://cli.github.com/packages stable main" > /etc/apt/sources.list.d/github-cli.list \
+  && apt-get update \
+  && apt-get install -y --no-install-recommends docker-ce-cli docker-compose-plugin gh \
+  && apt-get clean && rm -rf /var/lib/apt/lists/*
+
+# fd-find installs as fdfind on Debian/Ubuntu — symlink to fd
+RUN ln -sf "$(which fdfind)" /usr/local/bin/fd
+
+# Install uv (Python package manager)
+COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /usr/local/bin/
+
+# Create non-root dev user with passwordless sudo
+RUN useradd -m -s /bin/zsh dev && \
+  echo "dev ALL=(ALL) NOPASSWD:ALL" > /etc/sudoers.d/dev && \
+  chmod 0440 /etc/sudoers.d/dev
+
+ENV DEVCONTAINER=true
+
+RUN mkdir -p /workspace && \
+  chown -R dev:dev /workspace
+
+WORKDIR /workspace
+
+# Install Claude Code
+ARG CLAUDE_CODE_VERSION=latest
+RUN npm install -g @anthropic-ai/claude-code@${CLAUDE_CODE_VERSION}
+
+# Configure zsh — source the repo-local zshrc so shell customization
+# doesn't require an image rebuild.
+RUN chsh -s /bin/zsh root && \
+  for rc in /root/.zshrc /home/dev/.zshrc; do \
+    echo '[ -f /workspace/.devcontainer/zshrc ] && . /workspace/.devcontainer/zshrc' >> "$rc"; \
+  done && \
+  chown dev:dev /home/dev/.zshrc

.devcontainer/README.md

@@ -0,0 +1,99 @@
+# Onyx Dev Container
+
+A containerized development environment for working on Onyx.
+
+## What's included
+
+- Ubuntu 26.04 base image
+- Node.js 20, uv, Claude Code
+- Docker CLI, GitHub CLI (`gh`)
+- Neovim, ripgrep, fd, fzf, jq, make, wget, unzip
+- Zsh as default shell (sources host `~/.zshrc` if available)
+- Python venv auto-activation
+- Network firewall (default-deny, whitelists npm, GitHub, Anthropic APIs, Sentry, and VS Code update servers)
+
+## Usage
+
+### CLI (`ods dev`)
+
+The [`ods` devtools CLI](../tools/ods/README.md) provides workspace-aware wrappers
+for all devcontainer operations (also available as `ods dc`):
+
+```bash
+# Start the container
+ods dev up
+
+# Open a shell
+ods dev into
+
+# Run a command
+ods dev exec npm test
+
+# Stop the container
+ods dev stop
+```
+
+## Restarting the container
+
+```bash
+# Restart the container
+ods dev restart
+
+# Pull the latest published image and recreate
+ods dev rebuild
+```
+
+## Image
+
+The devcontainer uses a prebuilt image published to `onyxdotapp/onyx-devcontainer`.
+The tag is pinned in `devcontainer.json` — no local build is required.
+
+To build the image locally (e.g. while iterating on the Dockerfile):
+
+```bash
+docker buildx bake devcontainer
+```
+
+The `devcontainer` target is defined in `docker-bake.hcl` at the repo root.
+
+## User & permissions
+
+The container runs as the `dev` user by default (`remoteUser` in devcontainer.json).
+An init script (`init-dev-user.sh`) runs at container start to ensure the active
+user has read/write access to the bind-mounted workspace:
+
+- **Standard Docker** — `dev`'s UID/GID is remapped to match the workspace owner,
+  so file permissions work seamlessly.
+- **Rootless Docker** — The workspace appears as root-owned (UID 0) inside the
+  container due to user-namespace mapping. `ods dev up` auto-detects rootless Docker
+  and sets `DEVCONTAINER_REMOTE_USER=root` so the container runs as root — which
+  maps back to your host user via the user namespace. New files are owned by your
+  host UID and no ACL workarounds are needed.
+
+  To override the auto-detection, set `DEVCONTAINER_REMOTE_USER` before running
+  `ods dev up`.
+
+## Docker socket
+
+The container mounts the host's Docker socket so you can run `docker` commands
+from inside. `ods dev` auto-detects the socket path and sets `DOCKER_SOCK`:
+
+| Environment             | Socket path                    |
+| ----------------------- | ------------------------------ |
+| Linux (rootless Docker) | `$XDG_RUNTIME_DIR/docker.sock` |
+| macOS (Docker Desktop)  | `~/.docker/run/docker.sock`    |
+| Linux (standard Docker) | `/var/run/docker.sock`         |
+
+To override, set `DOCKER_SOCK` before running `ods dev up`.
+
+## Firewall
+
+The container starts with a default-deny firewall (`init-firewall.sh`) that only allows outbound traffic to:
+
+- npm registry
+- GitHub
+- Anthropic API
+- Sentry
+- VS Code update servers
+
+This requires the `NET_ADMIN` and `NET_RAW` capabilities, which are added via `runArgs` in `devcontainer.json`.

.devcontainer/devcontainer.json

@@ -0,0 +1,24 @@
+{
+  "name": "Onyx Dev Sandbox",
+  "image": "onyxdotapp/onyx-devcontainer@sha256:12184169c5bcc9cca0388286d5ffe504b569bc9c37bfa631b76ee8eee2064055",
+  "runArgs": ["--cap-add=NET_ADMIN", "--cap-add=NET_RAW"],
+  "mounts": [
+    "source=${localEnv:DOCKER_SOCK},target=/var/run/docker.sock,type=bind",
+    "source=${localEnv:HOME}/.claude,target=/home/dev/.claude,type=bind",
+    "source=${localEnv:HOME}/.claude.json,target=/home/dev/.claude.json,type=bind",
+    "source=${localEnv:HOME}/.zshrc,target=/home/dev/.zshrc.host,type=bind,readonly",
+    "source=${localEnv:HOME}/.gitconfig,target=/home/dev/.gitconfig,type=bind,readonly",
+    "source=${localEnv:HOME}/.config/nvim,target=/home/dev/.config/nvim,type=bind,readonly",
+    "source=onyx-devcontainer-cache,target=/home/dev/.cache,type=volume",
+    "source=onyx-devcontainer-local,target=/home/dev/.local,type=volume"
+  ],
+  "containerEnv": {
+    "SSH_AUTH_SOCK": "/tmp/ssh-agent.sock"
+  },
+  "remoteUser": "${localEnv:DEVCONTAINER_REMOTE_USER:dev}",
+  "updateRemoteUserUID": false,
+  "workspaceMount": "source=${localWorkspaceFolder},target=/workspace,type=bind,consistency=delegated",
+  "workspaceFolder": "/workspace",
+  "postStartCommand": "sudo bash /workspace/.devcontainer/init-dev-user.sh && sudo bash /workspace/.devcontainer/init-firewall.sh",
+  "waitFor": "postStartCommand"
+}

.devcontainer/init-dev-user.sh

@@ -0,0 +1,107 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# Remap the dev user's UID/GID to match the workspace owner so that
+# bind-mounted files are accessible without running as root.
+#
+# Standard Docker:   Workspace is owned by the host user's UID (e.g. 1000).
+#                    We remap dev to that UID -- fast and seamless.
+#
+# Rootless Docker:   Workspace appears as root-owned (UID 0) inside the
+#                    container due to user-namespace mapping.  Requires
+#                    DEVCONTAINER_REMOTE_USER=root (set automatically by
+#                    ods dev up).  Container root IS the host user, so
+#                    bind-mounts and named volumes are symlinked into /root.
+
+WORKSPACE=/workspace
+TARGET_USER=dev
+REMOTE_USER="${SUDO_USER:-$TARGET_USER}"
+
+WS_UID=$(stat -c '%u' "$WORKSPACE")
+WS_GID=$(stat -c '%g' "$WORKSPACE")
+DEV_UID=$(id -u "$TARGET_USER")
+DEV_GID=$(id -g "$TARGET_USER")
+
+# devcontainer.json bind-mounts and named volumes target /home/dev regardless
+# of remoteUser.  When running as root ($HOME=/root), Phase 1 bridges the gap
+# with symlinks from ACTIVE_HOME → MOUNT_HOME.
+MOUNT_HOME=/home/"$TARGET_USER"
+
+if [ "$REMOTE_USER" = "root" ]; then
+    ACTIVE_HOME="/root"
+else
+    ACTIVE_HOME="$MOUNT_HOME"
+fi
+
+# ── Phase 1: home directory setup ───────────────────────────────────
+
+# ~/.local and ~/.cache are named Docker volumes mounted under MOUNT_HOME.
+mkdir -p "$MOUNT_HOME"/.local/state "$MOUNT_HOME"/.local/share
+
+# When running as root, symlink bind-mounts and named volumes into /root
+# so that $HOME-relative tools (Claude Code, git, etc.) find them.
+if [ "$ACTIVE_HOME" != "$MOUNT_HOME" ]; then
+    for item in .claude .cache .local; do
+        [ -d "$MOUNT_HOME/$item" ] || continue
+        if [ -e "$ACTIVE_HOME/$item" ] && [ ! -L "$ACTIVE_HOME/$item" ]; then
+            echo "warning: replacing $ACTIVE_HOME/$item with symlink to $MOUNT_HOME/$item" >&2
+            rm -rf "$ACTIVE_HOME/$item"
+        fi
+        ln -sfn "$MOUNT_HOME/$item" "$ACTIVE_HOME/$item"
+    done
+    # Symlink files (not directories).
+    for file in .claude.json .gitconfig .zshrc.host; do
+        [ -f "$MOUNT_HOME/$file" ] && ln -sf "$MOUNT_HOME/$file" "$ACTIVE_HOME/$file"
+    done
+
+    # Nested mount: .config/nvim
+    if [ -d "$MOUNT_HOME/.config/nvim" ]; then
+        mkdir -p "$ACTIVE_HOME/.config"
+        if [ -e "$ACTIVE_HOME/.config/nvim" ] && [ ! -L "$ACTIVE_HOME/.config/nvim" ]; then
+            echo "warning: replacing $ACTIVE_HOME/.config/nvim with symlink" >&2
+            rm -rf "$ACTIVE_HOME/.config/nvim"
+        fi
+        ln -sfn "$MOUNT_HOME/.config/nvim" "$ACTIVE_HOME/.config/nvim"
+    fi
+fi
+
+# ── Phase 2: workspace access ───────────────────────────────────────
+
+# Root always has workspace access; Phase 1 handled home setup.
+if [ "$REMOTE_USER" = "root" ]; then
+    exit 0
+fi
+
+# Already matching -- nothing to do.
+if [ "$WS_UID" = "$DEV_UID" ] && [ "$WS_GID" = "$DEV_GID" ]; then
+    exit 0
+fi
+
+if [ "$WS_UID" != "0" ]; then
+    # ── Standard Docker ──────────────────────────────────────────────
+    # Workspace is owned by a non-root UID (the host user).
+    # Remap dev's UID/GID to match.
+    if [ "$DEV_GID" != "$WS_GID" ]; then
+        if ! groupmod -g "$WS_GID" "$TARGET_USER" 2>&1; then
+            echo "warning: failed to remap $TARGET_USER GID to $WS_GID" >&2
+        fi
+    fi
+    if [ "$DEV_UID" != "$WS_UID" ]; then
+        if ! usermod -u "$WS_UID" -g "$WS_GID" "$TARGET_USER" 2>&1; then
+            echo "warning: failed to remap $TARGET_USER UID to $WS_UID" >&2
+        fi
+    fi
+    if ! chown -R "$TARGET_USER":"$TARGET_USER" "$MOUNT_HOME" 2>&1; then
+        echo "warning: failed to chown $MOUNT_HOME" >&2
+    fi
+else
+    # ── Rootless Docker ──────────────────────────────────────────────
+    # Workspace is root-owned (UID 0) due to user-namespace mapping.
+    # The supported path is remoteUser=root (set DEVCONTAINER_REMOTE_USER=root),
+    # which is handled above.  If we reach here, the user is running as dev
+    # under rootless Docker without the override.
+    echo "error: rootless Docker detected but remoteUser is not root." >&2
+    echo "       Set DEVCONTAINER_REMOTE_USER=root before starting the container," >&2
+    echo "       or use 'ods dev up' which sets it automatically." >&2
+    exit 1
+fi

.devcontainer/init-firewall.sh

@@ -0,0 +1,104 @@
+#!/usr/bin/env bash
+
+set -euo pipefail
+
+echo "Setting up firewall..."
+
+# Preserve docker dns resolution
+DOCKER_DNS_RULES=$(iptables-save | grep -E "^-A.*-d 127.0.0.11/32" || true)
+
+# Flush all rules
+iptables -t nat -F
+iptables -t nat -X
+iptables -t mangle -F
+iptables -t mangle -X
+iptables -F
+iptables -X
+
+# Restore docker dns rules
+if [ -n "$DOCKER_DNS_RULES" ]; then
+    echo "$DOCKER_DNS_RULES" | iptables-restore -n
+fi
+
+# Create ipset for allowed destinations
+ipset create allowed-domains hash:net || true
+ipset flush allowed-domains
+
+# Fetch GitHub IP ranges (IPv4 only -- ipset hash:net and iptables are IPv4)
+GITHUB_IPS=$(curl -s https://api.github.com/meta | jq -r '.api[]' 2>/dev/null | grep -v ':' || echo "")
+for ip in $GITHUB_IPS; do
+    if ! ipset add allowed-domains "$ip" -exist 2>&1; then
+        echo "warning: failed to add GitHub IP $ip to allowlist" >&2
+    fi
+done
+
+# Resolve allowed domains
+ALLOWED_DOMAINS=(
+    "registry.npmjs.org"
+    "api.anthropic.com"
+    "api-staging.anthropic.com"
+    "files.anthropic.com"
+    "sentry.io"
+    "update.code.visualstudio.com"
+    "pypi.org"
+    "files.pythonhosted.org"
+    "go.dev"
+    "storage.googleapis.com"
+    "static.rust-lang.org"
+)
+
+for domain in "${ALLOWED_DOMAINS[@]}"; do
+    IPS=$(getent ahosts "$domain" 2>/dev/null | awk '{print $1}' | grep -v ':' | sort -u || echo "")
+    for ip in $IPS; do
+        if ! ipset add allowed-domains "$ip/32" -exist 2>&1; then
+            echo "warning: failed to add $domain ($ip) to allowlist" >&2
+        fi
+    done
+done
+
+# Detect host network
+if [[ "${DOCKER_HOST:-}" == "unix://"* ]]; then
+    DOCKER_GATEWAY=$(ip -4 route show | grep "^default" | awk '{print $3}')
+    if ! ipset add allowed-domains "$DOCKER_GATEWAY/32" -exist 2>&1; then
+        echo "warning: failed to add Docker gateway $DOCKER_GATEWAY to allowlist" >&2
+    fi
+fi
+
+# Set default policies to DROP
+iptables -P FORWARD DROP
+iptables -P INPUT DROP
+iptables -P OUTPUT DROP
+
+# Allow established connections
+iptables -A INPUT -m conntrack --ctstate ESTABLISHED,RELATED -j ACCEPT
+iptables -A OUTPUT -m conntrack --ctstate ESTABLISHED,RELATED -j ACCEPT
+
+# Allow loopback
+iptables -A INPUT -i lo -j ACCEPT
+iptables -A OUTPUT -o lo -j ACCEPT
+
+# Allow DNS
+iptables -A OUTPUT -p udp --dport 53 -j ACCEPT
+iptables -A OUTPUT -p tcp --dport 53 -j ACCEPT
+
+# Allow outbound to allowed destinations
+iptables -A OUTPUT -m set --match-set allowed-domains dst -j ACCEPT
+
+# Reject unauthorized outbound
+iptables -A OUTPUT -j REJECT --reject-with icmp-host-unreachable
+
+# Validate firewall configuration
+echo "Validating firewall configuration..."
+
+BLOCKED_SITES=("example.com" "google.com" "facebook.com")
+for site in "${BLOCKED_SITES[@]}"; do
+    if timeout 2 ping -c 1 "$site" &>/dev/null; then
+        echo "Warning: $site is still reachable"
+    fi
+done
+
+if ! timeout 5 curl -s https://api.github.com/meta > /dev/null; then
+    echo "Warning: GitHub API is not accessible"
+fi
+
+echo "Firewall setup complete"

.devcontainer/zshrc

@@ -0,0 +1,10 @@
+# Devcontainer zshrc — sourced automatically for both root and dev users.
+# Edit this file to customize the shell without rebuilding the image.
+
+# Auto-activate Python venv
+if [ -f /workspace/.venv/bin/activate ]; then
+  . /workspace/.venv/bin/activate
+fi
+
+# Source host zshrc if bind-mounted
+[ -f ~/.zshrc.host ] && . ~/.zshrc.host

backend/ee/onyx/db/license.py

@@ -13,6 +13,7 @@ from ee.onyx.server.license.models import LicenseSource
 from onyx.auth.schemas import UserRole
 from onyx.cache.factory import get_cache_backend
 from onyx.configs.constants import ANONYMOUS_USER_EMAIL
+from onyx.db.enums import AccountType
 from onyx.db.models import License
 from onyx.db.models import User
 from onyx.utils.logger import setup_logger
@@ -107,12 +108,13 @@ def get_used_seats(tenant_id: str | None = None) -> int:
     Get current seat usage directly from database.
 
     For multi-tenant: counts users in UserTenantMapping for this tenant.
-    For self-hosted: counts all active users (excludes EXT_PERM_USER role
-    and the anonymous system user).
+    For self-hosted: counts all active users.
 
-    TODO: Exclude API key dummy users from seat counting. API keys create
-    users with emails like `__DANSWER_API_KEY_*` that should not count toward
-    seat limits. See: https://linear.app/onyx-app/issue/ENG-3518
+    Only human accounts count toward seat limits.
+    SERVICE_ACCOUNT (API key dummy users), EXT_PERM_USER, and the
+    anonymous system user are excluded. BOT (Slack users) ARE counted
+    because they represent real humans and get upgraded to STANDARD
+    when they log in via web.
     """
     if MULTI_TENANT:
         from ee.onyx.server.tenants.user_mapping import get_tenant_count
@@ -129,6 +131,7 @@ def get_used_seats(tenant_id: str | None = None) -> int:
                     User.is_active == True,  # type: ignore  # noqa: E712
                     User.role != UserRole.EXT_PERM_USER,
                     User.email != ANONYMOUS_USER_EMAIL,  # type: ignore
+                    User.account_type != AccountType.SERVICE_ACCOUNT,
                 )
             )
             return result.scalar() or 0

backend/onyx/connectors/google_drive/connector.py

@@ -42,6 +42,9 @@ from onyx.connectors.google_drive.file_retrieval import (
     get_all_files_in_my_drive_and_shared,
 )
 from onyx.connectors.google_drive.file_retrieval import get_external_access_for_folder
+from onyx.connectors.google_drive.file_retrieval import (
+    get_files_by_web_view_links_batch,
+)
 from onyx.connectors.google_drive.file_retrieval import get_files_in_shared_drive
 from onyx.connectors.google_drive.file_retrieval import get_folder_metadata
 from onyx.connectors.google_drive.file_retrieval import get_root_folder_id
@@ -70,11 +73,13 @@ from onyx.connectors.interfaces import CheckpointedConnectorWithPermSync
 from onyx.connectors.interfaces import CheckpointOutput
 from onyx.connectors.interfaces import GenerateSlimDocumentOutput
 from onyx.connectors.interfaces import NormalizationResult
+from onyx.connectors.interfaces import Resolver
 from onyx.connectors.interfaces import SecondsSinceUnixEpoch
 from onyx.connectors.interfaces import SlimConnectorWithPermSync
 from onyx.connectors.models import ConnectorFailure
 from onyx.connectors.models import ConnectorMissingCredentialError
 from onyx.connectors.models import Document
+from onyx.connectors.models import DocumentFailure
 from onyx.connectors.models import EntityFailure
 from onyx.connectors.models import HierarchyNode
 from onyx.connectors.models import SlimDocument
@@ -202,7 +207,9 @@ class DriveIdStatus(Enum):
 
 
 class GoogleDriveConnector(
-    SlimConnectorWithPermSync, CheckpointedConnectorWithPermSync[GoogleDriveCheckpoint]
+    SlimConnectorWithPermSync,
+    CheckpointedConnectorWithPermSync[GoogleDriveCheckpoint],
+    Resolver,
 ):
     def __init__(
         self,
@@ -1665,6 +1672,82 @@ class GoogleDriveConnector(
             start, end, checkpoint, include_permissions=True
         )
 
+    @override
+    def resolve_errors(
+        self,
+        errors: list[ConnectorFailure],
+        include_permissions: bool = False,
+    ) -> Generator[Document | ConnectorFailure | HierarchyNode, None, None]:
+        if self._creds is None or self._primary_admin_email is None:
+            raise RuntimeError(
+                "Credentials missing, should not call this method before calling load_credentials"
+            )
+
+        logger.info(f"Resolving {len(errors)} errors")
+        doc_ids = [
+            failure.failed_document.document_id
+            for failure in errors
+            if failure.failed_document
+        ]
+        service = get_drive_service(self.creds, self.primary_admin_email)
+        field_type = (
+            DriveFileFieldType.WITH_PERMISSIONS
+            if include_permissions or self.exclude_domain_link_only
+            else DriveFileFieldType.STANDARD
+        )
+        batch_result = get_files_by_web_view_links_batch(service, doc_ids, field_type)
+
+        for doc_id, error in batch_result.errors.items():
+            yield ConnectorFailure(
+                failed_document=DocumentFailure(
+                    document_id=doc_id,
+                    document_link=doc_id,
+                ),
+                failure_message=f"Failed to retrieve file during error resolution: {error}",
+                exception=error,
+            )
+
+        permission_sync_context = (
+            PermissionSyncContext(
+                primary_admin_email=self.primary_admin_email,
+                google_domain=self.google_domain,
+            )
+            if include_permissions
+            else None
+        )
+
+        retrieved_files = [
+            RetrievedDriveFile(
+                drive_file=file,
+                user_email=self.primary_admin_email,
+                completion_stage=DriveRetrievalStage.DONE,
+            )
+            for file in batch_result.files.values()
+        ]
+
+        yield from self._get_new_ancestors_for_files(
+            files=retrieved_files,
+            seen_hierarchy_node_raw_ids=ThreadSafeSet(),
+            fully_walked_hierarchy_node_raw_ids=ThreadSafeSet(),
+            permission_sync_context=permission_sync_context,
+            add_prefix=True,
+        )
+
+        func_with_args = [
+            (
+                self._convert_retrieved_file_to_document,
+                (rf, permission_sync_context),
+            )
+            for rf in retrieved_files
+        ]
+        results = cast(
+            list[Document | ConnectorFailure | None],
+            run_functions_tuples_in_parallel(func_with_args, max_workers=8),
+        )
+        for result in results:
+            if result is not None:
+                yield result
+
     def _extract_slim_docs_from_google_drive(
         self,
         checkpoint: GoogleDriveCheckpoint,

backend/onyx/connectors/google_drive/file_retrieval.py

@@ -9,6 +9,7 @@ from urllib.parse import urlparse
 
 from googleapiclient.discovery import Resource  # type: ignore
 from googleapiclient.errors import HttpError  # type: ignore
+from googleapiclient.http import BatchHttpRequest  # type: ignore
 
 from onyx.access.models import ExternalAccess
 from onyx.connectors.google_drive.constants import DRIVE_FOLDER_TYPE
@@ -60,6 +61,8 @@ SLIM_FILE_FIELDS = (
 )
 FOLDER_FIELDS = "nextPageToken, files(id, name, permissions, modifiedTime, webViewLink, shortcutDetails)"
 
+MAX_BATCH_SIZE = 100
+
 HIERARCHY_FIELDS = "id, name, parents, webViewLink, mimeType, driveId"
 
 HIERARCHY_FIELDS_WITH_PERMISSIONS = (
@@ -216,7 +219,7 @@ def get_external_access_for_folder(
 
 
 def _get_fields_for_file_type(field_type: DriveFileFieldType) -> str:
-    """Get the appropriate fields string based on the field type enum"""
+    """Get the appropriate fields string for files().list() based on the field type enum."""
     if field_type == DriveFileFieldType.SLIM:
         return SLIM_FILE_FIELDS
     elif field_type == DriveFileFieldType.WITH_PERMISSIONS:
@@ -225,6 +228,25 @@ def _get_fields_for_file_type(field_type: DriveFileFieldType) -> str:
         return FILE_FIELDS
 
 
+def _extract_single_file_fields(list_fields: str) -> str:
+    """Convert a files().list() fields string to one suitable for files().get().
+
+    List fields look like "nextPageToken, files(field1, field2, ...)"
+    Single-file fields should be just "field1, field2, ..."
+    """
+    start = list_fields.find("files(")
+    if start == -1:
+        return list_fields
+    inner_start = start + len("files(")
+    inner_end = list_fields.rfind(")")
+    return list_fields[inner_start:inner_end]
+
+
+def _get_single_file_fields(field_type: DriveFileFieldType) -> str:
+    """Get the appropriate fields string for files().get() based on the field type enum."""
+    return _extract_single_file_fields(_get_fields_for_file_type(field_type))
+
+
 def _get_files_in_parent(
     service: Resource,
     parent_id: str,
@@ -536,3 +558,74 @@ def get_file_by_web_view_link(
         )
         .execute()
     )
+
+
+class BatchRetrievalResult:
+    """Result of a batch file retrieval, separating successes from errors."""
+
+    def __init__(self) -> None:
+        self.files: dict[str, GoogleDriveFileType] = {}
+        self.errors: dict[str, Exception] = {}
+
+
+def get_files_by_web_view_links_batch(
+    service: GoogleDriveService,
+    web_view_links: list[str],
+    field_type: DriveFileFieldType,
+) -> BatchRetrievalResult:
+    """Retrieve multiple Google Drive files by webViewLink using the batch API.
+
+    Returns a BatchRetrievalResult containing successful file retrievals
+    and errors for any files that could not be fetched.
+    Automatically splits into chunks of MAX_BATCH_SIZE.
+    """
+    fields = _get_single_file_fields(field_type)
+    if len(web_view_links) <= MAX_BATCH_SIZE:
+        return _get_files_by_web_view_links_batch(service, web_view_links, fields)
+
+    combined = BatchRetrievalResult()
+    for i in range(0, len(web_view_links), MAX_BATCH_SIZE):
+        chunk = web_view_links[i : i + MAX_BATCH_SIZE]
+        chunk_result = _get_files_by_web_view_links_batch(service, chunk, fields)
+        combined.files.update(chunk_result.files)
+        combined.errors.update(chunk_result.errors)
+    return combined
+
+
+def _get_files_by_web_view_links_batch(
+    service: GoogleDriveService,
+    web_view_links: list[str],
+    fields: str,
+) -> BatchRetrievalResult:
+    """Single-batch implementation."""
+
+    result = BatchRetrievalResult()
+
+    def callback(
+        request_id: str,
+        response: GoogleDriveFileType,
+        exception: Exception | None,
+    ) -> None:
+        if exception:
+            logger.warning(f"Error retrieving file {request_id}: {exception}")
+            result.errors[request_id] = exception
+        else:
+            result.files[request_id] = response
+
+    batch = cast(BatchHttpRequest, service.new_batch_http_request(callback=callback))
+
+    for web_view_link in web_view_links:
+        try:
+            file_id = _extract_file_id_from_web_view_link(web_view_link)
+            request = service.files().get(
+                fileId=file_id,
+                supportsAllDrives=True,
+                fields=fields,
+            )
+            batch.add(request, request_id=web_view_link)
+        except ValueError as e:
+            logger.warning(f"Failed to extract file ID from {web_view_link}: {e}")
+            result.errors[web_view_link] = e
+
+    batch.execute()
+    return result

backend/onyx/connectors/interfaces.py

@@ -298,6 +298,22 @@ class CheckpointedConnectorWithPermSync(CheckpointedConnector[CT]):
         raise NotImplementedError
 
 
+class Resolver(BaseConnector):
+    @abc.abstractmethod
+    def resolve_errors(
+        self,
+        errors: list[ConnectorFailure],
+        include_permissions: bool = False,
+    ) -> Generator[Document | ConnectorFailure | HierarchyNode, None, None]:
+        """Attempts to yield back ALL the documents described by the errors, no checkpointing.
+
+        Caller's responsibility is to delete the old ConnectorFailures and replace with the new ones.
+        If include_permissions is True, the documents will have permissions synced.
+        May also yield HierarchyNode objects for ancestor folders of resolved documents.
+        """
+        raise NotImplementedError
+
+
 class HierarchyConnector(BaseConnector):
     @abc.abstractmethod
     def load_hierarchy(

backend/onyx/connectors/jira/connector.py

@@ -60,8 +60,10 @@ logger = setup_logger()
 
 ONE_HOUR = 3600
 
-_MAX_RESULTS_FETCH_IDS = 5000  # 5000
+_MAX_RESULTS_FETCH_IDS = 5000
 _JIRA_FULL_PAGE_SIZE = 50
+# https://developer.atlassian.com/cloud/jira/platform/rest/v3/api-group-issues/
+_JIRA_BULK_FETCH_LIMIT = 100
 
 # Constants for Jira field names
 _FIELD_REPORTER = "reporter"
@@ -255,15 +257,13 @@ def _bulk_fetch_request(
     return resp.json()["issues"]
 
 
-def bulk_fetch_issues(
-    jira_client: JIRA, issue_ids: list[str], fields: str | None = None
-) -> list[Issue]:
-    # TODO(evan): move away from this jira library if they continue to not support
-    # the endpoints we need. Using private fields is not ideal, but
-    # is likely fine for now since we pin the library version
-
+def _bulk_fetch_batch(
+    jira_client: JIRA, issue_ids: list[str], fields: str | None
+) -> list[dict[str, Any]]:
+    """Fetch a single batch (must be <= _JIRA_BULK_FETCH_LIMIT).
+    On JSONDecodeError, recursively bisects until it succeeds or reaches size 1."""
     try:
-        raw_issues = _bulk_fetch_request(jira_client, issue_ids, fields)
+        return _bulk_fetch_request(jira_client, issue_ids, fields)
     except requests.exceptions.JSONDecodeError:
         if len(issue_ids) <= 1:
             logger.exception(
@@ -277,12 +277,25 @@ def bulk_fetch_issues(
             f"Jira bulk-fetch JSON decode failed for batch of {len(issue_ids)} issues. "
             f"Splitting into sub-batches of {mid} and {len(issue_ids) - mid}."
         )
-        left = bulk_fetch_issues(jira_client, issue_ids[:mid], fields)
-        right = bulk_fetch_issues(jira_client, issue_ids[mid:], fields)
+        left = _bulk_fetch_batch(jira_client, issue_ids[:mid], fields)
+        right = _bulk_fetch_batch(jira_client, issue_ids[mid:], fields)
         return left + right
-    except Exception as e:
-        logger.error(f"Error fetching issues: {e}")
-        raise
+
+
+def bulk_fetch_issues(
+    jira_client: JIRA, issue_ids: list[str], fields: str | None = None
+) -> list[Issue]:
+    # TODO(evan): move away from this jira library if they continue to not support
+    # the endpoints we need. Using private fields is not ideal, but
+    # is likely fine for now since we pin the library version
+
+    raw_issues: list[dict[str, Any]] = []
+    for batch in chunked(issue_ids, _JIRA_BULK_FETCH_LIMIT):
+        try:
+            raw_issues.extend(_bulk_fetch_batch(jira_client, list(batch), fields))
+        except Exception as e:
+            logger.error(f"Error fetching issues: {e}")
+            raise
 
     return [
         Issue(jira_client._options, jira_client._session, raw=issue)

backend/onyx/context/search/federated/models.py

@@ -1,3 +1,4 @@
+from dataclasses import dataclass
 from datetime import datetime
 from typing import TypedDict
 
@@ -6,6 +7,14 @@ from pydantic import BaseModel
 from onyx.onyxbot.slack.models import ChannelType
 
 
+@dataclass(frozen=True)
+class DirectThreadFetch:
+    """Request to fetch a Slack thread directly by channel and timestamp."""
+
+    channel_id: str
+    thread_ts: str
+
+
 class ChannelMetadata(TypedDict):
     """Type definition for cached channel metadata."""
 

backend/onyx/context/search/federated/slack_search.py

@@ -19,6 +19,7 @@ from onyx.configs.chat_configs import DOC_TIME_DECAY
 from onyx.connectors.models import IndexingDocument
 from onyx.connectors.models import TextSection
 from onyx.context.search.federated.models import ChannelMetadata
+from onyx.context.search.federated.models import DirectThreadFetch
 from onyx.context.search.federated.models import SlackMessage
 from onyx.context.search.federated.slack_search_utils import ALL_CHANNEL_TYPES
 from onyx.context.search.federated.slack_search_utils import build_channel_query_filter
@@ -49,7 +50,6 @@ from onyx.server.federated.models import FederatedConnectorDetail
 from onyx.utils.logger import setup_logger
 from onyx.utils.threadpool_concurrency import run_functions_tuples_in_parallel
 from onyx.utils.timing import log_function_time
-from shared_configs.configs import DOC_EMBEDDING_CONTEXT_SIZE
 
 logger = setup_logger()
 
@@ -58,7 +58,6 @@ HIGHLIGHT_END_CHAR = "\ue001"
 
 CHANNEL_METADATA_CACHE_TTL = 60 * 60 * 24  # 24 hours
 USER_PROFILE_CACHE_TTL = 60 * 60 * 24  # 24 hours
-SLACK_THREAD_CONTEXT_WINDOW = 3  # Number of messages before matched message to include
 CHANNEL_METADATA_MAX_RETRIES = 3  # Maximum retry attempts for channel metadata fetching
 CHANNEL_METADATA_RETRY_DELAY = 1  # Initial retry delay in seconds (exponential backoff)
 
@@ -421,6 +420,94 @@ class SlackQueryResult(BaseModel):
     filtered_channels: list[str]  # Channels filtered out during this query
 
 
+def _fetch_thread_from_url(
+    thread_fetch: DirectThreadFetch,
+    access_token: str,
+    channel_metadata_dict: dict[str, ChannelMetadata] | None = None,
+) -> SlackQueryResult:
+    """Fetch a thread directly from a Slack URL via conversations.replies."""
+    channel_id = thread_fetch.channel_id
+    thread_ts = thread_fetch.thread_ts
+
+    slack_client = WebClient(token=access_token)
+    try:
+        response = slack_client.conversations_replies(
+            channel=channel_id,
+            ts=thread_ts,
+        )
+        response.validate()
+        messages: list[dict[str, Any]] = response.get("messages", [])
+    except SlackApiError as e:
+        logger.warning(
+            f"Failed to fetch thread from URL (channel={channel_id}, ts={thread_ts}): {e}"
+        )
+        return SlackQueryResult(messages=[], filtered_channels=[])
+
+    if not messages:
+        logger.warning(
+            f"No messages found for URL override (channel={channel_id}, ts={thread_ts})"
+        )
+        return SlackQueryResult(messages=[], filtered_channels=[])
+
+    # Build thread text from all messages
+    thread_text = _build_thread_text(messages, access_token, None, slack_client)
+
+    # Get channel name from metadata cache or API
+    channel_name = "unknown"
+    if channel_metadata_dict and channel_id in channel_metadata_dict:
+        channel_name = channel_metadata_dict[channel_id].get("name", "unknown")
+    else:
+        try:
+            ch_response = slack_client.conversations_info(channel=channel_id)
+            ch_response.validate()
+            channel_info: dict[str, Any] = ch_response.get("channel", {})
+            channel_name = channel_info.get("name", "unknown")
+        except SlackApiError:
+            pass
+
+    # Build the SlackMessage
+    parent_msg = messages[0]
+    message_ts = parent_msg.get("ts", thread_ts)
+    username = parent_msg.get("user", "unknown_user")
+    parent_text = parent_msg.get("text", "")
+    snippet = (
+        parent_text[:50].rstrip() + "..." if len(parent_text) > 50 else parent_text
+    ).replace("\n", " ")
+
+    doc_time = datetime.fromtimestamp(float(message_ts))
+    decay_factor = DOC_TIME_DECAY
+    doc_age_years = (datetime.now() - doc_time).total_seconds() / (365 * 24 * 60 * 60)
+    recency_bias = max(1 / (1 + decay_factor * doc_age_years), 0.75)
+
+    permalink = (
+        f"https://slack.com/archives/{channel_id}/p{message_ts.replace('.', '')}"
+    )
+
+    slack_message = SlackMessage(
+        document_id=f"{channel_id}_{message_ts}",
+        channel_id=channel_id,
+        message_id=message_ts,
+        thread_id=None,  # Prevent double-enrichment in thread context fetch
+        link=permalink,
+        metadata={
+            "channel": channel_name,
+            "time": doc_time.isoformat(),
+        },
+        timestamp=doc_time,
+        recency_bias=recency_bias,
+        semantic_identifier=f"{username} in #{channel_name}: {snippet}",
+        text=thread_text,
+        highlighted_texts=set(),
+        slack_score=100000.0,  # High priority — user explicitly asked for this thread
+    )
+
+    logger.info(
+        f"URL override: fetched thread from channel={channel_id}, ts={thread_ts}, {len(messages)} messages"
+    )
+
+    return SlackQueryResult(messages=[slack_message], filtered_channels=[])
+
+
 def query_slack(
     query_string: str,
     access_token: str,
@@ -432,7 +519,6 @@ def query_slack(
     available_channels: list[str] | None = None,
     channel_metadata_dict: dict[str, ChannelMetadata] | None = None,
 ) -> SlackQueryResult:
-
     # Check if query has channel override (user specified channels in query)
     has_channel_override = query_string.startswith("__CHANNEL_OVERRIDE__")
 
@@ -662,7 +748,6 @@ def _fetch_thread_context(
     """
     channel_id = message.channel_id
     thread_id = message.thread_id
-    message_id = message.message_id
 
     # If not a thread, return original text as success
     if thread_id is None:
@@ -695,62 +780,37 @@ def _fetch_thread_context(
     if len(messages) <= 1:
         return ThreadContextResult.success(message.text)
 
-    # Build thread text from thread starter + context window around matched message
-    thread_text = _build_thread_text(
-        messages, message_id, thread_id, access_token, team_id, slack_client
-    )
+    # Build thread text from thread starter + all replies
+    thread_text = _build_thread_text(messages, access_token, team_id, slack_client)
     return ThreadContextResult.success(thread_text)
 
 
 def _build_thread_text(
     messages: list[dict[str, Any]],
-    message_id: str,
-    thread_id: str,
     access_token: str,
     team_id: str | None,
     slack_client: WebClient,
 ) -> str:
-    """Build the thread text from messages."""
+    """Build thread text including all replies.
+
+    Includes the thread parent message followed by all replies in order.
+    """
     msg_text = messages[0].get("text", "")
     msg_sender = messages[0].get("user", "")
     thread_text = f"<@{msg_sender}>: {msg_text}"
 
+    # All messages after index 0 are replies
+    replies = messages[1:]
+    if not replies:
+        return thread_text
+
+    logger.debug(f"Thread {messages[0].get('ts')}: {len(replies)} replies included")
     thread_text += "\n\nReplies:"
-    if thread_id == message_id:
-        message_id_idx = 0
-    else:
-        message_id_idx = next(
-            (i for i, msg in enumerate(messages) if msg.get("ts") == message_id), 0
-        )
-        if not message_id_idx:
-            return thread_text
 
-        start_idx = max(1, message_id_idx - SLACK_THREAD_CONTEXT_WINDOW)
-
-        if start_idx > 1:
-            thread_text += "\n..."
-
-        for i in range(start_idx, message_id_idx):
-            msg_text = messages[i].get("text", "")
-            msg_sender = messages[i].get("user", "")
-            thread_text += f"\n\n<@{msg_sender}>: {msg_text}"
-
-        msg_text = messages[message_id_idx].get("text", "")
-        msg_sender = messages[message_id_idx].get("user", "")
-        thread_text += f"\n\n<@{msg_sender}>: {msg_text}"
-
-    # Add following replies
-    len_replies = 0
-    for msg in messages[message_id_idx + 1 :]:
+    for msg in replies:
         msg_text = msg.get("text", "")
         msg_sender = msg.get("user", "")
-        reply = f"\n\n<@{msg_sender}>: {msg_text}"
-        thread_text += reply
-
-        len_replies += len(reply)
-        if len_replies >= DOC_EMBEDDING_CONTEXT_SIZE * 4:
-            thread_text += "\n..."
-            break
+        thread_text += f"\n\n<@{msg_sender}>: {msg_text}"
 
     # Replace user IDs with names using cached lookups
     userids: set[str] = set(re.findall(r"<@([A-Z0-9]+)>", thread_text))
@@ -976,7 +1036,16 @@ def slack_retrieval(
 
     # Query slack with entity filtering
     llm = get_default_llm()
-    query_strings = build_slack_queries(query, llm, entities, available_channels)
+    query_items = build_slack_queries(query, llm, entities, available_channels)
+
+    # Partition into direct thread fetches and search query strings
+    direct_fetches: list[DirectThreadFetch] = []
+    query_strings: list[str] = []
+    for item in query_items:
+        if isinstance(item, DirectThreadFetch):
+            direct_fetches.append(item)
+        else:
+            query_strings.append(item)
 
     # Determine filtering based on entities OR context (bot)
     include_dm = False
@@ -993,8 +1062,16 @@ def slack_retrieval(
                 f"Private channel context: will only allow messages from {allowed_private_channel} + public channels"
             )
 
-    # Build search tasks
-    search_tasks = [
+    # Build search tasks — direct thread fetches + keyword searches
+    search_tasks: list[tuple] = [
+        (
+            _fetch_thread_from_url,
+            (fetch, access_token, channel_metadata_dict),
+        )
+        for fetch in direct_fetches
+    ]
+
+    search_tasks.extend(
         (
             query_slack,
             (
@@ -1010,7 +1087,7 @@ def slack_retrieval(
             ),
         )
         for query_string in query_strings
-    ]
+    )
 
     # If include_dm is True AND we're not already searching all channels,
     # add additional searches without channel filters.

backend/onyx/context/search/federated/slack_search_utils.py

@@ -10,6 +10,7 @@ from pydantic import ValidationError
 
 from onyx.configs.app_configs import MAX_SLACK_QUERY_EXPANSIONS
 from onyx.context.search.federated.models import ChannelMetadata
+from onyx.context.search.federated.models import DirectThreadFetch
 from onyx.context.search.models import ChunkIndexRequest
 from onyx.federated_connectors.slack.models import SlackEntities
 from onyx.llm.interfaces import LLM
@@ -638,12 +639,38 @@ def expand_query_with_llm(query_text: str, llm: LLM) -> list[str]:
         return [query_text]
 
 
+SLACK_URL_PATTERN = re.compile(
+    r"https?://[a-z0-9-]+\.slack\.com/archives/([A-Z0-9]+)/p(\d{16})"
+)
+
+
+def extract_slack_message_urls(
+    query_text: str,
+) -> list[tuple[str, str]]:
+    """Extract Slack message URLs from query text.
+
+    Parses URLs like:
+      https://onyx-company.slack.com/archives/C097NBWMY8Y/p1775491616524769
+
+    Returns list of (channel_id, thread_ts) tuples.
+    The 16-digit timestamp is converted to Slack ts format (with dot).
+    """
+    results = []
+    for match in SLACK_URL_PATTERN.finditer(query_text):
+        channel_id = match.group(1)
+        raw_ts = match.group(2)
+        # Convert p1775491616524769 -> 1775491616.524769
+        thread_ts = f"{raw_ts[:10]}.{raw_ts[10:]}"
+        results.append((channel_id, thread_ts))
+    return results
+
+
 def build_slack_queries(
     query: ChunkIndexRequest,
     llm: LLM,
     entities: dict[str, Any] | None = None,
     available_channels: list[str] | None = None,
-) -> list[str]:
+) -> list[str | DirectThreadFetch]:
     """Build Slack query strings with date filtering and query expansion."""
     default_search_days = 30
     if entities:
@@ -668,6 +695,15 @@ def build_slack_queries(
             cutoff_date = datetime.now(timezone.utc) - timedelta(days=days_back)
             time_filter = f" after:{cutoff_date.strftime('%Y-%m-%d')}"
 
+    # Check for Slack message URLs — if found, add direct fetch requests
+    url_fetches: list[DirectThreadFetch] = []
+    slack_urls = extract_slack_message_urls(query.query)
+    for channel_id, thread_ts in slack_urls:
+        url_fetches.append(
+            DirectThreadFetch(channel_id=channel_id, thread_ts=thread_ts)
+        )
+        logger.info(f"Detected Slack URL: channel={channel_id}, ts={thread_ts}")
+
     # ALWAYS extract channel references from the query (not just for recency queries)
     channel_references = extract_channel_references_from_query(query.query)
 
@@ -684,7 +720,9 @@ def build_slack_queries(
 
             # If valid channels detected, use ONLY those channels with NO keywords
             # Return query with ONLY time filter + channel filter (no keywords)
-            return [build_channel_override_query(channel_references, time_filter)]
+            return url_fetches + [
+                build_channel_override_query(channel_references, time_filter)
+            ]
         except ValueError as e:
             # If validation fails, log the error and continue with normal flow
             logger.warning(f"Channel reference validation failed: {e}")
@@ -702,7 +740,8 @@ def build_slack_queries(
         rephrased_queries = expand_query_with_llm(query.query, llm)
 
     # Build final query strings with time filters
-    return [
+    search_queries = [
         rephrased_query.strip() + time_filter
         for rephrased_query in rephrased_queries[:MAX_SLACK_QUERY_EXPANSIONS]
     ]
+    return url_fetches + search_queries

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

@@ -111,6 +111,43 @@ def _mask_string(value: str) -> str:
     return value[:4] + "****" + value[-4:]
 
 
+def _resolve_api_key(
+    api_key: str | None,
+    provider_name: str | None,
+    api_base: str | None,
+    db_session: Session,
+) -> str | None:
+    """Return the real API key for model-fetch endpoints.
+
+    When editing an existing provider the form value is masked (e.g.
+    ``sk-a****b1c2``).  If *provider_name* is supplied we can look up
+    the unmasked key from the database so the external request succeeds.
+
+    The stored key is only returned when the request's *api_base*
+    matches the value stored in the database.
+    """
+    if not provider_name:
+        return api_key
+
+    existing_provider = fetch_existing_llm_provider(
+        name=provider_name, db_session=db_session
+    )
+    if existing_provider and existing_provider.api_key:
+        # Normalise both URLs before comparing so trailing-slash
+        # differences don't cause a false mismatch.
+        stored_base = (existing_provider.api_base or "").strip().rstrip("/")
+        request_base = (api_base or "").strip().rstrip("/")
+        if stored_base != request_base:
+            return api_key
+
+        stored_key = existing_provider.api_key.get_value(apply_mask=False)
+        # Only resolve when the incoming value is the masked form of the
+        # stored key — i.e. the user hasn't typed a new key.
+        if api_key and api_key == _mask_string(stored_key):
+            return stored_key
+    return api_key
+
+
 def _sync_fetched_models(
     db_session: Session,
     provider_name: str,
@@ -1174,16 +1211,17 @@ def get_ollama_available_models(
     return sorted_results
 
 
-def _get_openrouter_models_response(api_base: str, api_key: str) -> dict:
+def _get_openrouter_models_response(api_base: str, api_key: str | None) -> dict:
     """Perform GET to OpenRouter /models and return parsed JSON."""
     cleaned_api_base = api_base.strip().rstrip("/")
     url = f"{cleaned_api_base}/models"
-    headers = {
-        "Authorization": f"Bearer {api_key}",
+    headers: dict[str, str] = {
         # Optional headers recommended by OpenRouter for attribution
         "HTTP-Referer": "https://onyx.app",
         "X-Title": "Onyx",
     }
+    if api_key:
+        headers["Authorization"] = f"Bearer {api_key}"
     try:
         response = httpx.get(url, headers=headers, timeout=10.0)
         response.raise_for_status()
@@ -1206,8 +1244,12 @@ def get_openrouter_available_models(
     Parses id, name (display), context_length, and architecture.input_modalities.
     """
 
+    api_key = _resolve_api_key(
+        request.api_key, request.provider_name, request.api_base, db_session
+    )
+
     response_json = _get_openrouter_models_response(
-        api_base=request.api_base, api_key=request.api_key
+        api_base=request.api_base, api_key=api_key
     )
 
     data = response_json.get("data", [])
@@ -1300,13 +1342,18 @@ def get_lm_studio_available_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).
+    # Only do so when the api_base matches what is stored.
     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)
+            stored_base = (existing_provider.api_base or "").strip().rstrip("/")
+            if stored_base == cleaned_api_base:
+                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] = {}
@@ -1390,8 +1437,12 @@ def get_litellm_available_models(
     db_session: Session = Depends(get_session),
 ) -> list[LitellmFinalModelResponse]:
     """Fetch available models from Litellm proxy /v1/models endpoint."""
+    api_key = _resolve_api_key(
+        request.api_key, request.provider_name, request.api_base, db_session
+    )
+
     response_json = _get_litellm_models_response(
-        api_key=request.api_key, api_base=request.api_base
+        api_key=api_key, api_base=request.api_base
     )
 
     models = response_json.get("data", [])
@@ -1448,7 +1499,7 @@ def get_litellm_available_models(
     return sorted_results
 
 
-def _get_litellm_models_response(api_key: str, api_base: str) -> dict:
+def _get_litellm_models_response(api_key: str | None, api_base: str) -> dict:
     """Perform GET to Litellm proxy /api/v1/models and return parsed JSON."""
     cleaned_api_base = api_base.strip().rstrip("/")
     url = f"{cleaned_api_base}/v1/models"
@@ -1523,8 +1574,12 @@ def get_bifrost_available_models(
     db_session: Session = Depends(get_session),
 ) -> list[BifrostFinalModelResponse]:
     """Fetch available models from Bifrost gateway /v1/models endpoint."""
+    api_key = _resolve_api_key(
+        request.api_key, request.provider_name, request.api_base, db_session
+    )
+
     response_json = _get_bifrost_models_response(
-        api_base=request.api_base, api_key=request.api_key
+        api_base=request.api_base, api_key=api_key
     )
 
     models = response_json.get("data", [])
@@ -1613,8 +1668,12 @@ def get_openai_compatible_server_available_models(
     db_session: Session = Depends(get_session),
 ) -> list[OpenAICompatibleFinalModelResponse]:
     """Fetch available models from a generic OpenAI-compatible /v1/models endpoint."""
+    api_key = _resolve_api_key(
+        request.api_key, request.provider_name, request.api_base, db_session
+    )
+
     response_json = _get_openai_compatible_server_response(
-        api_base=request.api_base, api_key=request.api_key
+        api_base=request.api_base, api_key=api_key
     )
 
     models = response_json.get("data", [])

backend/requirements/dev.txt

@@ -254,7 +254,7 @@ oauthlib==3.2.2
     # via
     #   kubernetes
     #   requests-oauthlib
-onyx-devtools==0.7.3
+onyx-devtools==0.7.5
 openai==2.14.0
     # via
     #   litellm

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

@@ -0,0 +1,239 @@
+"""Tests for GoogleDriveConnector.resolve_errors against real Google Drive."""
+
+import json
+import os
+from collections.abc import Callable
+from unittest.mock import patch
+
+from onyx.connectors.google_drive.connector import GoogleDriveConnector
+from onyx.connectors.models import ConnectorFailure
+from onyx.connectors.models import Document
+from onyx.connectors.models import DocumentFailure
+from onyx.connectors.models import HierarchyNode
+from tests.daily.connectors.google_drive.consts_and_utils import ADMIN_EMAIL
+from tests.daily.connectors.google_drive.consts_and_utils import (
+    ALL_EXPECTED_HIERARCHY_NODES,
+)
+from tests.daily.connectors.google_drive.consts_and_utils import FOLDER_1_ID
+from tests.daily.connectors.google_drive.consts_and_utils import SHARED_DRIVE_1_ID
+
+_DRIVE_ID_MAPPING_PATH = os.path.join(
+    os.path.dirname(__file__), "drive_id_mapping.json"
+)
+
+
+def _load_web_view_links(file_ids: list[int]) -> list[str]:
+    with open(_DRIVE_ID_MAPPING_PATH) as f:
+        mapping: dict[str, str] = json.load(f)
+    return [mapping[str(fid)] for fid in file_ids]
+
+
+def _build_failures(web_view_links: list[str]) -> list[ConnectorFailure]:
+    return [
+        ConnectorFailure(
+            failed_document=DocumentFailure(
+                document_id=link,
+                document_link=link,
+            ),
+            failure_message=f"Synthetic failure for {link}",
+        )
+        for link in web_view_links
+    ]
+
+
+@patch("onyx.file_processing.extract_file_text.get_unstructured_api_key")
+def test_resolve_single_file(
+    mock_api_key: None,  # noqa: ARG001
+    google_drive_service_acct_connector_factory: Callable[..., GoogleDriveConnector],
+) -> None:
+    """Resolve a single known file and verify we get back exactly one Document."""
+    connector = google_drive_service_acct_connector_factory(
+        primary_admin_email=ADMIN_EMAIL,
+        include_shared_drives=True,
+        shared_drive_urls=None,
+        include_my_drives=True,
+        my_drive_emails=None,
+        shared_folder_urls=None,
+        include_files_shared_with_me=False,
+    )
+
+    web_view_links = _load_web_view_links([0])
+    failures = _build_failures(web_view_links)
+
+    results = list(connector.resolve_errors(failures))
+
+    docs = [r for r in results if isinstance(r, Document)]
+    new_failures = [r for r in results if isinstance(r, ConnectorFailure)]
+    hierarchy_nodes = [r for r in results if isinstance(r, HierarchyNode)]
+
+    assert len(docs) == 1
+    assert len(new_failures) == 0
+    assert docs[0].semantic_identifier == "file_0.txt"
+
+    # Should yield at least one hierarchy node (the file's parent folder chain)
+    assert len(hierarchy_nodes) > 0
+
+
+@patch("onyx.file_processing.extract_file_text.get_unstructured_api_key")
+def test_resolve_multiple_files(
+    mock_api_key: None,  # noqa: ARG001
+    google_drive_service_acct_connector_factory: Callable[..., GoogleDriveConnector],
+) -> None:
+    """Resolve multiple files across different folders via batch API."""
+    connector = google_drive_service_acct_connector_factory(
+        primary_admin_email=ADMIN_EMAIL,
+        include_shared_drives=True,
+        shared_drive_urls=None,
+        include_my_drives=True,
+        my_drive_emails=None,
+        shared_folder_urls=None,
+        include_files_shared_with_me=False,
+    )
+
+    # Pick files from different folders: admin files (0-4), shared drive 1 (20-24), folder_2 (45-49)
+    file_ids = [0, 1, 20, 21, 45]
+    web_view_links = _load_web_view_links(file_ids)
+    failures = _build_failures(web_view_links)
+
+    results = list(connector.resolve_errors(failures))
+
+    docs = [r for r in results if isinstance(r, Document)]
+    new_failures = [r for r in results if isinstance(r, ConnectorFailure)]
+    hierarchy_nodes = [r for r in results if isinstance(r, HierarchyNode)]
+
+    assert len(new_failures) == 0
+    retrieved_names = {doc.semantic_identifier for doc in docs}
+    expected_names = {f"file_{fid}.txt" for fid in file_ids}
+    assert expected_names == retrieved_names
+
+    # Files span multiple folders, so we should get hierarchy nodes
+    assert len(hierarchy_nodes) > 0
+
+
+@patch("onyx.file_processing.extract_file_text.get_unstructured_api_key")
+def test_resolve_hierarchy_nodes_are_valid(
+    mock_api_key: None,  # noqa: ARG001
+    google_drive_service_acct_connector_factory: Callable[..., GoogleDriveConnector],
+) -> None:
+    """Verify that hierarchy nodes from resolve_errors match expected structure."""
+    connector = google_drive_service_acct_connector_factory(
+        primary_admin_email=ADMIN_EMAIL,
+        include_shared_drives=True,
+        shared_drive_urls=None,
+        include_my_drives=True,
+        my_drive_emails=None,
+        shared_folder_urls=None,
+        include_files_shared_with_me=False,
+    )
+
+    # File in folder_1 (inside shared_drive_1) — should walk up to shared_drive_1 root
+    web_view_links = _load_web_view_links([25])
+    failures = _build_failures(web_view_links)
+
+    results = list(connector.resolve_errors(failures))
+
+    hierarchy_nodes = [r for r in results if isinstance(r, HierarchyNode)]
+    node_ids = {node.raw_node_id for node in hierarchy_nodes}
+
+    # File 25 is in folder_1 which is inside shared_drive_1.
+    # The parent walk must yield at least these two ancestors.
+    assert (
+        FOLDER_1_ID in node_ids
+    ), f"Expected folder_1 ({FOLDER_1_ID}) in hierarchy nodes, got: {node_ids}"
+    assert (
+        SHARED_DRIVE_1_ID in node_ids
+    ), f"Expected shared_drive_1 ({SHARED_DRIVE_1_ID}) in hierarchy nodes, got: {node_ids}"
+
+    for node in hierarchy_nodes:
+        if node.raw_node_id not in ALL_EXPECTED_HIERARCHY_NODES:
+            continue
+        expected = ALL_EXPECTED_HIERARCHY_NODES[node.raw_node_id]
+        assert node.display_name == expected.display_name, (
+            f"Display name mismatch for {node.raw_node_id}: "
+            f"expected '{expected.display_name}', got '{node.display_name}'"
+        )
+        assert node.node_type == expected.node_type, (
+            f"Node type mismatch for {node.raw_node_id}: "
+            f"expected '{expected.node_type}', got '{node.node_type}'"
+        )
+
+
+@patch("onyx.file_processing.extract_file_text.get_unstructured_api_key")
+def test_resolve_with_invalid_link(
+    mock_api_key: None,  # noqa: ARG001
+    google_drive_service_acct_connector_factory: Callable[..., GoogleDriveConnector],
+) -> None:
+    """Resolve with a mix of valid and invalid links — invalid ones yield ConnectorFailure."""
+    connector = google_drive_service_acct_connector_factory(
+        primary_admin_email=ADMIN_EMAIL,
+        include_shared_drives=True,
+        shared_drive_urls=None,
+        include_my_drives=True,
+        my_drive_emails=None,
+        shared_folder_urls=None,
+        include_files_shared_with_me=False,
+    )
+
+    valid_links = _load_web_view_links([0])
+    invalid_link = "https://drive.google.com/file/d/NONEXISTENT_FILE_ID_12345"
+    failures = _build_failures(valid_links + [invalid_link])
+
+    results = list(connector.resolve_errors(failures))
+
+    docs = [r for r in results if isinstance(r, Document)]
+    new_failures = [r for r in results if isinstance(r, ConnectorFailure)]
+
+    assert len(docs) == 1
+    assert docs[0].semantic_identifier == "file_0.txt"
+    assert len(new_failures) == 1
+    assert new_failures[0].failed_document is not None
+    assert new_failures[0].failed_document.document_id == invalid_link
+
+
+@patch("onyx.file_processing.extract_file_text.get_unstructured_api_key")
+def test_resolve_empty_errors(
+    mock_api_key: None,  # noqa: ARG001
+    google_drive_service_acct_connector_factory: Callable[..., GoogleDriveConnector],
+) -> None:
+    """Resolving an empty error list should yield nothing."""
+    connector = google_drive_service_acct_connector_factory(
+        primary_admin_email=ADMIN_EMAIL,
+        include_shared_drives=True,
+        shared_drive_urls=None,
+        include_my_drives=True,
+        my_drive_emails=None,
+        shared_folder_urls=None,
+        include_files_shared_with_me=False,
+    )
+
+    results = list(connector.resolve_errors([]))
+
+    assert len(results) == 0
+
+
+@patch("onyx.file_processing.extract_file_text.get_unstructured_api_key")
+def test_resolve_entity_failures_are_skipped(
+    mock_api_key: None,  # noqa: ARG001
+    google_drive_service_acct_connector_factory: Callable[..., GoogleDriveConnector],
+) -> None:
+    """Entity failures (not document failures) should be skipped by resolve_errors."""
+    from onyx.connectors.models import EntityFailure
+
+    connector = google_drive_service_acct_connector_factory(
+        primary_admin_email=ADMIN_EMAIL,
+        include_shared_drives=True,
+        shared_drive_urls=None,
+        include_my_drives=True,
+        my_drive_emails=None,
+        shared_folder_urls=None,
+        include_files_shared_with_me=False,
+    )
+
+    entity_failure = ConnectorFailure(
+        failed_entity=EntityFailure(entity_id="some_stage"),
+        failure_message="retrieval failure",
+    )
+
+    results = list(connector.resolve_errors([entity_failure]))
+
+    assert len(results) == 0

backend/tests/unit/ee/onyx/db/test_license.py

@@ -9,6 +9,7 @@ from unittest.mock import patch
 from ee.onyx.db.license import check_seat_availability
 from ee.onyx.db.license import delete_license
 from ee.onyx.db.license import get_license
+from ee.onyx.db.license import get_used_seats
 from ee.onyx.db.license import upsert_license
 from ee.onyx.server.license.models import LicenseMetadata
 from ee.onyx.server.license.models import LicenseSource
@@ -214,3 +215,43 @@ class TestCheckSeatAvailabilityMultiTenant:
         assert result.available is False
         assert result.error_message is not None
         mock_tenant_count.assert_called_once_with("tenant-abc")
+
+
+class TestGetUsedSeatsAccountTypeFiltering:
+    """Verify get_used_seats query excludes SERVICE_ACCOUNT but includes BOT."""
+
+    @patch("ee.onyx.db.license.MULTI_TENANT", False)
+    @patch("onyx.db.engine.sql_engine.get_session_with_current_tenant")
+    def test_excludes_service_accounts(self, mock_get_session: MagicMock) -> None:
+        """SERVICE_ACCOUNT users should not count toward seats."""
+        mock_session = MagicMock()
+        mock_get_session.return_value.__enter__ = MagicMock(return_value=mock_session)
+        mock_get_session.return_value.__exit__ = MagicMock(return_value=False)
+        mock_session.execute.return_value.scalar.return_value = 5
+
+        result = get_used_seats()
+
+        assert result == 5
+        # Inspect the compiled query to verify account_type filter
+        call_args = mock_session.execute.call_args
+        query = call_args[0][0]
+        compiled = str(query.compile(compile_kwargs={"literal_binds": True}))
+        assert "SERVICE_ACCOUNT" in compiled
+        # BOT should NOT be excluded
+        assert "BOT" not in compiled
+
+    @patch("ee.onyx.db.license.MULTI_TENANT", False)
+    @patch("onyx.db.engine.sql_engine.get_session_with_current_tenant")
+    def test_still_excludes_ext_perm_user(self, mock_get_session: MagicMock) -> None:
+        """EXT_PERM_USER exclusion should still be present."""
+        mock_session = MagicMock()
+        mock_get_session.return_value.__enter__ = MagicMock(return_value=mock_session)
+        mock_get_session.return_value.__exit__ = MagicMock(return_value=False)
+        mock_session.execute.return_value.scalar.return_value = 3
+
+        get_used_seats()
+
+        call_args = mock_session.execute.call_args
+        query = call_args[0][0]
+        compiled = str(query.compile(compile_kwargs={"literal_binds": True}))
+        assert "EXT_PERM_USER" in compiled

backend/tests/unit/onyx/connectors/jira/test_jira_bulk_fetch.py

@@ -6,6 +6,7 @@ import requests
 from jira import JIRA
 from jira.resources import Issue
 
+from onyx.connectors.jira.connector import _JIRA_BULK_FETCH_LIMIT
 from onyx.connectors.jira.connector import bulk_fetch_issues
 
 
@@ -145,3 +146,29 @@ def test_bulk_fetch_recursive_splitting_raises_on_bad_issue() -> None:
 
     with pytest.raises(requests.exceptions.JSONDecodeError):
         bulk_fetch_issues(client, ["1", "2", bad_id, "3", "4", "5"])
+
+
+def test_bulk_fetch_respects_api_batch_limit() -> None:
+    """Requests to the bulkfetch endpoint never exceed _JIRA_BULK_FETCH_LIMIT IDs."""
+    client = _mock_jira_client()
+    total_issues = _JIRA_BULK_FETCH_LIMIT * 3 + 7
+    all_ids = [str(i) for i in range(total_issues)]
+
+    batch_sizes: list[int] = []
+
+    def _post_side_effect(url: str, json: dict[str, Any]) -> MagicMock:  # noqa: ARG001
+        ids = json["issueIdsOrKeys"]
+        batch_sizes.append(len(ids))
+        resp = MagicMock()
+        resp.json.return_value = {"issues": [_make_raw_issue(i) for i in ids]}
+        return resp
+
+    client._session.post.side_effect = _post_side_effect
+
+    result = bulk_fetch_issues(client, all_ids)
+
+    assert len(result) == total_issues
+    # keeping this hardcoded because it's the documented limit
+    # https://developer.atlassian.com/cloud/jira/platform/rest/v3/api-group-issues/
+    assert all(size <= 100 for size in batch_sizes)
+    assert len(batch_sizes) == 4

backend/tests/unit/onyx/context/search/federated/test_build_thread_text.py

@@ -0,0 +1,67 @@
+"""Tests for _build_thread_text function."""
+
+from unittest.mock import MagicMock
+from unittest.mock import patch
+
+from onyx.context.search.federated.slack_search import _build_thread_text
+
+
+def _make_msg(user: str, text: str, ts: str) -> dict[str, str]:
+    return {"user": user, "text": text, "ts": ts}
+
+
+class TestBuildThreadText:
+    """Verify _build_thread_text includes full thread replies up to cap."""
+
+    @patch("onyx.context.search.federated.slack_search.batch_get_user_profiles")
+    def test_includes_all_replies(self, mock_profiles: MagicMock) -> None:
+        """All replies within cap are included in output."""
+        mock_profiles.return_value = {}
+        messages = [
+            _make_msg("U1", "parent msg", "1000.0"),
+            _make_msg("U2", "reply 1", "1001.0"),
+            _make_msg("U3", "reply 2", "1002.0"),
+            _make_msg("U4", "reply 3", "1003.0"),
+        ]
+        result = _build_thread_text(messages, "token", "T123", MagicMock())
+        assert "parent msg" in result
+        assert "reply 1" in result
+        assert "reply 2" in result
+        assert "reply 3" in result
+        assert "..." not in result
+
+    @patch("onyx.context.search.federated.slack_search.batch_get_user_profiles")
+    def test_non_thread_returns_parent_only(self, mock_profiles: MagicMock) -> None:
+        """Single message (no replies) returns just the parent text."""
+        mock_profiles.return_value = {}
+        messages = [_make_msg("U1", "just a message", "1000.0")]
+        result = _build_thread_text(messages, "token", "T123", MagicMock())
+        assert "just a message" in result
+        assert "Replies:" not in result
+
+    @patch("onyx.context.search.federated.slack_search.batch_get_user_profiles")
+    def test_parent_always_first(self, mock_profiles: MagicMock) -> None:
+        """Thread parent message is always the first line of output."""
+        mock_profiles.return_value = {}
+        messages = [
+            _make_msg("U1", "I am the parent", "1000.0"),
+            _make_msg("U2", "I am a reply", "1001.0"),
+        ]
+        result = _build_thread_text(messages, "token", "T123", MagicMock())
+        parent_pos = result.index("I am the parent")
+        reply_pos = result.index("I am a reply")
+        assert parent_pos < reply_pos
+
+    @patch("onyx.context.search.federated.slack_search.batch_get_user_profiles")
+    def test_user_profiles_resolved(self, mock_profiles: MagicMock) -> None:
+        """User IDs in thread text are replaced with display names."""
+        mock_profiles.return_value = {"U1": "Alice", "U2": "Bob"}
+        messages = [
+            _make_msg("U1", "hello", "1000.0"),
+            _make_msg("U2", "world", "1001.0"),
+        ]
+        result = _build_thread_text(messages, "token", "T123", MagicMock())
+        assert "Alice" in result
+        assert "Bob" in result
+        assert "<@U1>" not in result
+        assert "<@U2>" not in result

backend/tests/unit/onyx/context/search/federated/test_url_override.py

@@ -0,0 +1,108 @@
+"""Tests for Slack URL parsing and direct thread fetch via URL override."""
+
+from unittest.mock import MagicMock
+from unittest.mock import patch
+
+from onyx.context.search.federated.models import DirectThreadFetch
+from onyx.context.search.federated.slack_search import _fetch_thread_from_url
+from onyx.context.search.federated.slack_search_utils import extract_slack_message_urls
+
+
+class TestExtractSlackMessageUrls:
+    """Verify URL parsing extracts channel_id and timestamp correctly."""
+
+    def test_standard_url(self) -> None:
+        query = "summarize https://mycompany.slack.com/archives/C097NBWMY8Y/p1775491616524769"
+        results = extract_slack_message_urls(query)
+        assert len(results) == 1
+        assert results[0] == ("C097NBWMY8Y", "1775491616.524769")
+
+    def test_multiple_urls(self) -> None:
+        query = (
+            "compare https://co.slack.com/archives/C111/p1234567890123456 "
+            "and https://co.slack.com/archives/C222/p9876543210987654"
+        )
+        results = extract_slack_message_urls(query)
+        assert len(results) == 2
+        assert results[0] == ("C111", "1234567890.123456")
+        assert results[1] == ("C222", "9876543210.987654")
+
+    def test_no_urls(self) -> None:
+        query = "what happened in #general last week?"
+        results = extract_slack_message_urls(query)
+        assert len(results) == 0
+
+    def test_non_slack_url_ignored(self) -> None:
+        query = "check https://google.com/archives/C111/p1234567890123456"
+        results = extract_slack_message_urls(query)
+        assert len(results) == 0
+
+    def test_timestamp_conversion(self) -> None:
+        """p prefix removed, dot inserted after 10th digit."""
+        query = "https://x.slack.com/archives/CABC123/p1775491616524769"
+        results = extract_slack_message_urls(query)
+        channel_id, ts = results[0]
+        assert channel_id == "CABC123"
+        assert ts == "1775491616.524769"
+        assert not ts.startswith("p")
+        assert "." in ts
+
+
+class TestFetchThreadFromUrl:
+    """Verify _fetch_thread_from_url calls conversations.replies and returns SlackMessage."""
+
+    @patch("onyx.context.search.federated.slack_search._build_thread_text")
+    @patch("onyx.context.search.federated.slack_search.WebClient")
+    def test_successful_fetch(
+        self, mock_webclient_cls: MagicMock, mock_build_thread: MagicMock
+    ) -> None:
+        mock_client = MagicMock()
+        mock_webclient_cls.return_value = mock_client
+
+        # Mock conversations_replies
+        mock_response = MagicMock()
+        mock_response.get.return_value = [
+            {"user": "U1", "text": "parent", "ts": "1775491616.524769"},
+            {"user": "U2", "text": "reply 1", "ts": "1775491617.000000"},
+            {"user": "U3", "text": "reply 2", "ts": "1775491618.000000"},
+        ]
+        mock_client.conversations_replies.return_value = mock_response
+
+        # Mock channel info
+        mock_ch_response = MagicMock()
+        mock_ch_response.get.return_value = {"name": "general"}
+        mock_client.conversations_info.return_value = mock_ch_response
+
+        mock_build_thread.return_value = (
+            "U1: parent\n\nReplies:\n\nU2: reply 1\n\nU3: reply 2"
+        )
+
+        fetch = DirectThreadFetch(
+            channel_id="C097NBWMY8Y", thread_ts="1775491616.524769"
+        )
+        result = _fetch_thread_from_url(fetch, "xoxp-token")
+
+        assert len(result.messages) == 1
+        msg = result.messages[0]
+        assert msg.channel_id == "C097NBWMY8Y"
+        assert msg.thread_id is None  # Prevents double-enrichment
+        assert msg.slack_score == 100000.0
+        assert "parent" in msg.text
+        mock_client.conversations_replies.assert_called_once_with(
+            channel="C097NBWMY8Y", ts="1775491616.524769"
+        )
+
+    @patch("onyx.context.search.federated.slack_search.WebClient")
+    def test_api_error_returns_empty(self, mock_webclient_cls: MagicMock) -> None:
+        from slack_sdk.errors import SlackApiError
+
+        mock_client = MagicMock()
+        mock_webclient_cls.return_value = mock_client
+        mock_client.conversations_replies.side_effect = SlackApiError(
+            message="channel_not_found",
+            response=MagicMock(status_code=404),
+        )
+
+        fetch = DirectThreadFetch(channel_id="CBAD", thread_ts="1234567890.123456")
+        result = _fetch_thread_from_url(fetch, "xoxp-token")
+        assert len(result.messages) == 0

backend/tests/unit/onyx/server/manage/llm/test_fetch_models_api.py

@@ -505,6 +505,7 @@ class TestGetLMStudioAvailableModels:
 
         mock_session = MagicMock()
         mock_provider = MagicMock()
+        mock_provider.api_base = "http://localhost:1234"
         mock_provider.custom_config = {"LM_STUDIO_API_KEY": "stored-secret"}
 
         response = {

deployment/helm/charts/onyx/Chart.yaml

@@ -5,7 +5,7 @@ home: https://www.onyx.app/
 sources:
   - "https://github.com/onyx-dot-app/onyx"
 type: application
-version: 0.4.40
+version: 0.4.42
 appVersion: latest
 annotations:
   category: Productivity

deployment/helm/charts/onyx/dashboards/opensearch-search-latency.json

@@ -0,0 +1,349 @@
+{
+  "annotations": {
+    "list": [
+      {
+        "builtIn": 1,
+        "datasource": { "type": "grafana", "uid": "-- Grafana --" },
+        "enable": true,
+        "hide": true,
+        "iconColor": "rgba(0, 211, 255, 1)",
+        "name": "Annotations & Alerts",
+        "type": "dashboard"
+      }
+    ]
+  },
+  "editable": true,
+  "fiscalYearStartMonth": 0,
+  "graphTooltip": 1,
+  "id": null,
+  "links": [],
+  "liveNow": true,
+  "panels": [
+    {
+      "title": "Client-Side Search Latency (P50 / P95 / P99)",
+      "description": "End-to-end latency as measured by the Python client, including network round-trip and serialization overhead.",
+      "type": "timeseries",
+      "gridPos": { "h": 10, "w": 12, "x": 0, "y": 0 },
+      "id": 1,
+      "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+      "fieldConfig": {
+        "defaults": {
+          "color": { "mode": "palette-classic" },
+          "custom": {
+            "axisBorderShow": false,
+            "axisCenteredZero": false,
+            "axisLabel": "seconds",
+            "axisPlacement": "auto",
+            "drawStyle": "line",
+            "fillOpacity": 0,
+            "gradientMode": "none",
+            "lineInterpolation": "smooth",
+            "lineWidth": 2,
+            "pointSize": 5,
+            "scaleDistribution": { "type": "linear" },
+            "showPoints": "never",
+            "spanNulls": false,
+            "stacking": { "group": "A", "mode": "none" },
+            "thresholdsStyle": { "mode": "dashed" }
+          },
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              { "color": "green", "value": null },
+              { "color": "yellow", "value": 0.5 },
+              { "color": "red", "value": 2.0 }
+            ]
+          },
+          "unit": "s",
+          "min": 0
+        },
+        "overrides": []
+      },
+      "targets": [
+        {
+          "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+          "expr": "histogram_quantile(0.5, sum by (le) (rate(onyx_opensearch_search_client_duration_seconds_bucket[5m])))",
+          "legendFormat": "P50",
+          "refId": "A"
+        },
+        {
+          "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+          "expr": "histogram_quantile(0.95, sum by (le) (rate(onyx_opensearch_search_client_duration_seconds_bucket[5m])))",
+          "legendFormat": "P95",
+          "refId": "B"
+        },
+        {
+          "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+          "expr": "histogram_quantile(0.99, sum by (le) (rate(onyx_opensearch_search_client_duration_seconds_bucket[5m])))",
+          "legendFormat": "P99",
+          "refId": "C"
+        }
+      ]
+    },
+    {
+      "title": "Server-Side Search Latency (P50 / P95 / P99)",
+      "description": "OpenSearch server-side execution time from the 'took' field in the response. Does not include network or client-side overhead.",
+      "type": "timeseries",
+      "gridPos": { "h": 10, "w": 12, "x": 12, "y": 0 },
+      "id": 2,
+      "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+      "fieldConfig": {
+        "defaults": {
+          "color": { "mode": "palette-classic" },
+          "custom": {
+            "axisBorderShow": false,
+            "axisCenteredZero": false,
+            "axisLabel": "seconds",
+            "axisPlacement": "auto",
+            "drawStyle": "line",
+            "fillOpacity": 0,
+            "gradientMode": "none",
+            "lineInterpolation": "smooth",
+            "lineWidth": 2,
+            "pointSize": 5,
+            "scaleDistribution": { "type": "linear" },
+            "showPoints": "never",
+            "spanNulls": false,
+            "stacking": { "group": "A", "mode": "none" },
+            "thresholdsStyle": { "mode": "dashed" }
+          },
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              { "color": "green", "value": null },
+              { "color": "yellow", "value": 0.5 },
+              { "color": "red", "value": 2.0 }
+            ]
+          },
+          "unit": "s",
+          "min": 0
+        },
+        "overrides": []
+      },
+      "targets": [
+        {
+          "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+          "expr": "histogram_quantile(0.5, sum by (le) (rate(onyx_opensearch_search_server_duration_seconds_bucket[5m])))",
+          "legendFormat": "P50",
+          "refId": "A"
+        },
+        {
+          "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+          "expr": "histogram_quantile(0.95, sum by (le) (rate(onyx_opensearch_search_server_duration_seconds_bucket[5m])))",
+          "legendFormat": "P95",
+          "refId": "B"
+        },
+        {
+          "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+          "expr": "histogram_quantile(0.99, sum by (le) (rate(onyx_opensearch_search_server_duration_seconds_bucket[5m])))",
+          "legendFormat": "P99",
+          "refId": "C"
+        }
+      ]
+    },
+    {
+      "title": "Client-Side Latency by Search Type (P95)",
+      "description": "P95 client-side latency broken down by search type (hybrid, keyword, semantic, random, doc_id_retrieval).",
+      "type": "timeseries",
+      "gridPos": { "h": 10, "w": 12, "x": 0, "y": 10 },
+      "id": 3,
+      "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+      "fieldConfig": {
+        "defaults": {
+          "color": { "mode": "palette-classic" },
+          "custom": {
+            "axisBorderShow": false,
+            "axisCenteredZero": false,
+            "axisLabel": "seconds",
+            "axisPlacement": "auto",
+            "drawStyle": "line",
+            "fillOpacity": 0,
+            "gradientMode": "none",
+            "lineInterpolation": "smooth",
+            "lineWidth": 2,
+            "pointSize": 5,
+            "scaleDistribution": { "type": "linear" },
+            "showPoints": "never",
+            "spanNulls": false,
+            "stacking": { "group": "A", "mode": "none" },
+            "thresholdsStyle": { "mode": "off" }
+          },
+          "unit": "s",
+          "min": 0
+        },
+        "overrides": []
+      },
+      "targets": [
+        {
+          "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+          "expr": "histogram_quantile(0.95, sum by (search_type, le) (rate(onyx_opensearch_search_client_duration_seconds_bucket[5m])))",
+          "legendFormat": "{{ search_type }}",
+          "refId": "A"
+        }
+      ]
+    },
+    {
+      "title": "Search Throughput by Type",
+      "description": "Searches per second broken down by search type.",
+      "type": "timeseries",
+      "gridPos": { "h": 10, "w": 12, "x": 12, "y": 10 },
+      "id": 4,
+      "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+      "fieldConfig": {
+        "defaults": {
+          "color": { "mode": "palette-classic" },
+          "custom": {
+            "axisBorderShow": false,
+            "axisCenteredZero": false,
+            "axisLabel": "searches/s",
+            "axisPlacement": "auto",
+            "drawStyle": "line",
+            "fillOpacity": 0,
+            "gradientMode": "none",
+            "lineInterpolation": "smooth",
+            "lineWidth": 2,
+            "pointSize": 5,
+            "scaleDistribution": { "type": "linear" },
+            "showPoints": "never",
+            "spanNulls": false,
+            "stacking": { "group": "A", "mode": "normal" },
+            "thresholdsStyle": { "mode": "off" }
+          },
+          "unit": "ops",
+          "min": 0
+        },
+        "overrides": []
+      },
+      "targets": [
+        {
+          "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+          "expr": "sum by (search_type) (rate(onyx_opensearch_search_total[5m]))",
+          "legendFormat": "{{ search_type }}",
+          "refId": "A"
+        }
+      ]
+    },
+    {
+      "title": "Concurrent Searches In Progress",
+      "description": "Number of OpenSearch searches currently in flight, broken down by search type. Summed across all instances.",
+      "type": "timeseries",
+      "gridPos": { "h": 10, "w": 12, "x": 0, "y": 20 },
+      "id": 5,
+      "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+      "fieldConfig": {
+        "defaults": {
+          "color": { "mode": "palette-classic" },
+          "custom": {
+            "axisBorderShow": false,
+            "axisCenteredZero": false,
+            "axisLabel": "searches",
+            "axisPlacement": "auto",
+            "drawStyle": "line",
+            "fillOpacity": 0,
+            "gradientMode": "none",
+            "lineInterpolation": "smooth",
+            "lineWidth": 2,
+            "pointSize": 5,
+            "scaleDistribution": { "type": "linear" },
+            "showPoints": "never",
+            "spanNulls": false,
+            "stacking": { "group": "A", "mode": "normal" },
+            "thresholdsStyle": { "mode": "off" }
+          },
+          "min": 0
+        },
+        "overrides": []
+      },
+      "targets": [
+        {
+          "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+          "expr": "sum by (search_type) (onyx_opensearch_searches_in_progress)",
+          "legendFormat": "{{ search_type }}",
+          "refId": "A"
+        }
+      ]
+    },
+    {
+      "title": "Client vs Server Latency Overhead (P50)",
+      "description": "Difference between client-side and server-side P50 latency. Reveals network, serialization, and untracked OpenSearch overhead.",
+      "type": "timeseries",
+      "gridPos": { "h": 10, "w": 12, "x": 12, "y": 20 },
+      "id": 6,
+      "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+      "fieldConfig": {
+        "defaults": {
+          "color": { "mode": "palette-classic" },
+          "custom": {
+            "axisBorderShow": false,
+            "axisCenteredZero": false,
+            "axisLabel": "seconds",
+            "axisPlacement": "auto",
+            "drawStyle": "line",
+            "fillOpacity": 0,
+            "gradientMode": "none",
+            "lineInterpolation": "smooth",
+            "lineWidth": 2,
+            "pointSize": 5,
+            "scaleDistribution": { "type": "linear" },
+            "showPoints": "never",
+            "spanNulls": false,
+            "stacking": { "group": "A", "mode": "none" },
+            "thresholdsStyle": { "mode": "off" }
+          },
+          "unit": "s",
+          "min": 0
+        },
+        "overrides": []
+      },
+      "targets": [
+        {
+          "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+          "expr": "histogram_quantile(0.5, sum by (le) (rate(onyx_opensearch_search_client_duration_seconds_bucket[5m]))) - histogram_quantile(0.5, sum by (le) (rate(onyx_opensearch_search_server_duration_seconds_bucket[5m])))",
+          "legendFormat": "Client - Server overhead (P50)",
+          "refId": "A"
+        },
+        {
+          "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+          "expr": "histogram_quantile(0.5, sum by (le) (rate(onyx_opensearch_search_client_duration_seconds_bucket[5m])))",
+          "legendFormat": "Client P50",
+          "refId": "B"
+        },
+        {
+          "datasource": { "type": "prometheus", "uid": "${DS_PROMETHEUS}" },
+          "expr": "histogram_quantile(0.5, sum by (le) (rate(onyx_opensearch_search_server_duration_seconds_bucket[5m])))",
+          "legendFormat": "Server P50",
+          "refId": "C"
+        }
+      ]
+    }
+  ],
+  "refresh": "5s",
+  "schemaVersion": 37,
+  "style": "dark",
+  "tags": ["onyx", "opensearch", "search", "latency"],
+  "templating": {
+    "list": [
+      {
+        "current": {
+          "text": "Prometheus",
+          "value": "prometheus"
+        },
+        "includeAll": false,
+        "name": "DS_PROMETHEUS",
+        "options": [],
+        "query": "prometheus",
+        "refresh": 1,
+        "type": "datasource"
+      }
+    ]
+  },
+  "time": { "from": "now-60m", "to": "now" },
+  "timepicker": {
+    "refresh_intervals": ["5s", "10s", "30s", "1m"]
+  },
+  "timezone": "",
+  "title": "Onyx OpenSearch Search Latency",
+  "uid": "onyx-opensearch-search-latency",
+  "version": 0,
+  "weekStart": ""
+}

deployment/helm/charts/onyx/dashboards/redis-queues.json

@@ -0,0 +1,606 @@
+{
+  "id": null,
+  "annotations": {
+    "list": [
+      {
+        "builtIn": 1,
+        "datasource": {
+          "type": "grafana",
+          "uid": "-- Grafana --"
+        },
+        "enable": true,
+        "hide": true,
+        "iconColor": "rgba(0, 211, 255, 1)",
+        "name": "Annotations & Alerts",
+        "type": "dashboard"
+      }
+    ]
+  },
+  "editable": true,
+  "fiscalYearStartMonth": 0,
+  "graphTooltip": 0,
+  "links": [],
+  "panels": [
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "palette-classic"
+          },
+          "custom": {
+            "axisBorderShow": false,
+            "axisCenteredZero": false,
+            "axisColorMode": "text",
+            "axisLabel": "",
+            "axisPlacement": "auto",
+            "barAlignment": 0,
+            "drawStyle": "line",
+            "fillOpacity": 18,
+            "gradientMode": "none",
+            "hideFrom": {
+              "legend": false,
+              "tooltip": false,
+              "viz": false
+            },
+            "insertNulls": false,
+            "lineInterpolation": "linear",
+            "lineWidth": 2,
+            "pointSize": 4,
+            "scaleDistribution": {
+              "type": "linear"
+            },
+            "showPoints": "never",
+            "spanNulls": false,
+            "stacking": {
+              "group": "A",
+              "mode": "none"
+            },
+            "thresholdsStyle": {
+              "mode": "off"
+            }
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "orange",
+                "value": 10
+              },
+              {
+                "color": "red",
+                "value": 50
+              }
+            ]
+          }
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 10,
+        "w": 24,
+        "x": 0,
+        "y": 0
+      },
+      "id": 1,
+      "options": {
+        "legend": {
+          "calcs": ["lastNotNull", "max"],
+          "displayMode": "table",
+          "placement": "right",
+          "showLegend": true
+        },
+        "tooltip": {
+          "mode": "multi",
+          "sort": "desc"
+        }
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "onyx_celery_queue_depth{queue=~\"$queue\"}",
+          "legendFormat": "{{queue}}",
+          "range": true,
+          "refId": "A"
+        }
+      ],
+      "title": "Queue Depth by Queue",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "orange",
+                "value": 20
+              },
+              {
+                "color": "red",
+                "value": 100
+              }
+            ]
+          }
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 4,
+        "w": 6,
+        "x": 0,
+        "y": 10
+      },
+      "id": 2,
+      "options": {
+        "colorMode": "background",
+        "graphMode": "area",
+        "justifyMode": "auto",
+        "orientation": "auto",
+        "reduceOptions": {
+          "calcs": ["lastNotNull"],
+          "fields": "",
+          "values": false
+        },
+        "textMode": "auto"
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "sum(onyx_celery_queue_depth)",
+          "refId": "A"
+        }
+      ],
+      "title": "Total Queued Tasks",
+      "type": "stat"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "orange",
+                "value": 20
+              },
+              {
+                "color": "red",
+                "value": 100
+              }
+            ]
+          }
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 4,
+        "w": 6,
+        "x": 6,
+        "y": 10
+      },
+      "id": 3,
+      "options": {
+        "colorMode": "background",
+        "graphMode": "area",
+        "justifyMode": "auto",
+        "orientation": "auto",
+        "reduceOptions": {
+          "calcs": ["lastNotNull"],
+          "fields": "",
+          "values": false
+        },
+        "textMode": "auto"
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "onyx_celery_unacked_tasks",
+          "refId": "A"
+        }
+      ],
+      "title": "Unacked Tasks",
+      "type": "stat"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "orange",
+                "value": 10
+              },
+              {
+                "color": "red",
+                "value": 50
+              }
+            ]
+          }
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 4,
+        "w": 6,
+        "x": 12,
+        "y": 10
+      },
+      "id": 4,
+      "options": {
+        "colorMode": "background",
+        "graphMode": "none",
+        "justifyMode": "center",
+        "orientation": "auto",
+        "reduceOptions": {
+          "calcs": ["lastNotNull"],
+          "fields": "",
+          "values": false
+        },
+        "textMode": "auto"
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "onyx_celery_queue_depth{queue=\"docprocessing\"}",
+          "refId": "A"
+        }
+      ],
+      "title": "Docprocessing Queue",
+      "type": "stat"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "orange",
+                "value": 10
+              },
+              {
+                "color": "red",
+                "value": 50
+              }
+            ]
+          }
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 4,
+        "w": 6,
+        "x": 18,
+        "y": 10
+      },
+      "id": 5,
+      "options": {
+        "colorMode": "background",
+        "graphMode": "none",
+        "justifyMode": "center",
+        "orientation": "auto",
+        "reduceOptions": {
+          "calcs": ["lastNotNull"],
+          "fields": "",
+          "values": false
+        },
+        "textMode": "auto"
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "onyx_celery_queue_depth{queue=\"connector_doc_fetching\"}",
+          "refId": "A"
+        }
+      ],
+      "title": "Docfetching Queue",
+      "type": "stat"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "palette-classic"
+          },
+          "custom": {
+            "axisBorderShow": false,
+            "axisCenteredZero": false,
+            "axisColorMode": "text",
+            "axisLabel": "",
+            "axisPlacement": "auto",
+            "barAlignment": 0,
+            "drawStyle": "bars",
+            "fillOpacity": 80,
+            "gradientMode": "none",
+            "hideFrom": {
+              "legend": false,
+              "tooltip": false,
+              "viz": false
+            },
+            "lineWidth": 1,
+            "scaleDistribution": {
+              "type": "linear"
+            },
+            "showPoints": "never",
+            "stacking": {
+              "group": "A",
+              "mode": "none"
+            }
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "orange",
+                "value": 10
+              },
+              {
+                "color": "red",
+                "value": 50
+              }
+            ]
+          }
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 10,
+        "w": 12,
+        "x": 0,
+        "y": 14
+      },
+      "id": 6,
+      "options": {
+        "legend": {
+          "calcs": ["lastNotNull"],
+          "displayMode": "list",
+          "placement": "bottom",
+          "showLegend": false
+        },
+        "tooltip": {
+          "mode": "single",
+          "sort": "none"
+        }
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "topk(10, onyx_celery_queue_depth)",
+          "legendFormat": "{{queue}}",
+          "range": true,
+          "refId": "A"
+        }
+      ],
+      "title": "Top 10 Queue Backlogs",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "custom": {
+            "align": "auto",
+            "cellOptions": {
+              "type": "auto"
+            },
+            "inspect": false
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "orange",
+                "value": 10
+              },
+              {
+                "color": "red",
+                "value": 50
+              }
+            ]
+          }
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 10,
+        "w": 12,
+        "x": 12,
+        "y": 14
+      },
+      "id": 7,
+      "options": {
+        "cellHeight": "sm",
+        "footer": {
+          "countRows": false,
+          "fields": "",
+          "reducer": ["sum"],
+          "show": false
+        },
+        "showHeader": true,
+        "sortBy": [
+          {
+            "desc": true,
+            "displayName": "Value"
+          }
+        ]
+      },
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "expr": "sort_desc(onyx_celery_queue_depth)",
+          "format": "table",
+          "instant": true,
+          "refId": "A"
+        }
+      ],
+      "title": "Current Queue Depth",
+      "transformations": [
+        {
+          "id": "labelsToFields",
+          "options": {
+            "mode": "columns"
+          }
+        }
+      ],
+      "type": "table"
+    }
+  ],
+  "refresh": "30s",
+  "schemaVersion": 39,
+  "style": "dark",
+  "tags": ["onyx", "redis", "celery"],
+  "templating": {
+    "list": [
+      {
+        "current": {
+          "selected": true,
+          "text": "Prometheus",
+          "value": "Prometheus"
+        },
+        "hide": 0,
+        "includeAll": false,
+        "label": "Datasource",
+        "name": "DS_PROMETHEUS",
+        "options": [],
+        "query": "prometheus",
+        "refresh": 1,
+        "regex": "",
+        "type": "datasource"
+      },
+      {
+        "allValue": ".*",
+        "current": {
+          "selected": true,
+          "text": "All",
+          "value": ".*"
+        },
+        "datasource": {
+          "type": "prometheus",
+          "uid": "${DS_PROMETHEUS}"
+        },
+        "definition": "label_values(onyx_celery_queue_depth, queue)",
+        "hide": 0,
+        "includeAll": true,
+        "label": "Queue",
+        "multi": true,
+        "name": "queue",
+        "options": [],
+        "query": {
+          "query": "label_values(onyx_celery_queue_depth, queue)",
+          "refId": "StandardVariableQuery"
+        },
+        "refresh": 2,
+        "regex": "",
+        "sort": 1,
+        "type": "query"
+      }
+    ]
+  },
+  "time": {
+    "from": "now-6h",
+    "to": "now"
+  },
+  "timepicker": {},
+  "timezone": "",
+  "title": "Onyx Redis Queues",
+  "uid": "onyx-redis-queues",
+  "version": 1,
+  "weekStart": ""
+}

deployment/helm/charts/onyx/templates/api-servicemonitor.yaml

@@ -0,0 +1,23 @@
+{{- if .Values.monitoring.serviceMonitors.enabled }}
+apiVersion: monitoring.coreos.com/v1
+kind: ServiceMonitor
+metadata:
+  name: {{ include "onyx.fullname" . }}-api
+  labels:
+    {{- include "onyx.labels" . | nindent 4 }}
+    {{- with .Values.monitoring.serviceMonitors.labels }}
+    {{- toYaml . | nindent 4 }}
+    {{- end }}
+spec:
+  namespaceSelector:
+    matchNames:
+      - {{ .Release.Namespace }}
+  selector:
+    matchLabels:
+      app: {{ .Values.api.deploymentLabels.app }}
+  endpoints:
+    - port: api-server-port
+      path: /metrics
+      interval: 30s
+      scrapeTimeout: 10s
+{{- end }}

deployment/helm/charts/onyx/templates/celery-worker-servicemonitors.yaml

@@ -74,4 +74,29 @@ spec:
       interval: 30s
       scrapeTimeout: 10s
 {{- end }}
+{{- if gt (int .Values.celery_worker_heavy.replicaCount) 0 }}
+---
+apiVersion: monitoring.coreos.com/v1
+kind: ServiceMonitor
+metadata:
+  name: {{ include "onyx.fullname" . }}-celery-worker-heavy
+  labels:
+    {{- include "onyx.labels" . | nindent 4 }}
+    {{- with .Values.monitoring.serviceMonitors.labels }}
+    {{- toYaml . | nindent 4 }}
+    {{- end }}
+spec:
+  namespaceSelector:
+    matchNames:
+      - {{ .Release.Namespace }}
+  selector:
+    matchLabels:
+      app: {{ .Values.celery_worker_heavy.deploymentLabels.app }}
+      metrics: "true"
+  endpoints:
+    - port: metrics
+      path: /metrics
+      interval: 30s
+      scrapeTimeout: 10s
+{{- end }}
 {{- end }}

deployment/helm/charts/onyx/templates/grafana-dashboards.yaml

@@ -12,4 +12,30 @@ metadata:
 data:
   onyx-indexing-pipeline.json: |
     {{- .Files.Get "dashboards/indexing-pipeline.json" | nindent 4 }}
+---
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: {{ include "onyx.fullname" . }}-opensearch-search-latency-dashboard
+  labels:
+    {{- include "onyx.labels" . | nindent 4 }}
+    grafana_dashboard: "1"
+  annotations:
+    grafana_folder: "Onyx"
+data:
+  onyx-opensearch-search-latency.json: |
+    {{- .Files.Get "dashboards/opensearch-search-latency.json" | nindent 4 }}
+---
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: {{ include "onyx.fullname" . }}-redis-queues-dashboard
+  labels:
+    {{- include "onyx.labels" . | nindent 4 }}
+    grafana_dashboard: "1"
+  annotations:
+    grafana_folder: "Onyx"
+data:
+  onyx-redis-queues.json: |
+    {{- .Files.Get "dashboards/redis-queues.json" | nindent 4 }}
 {{- end }}

deployment/helm/charts/onyx/values.yaml

@@ -264,7 +264,7 @@ monitoring:
       # The sidecar must be configured with label selector: grafana_dashboard=1
       enabled: false
   serviceMonitors:
-    # -- Set to true to deploy ServiceMonitor resources for Celery worker metrics endpoints.
+    # -- Set to true to deploy ServiceMonitor resources for API server and Celery worker metrics endpoints.
     # Requires the Prometheus Operator CRDs (included in kube-prometheus-stack).
     # Use `labels` to match your Prometheus CR's serviceMonitorSelector (e.g. release: onyx-monitoring).
     enabled: false

docker-bake.hcl

@@ -22,6 +22,10 @@ variable "CLI_REPOSITORY" {
   default = "onyxdotapp/onyx-cli"
 }
 
+variable "DEVCONTAINER_REPOSITORY" {
+  default = "onyxdotapp/onyx-devcontainer"
+}
+
 variable "TAG" {
   default = "latest"
 }
@@ -90,3 +94,16 @@ target "cli" {
 
   tags      = ["${CLI_REPOSITORY}:${TAG}"]
 }
+
+target "devcontainer" {
+  context    = ".devcontainer"
+  dockerfile = "Dockerfile"
+
+  cache-from = [
+    "type=registry,ref=${DEVCONTAINER_REPOSITORY}:latest",
+    "type=registry,ref=${DEVCONTAINER_REPOSITORY}:edge",
+  ]
+  cache-to   = ["type=inline"]
+
+  tags      = ["${DEVCONTAINER_REPOSITORY}:${TAG}"]
+}

pyproject.toml

@@ -148,7 +148,7 @@ dev = [
     "matplotlib==3.10.8",
     "mypy-extensions==1.0.0",
     "mypy==1.13.0",
-    "onyx-devtools==0.7.3",
+    "onyx-devtools==0.7.5",
     "openapi-generator-cli==7.17.0",
     "pandas-stubs~=2.3.3",
     "pre-commit==3.2.2",

tools/ods/README.md

@@ -244,6 +244,54 @@ ods web lint
 ods web test --watch
 ```
 
+### `dev` - Devcontainer Management
+
+Manage the Onyx devcontainer. Also available as `ods dc`.
+
+Requires the [devcontainer CLI](https://github.com/devcontainers/cli) (`npm install -g @devcontainers/cli`).
+
+```shell
+ods dev <subcommand>
+```
+
+**Subcommands:**
+
+- `up` - Start the devcontainer (pulls the image if needed)
+- `into` - Open a zsh shell inside the running devcontainer
+- `exec` - Run an arbitrary command inside the devcontainer
+- `restart` - Remove and recreate the devcontainer
+- `rebuild` - Pull the latest published image and recreate
+- `stop` - Stop the running devcontainer
+
+The devcontainer image is published to `onyxdotapp/onyx-devcontainer` and
+referenced by tag in `.devcontainer/devcontainer.json` — no local build needed.
+
+**Examples:**
+
+```shell
+# Start the devcontainer
+ods dev up
+
+# Open a shell
+ods dev into
+
+# Run a command
+ods dev exec -- npm test
+
+# Restart the container
+ods dev restart
+
+# Pull latest image and recreate
+ods dev rebuild
+
+# Stop the container
+ods dev stop
+
+# Same commands work with the dc alias
+ods dc up
+ods dc into
+```
+
 ### `db` - Database Administration
 
 Manage PostgreSQL database dumps, restores, and migrations.

tools/ods/cmd/dev.go

@@ -0,0 +1,34 @@
+package cmd
+
+import (
+	"github.com/spf13/cobra"
+)
+
+// NewDevCommand creates the parent dev command for devcontainer operations.
+func NewDevCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:     "dev",
+		Aliases: []string{"dc"},
+		Short:   "Manage the devcontainer",
+		Long: `Manage the Onyx devcontainer.
+
+Wraps the devcontainer CLI with workspace-aware defaults.
+
+Commands:
+  up        Start the devcontainer
+  into      Open a shell inside the running devcontainer
+  exec      Run a command inside the devcontainer
+  restart   Remove and recreate the devcontainer
+  rebuild   Pull the latest image and recreate
+  stop      Stop the running devcontainer`,
+	}
+
+	cmd.AddCommand(newDevUpCommand())
+	cmd.AddCommand(newDevIntoCommand())
+	cmd.AddCommand(newDevExecCommand())
+	cmd.AddCommand(newDevRestartCommand())
+	cmd.AddCommand(newDevRebuildCommand())
+	cmd.AddCommand(newDevStopCommand())
+
+	return cmd
+}

tools/ods/cmd/dev_exec.go

@@ -0,0 +1,29 @@
+package cmd
+
+import (
+	"github.com/spf13/cobra"
+)
+
+func newDevExecCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "exec [--] <command> [args...]",
+		Short: "Run a command inside the devcontainer",
+		Long: `Run an arbitrary command inside the running devcontainer.
+All arguments are treated as positional (flags like -it are passed through).
+
+Examples:
+  ods dev exec npm test
+  ods dev exec -- ls -la
+  ods dev exec -it echo hello`,
+		Args:               cobra.MinimumNArgs(1),
+		DisableFlagParsing: true,
+		Run: func(cmd *cobra.Command, args []string) {
+			if len(args) > 0 && args[0] == "--" {
+				args = args[1:]
+			}
+			runDevExec(args)
+		},
+	}
+
+	return cmd
+}

tools/ods/cmd/dev_into.go

@@ -0,0 +1,53 @@
+package cmd
+
+import (
+	"os"
+	"os/exec"
+
+	log "github.com/sirupsen/logrus"
+	"github.com/spf13/cobra"
+
+	"github.com/onyx-dot-app/onyx/tools/ods/internal/paths"
+)
+
+func newDevIntoCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "into",
+		Short: "Open a shell inside the running devcontainer",
+		Long: `Open an interactive zsh shell inside the running devcontainer.
+
+Examples:
+  ods dev into`,
+		Run: func(cmd *cobra.Command, args []string) {
+			runDevExec([]string{"zsh"})
+		},
+	}
+
+	return cmd
+}
+
+// runDevExec executes "devcontainer exec --workspace-folder <root> <command...>".
+func runDevExec(command []string) {
+	checkDevcontainerCLI()
+	ensureDockerSock()
+	ensureRemoteUser()
+
+	root, err := paths.GitRoot()
+	if err != nil {
+		log.Fatalf("Failed to find git root: %v", err)
+	}
+
+	args := []string{"exec", "--workspace-folder", root}
+	args = append(args, command...)
+
+	log.Debugf("Running: devcontainer %v", args)
+
+	c := exec.Command("devcontainer", args...)
+	c.Stdout = os.Stdout
+	c.Stderr = os.Stderr
+	c.Stdin = os.Stdin
+
+	if err := c.Run(); err != nil {
+		log.Fatalf("devcontainer exec failed: %v", err)
+	}
+}

tools/ods/cmd/dev_rebuild.go

@@ -0,0 +1,41 @@
+package cmd
+
+import (
+	"os"
+	"os/exec"
+
+	log "github.com/sirupsen/logrus"
+	"github.com/spf13/cobra"
+)
+
+func newDevRebuildCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "rebuild",
+		Short: "Pull the latest devcontainer image and recreate",
+		Long: `Pull the latest devcontainer image and recreate the container.
+
+Use after the published image has been updated or after changing devcontainer.json.
+
+Examples:
+  ods dev rebuild`,
+		Run: func(cmd *cobra.Command, args []string) {
+			runDevRebuild()
+		},
+	}
+
+	return cmd
+}
+
+func runDevRebuild() {
+	image := devcontainerImage()
+
+	log.Infof("Pulling %s...", image)
+	pull := exec.Command("docker", "pull", image)
+	pull.Stdout = os.Stdout
+	pull.Stderr = os.Stderr
+	if err := pull.Run(); err != nil {
+		log.Warnf("Failed to pull image (continuing with local copy): %v", err)
+	}
+
+	runDevcontainer("up", []string{"--remove-existing-container"})
+}

tools/ods/cmd/dev_restart.go

@@ -0,0 +1,23 @@
+package cmd
+
+import (
+	"github.com/spf13/cobra"
+)
+
+func newDevRestartCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "restart",
+		Short: "Remove and recreate the devcontainer",
+		Long: `Remove the existing devcontainer and recreate it.
+
+Uses the cached image — for a full image rebuild, use "ods dev rebuild".
+
+Examples:
+  ods dev restart`,
+		Run: func(cmd *cobra.Command, args []string) {
+			runDevcontainer("up", []string{"--remove-existing-container"})
+		},
+	}
+
+	return cmd
+}

tools/ods/cmd/dev_stop.go

@@ -0,0 +1,56 @@
+package cmd
+
+import (
+	"os/exec"
+	"strings"
+
+	log "github.com/sirupsen/logrus"
+	"github.com/spf13/cobra"
+
+	"github.com/onyx-dot-app/onyx/tools/ods/internal/paths"
+)
+
+func newDevStopCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "stop",
+		Short: "Stop the running devcontainer",
+		Long: `Stop the running devcontainer.
+
+Examples:
+  ods dev stop`,
+		Run: func(cmd *cobra.Command, args []string) {
+			runDevStop()
+		},
+	}
+
+	return cmd
+}
+
+func runDevStop() {
+	root, err := paths.GitRoot()
+	if err != nil {
+		log.Fatalf("Failed to find git root: %v", err)
+	}
+
+	// Find the container by the devcontainer label
+	out, err := exec.Command(
+		"docker", "ps", "-q",
+		"--filter", "label=devcontainer.local_folder="+root,
+	).Output()
+	if err != nil {
+		log.Fatalf("Failed to find devcontainer: %v", err)
+	}
+
+	containerID := strings.TrimSpace(string(out))
+	if containerID == "" {
+		log.Info("No running devcontainer found")
+		return
+	}
+
+	log.Infof("Stopping devcontainer %s...", containerID)
+	c := exec.Command("docker", "stop", containerID)
+	if err := c.Run(); err != nil {
+		log.Fatalf("Failed to stop devcontainer: %v", err)
+	}
+	log.Info("Devcontainer stopped")
+}

tools/ods/cmd/dev_up.go

@@ -0,0 +1,223 @@
+package cmd
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"runtime"
+	"strings"
+
+	log "github.com/sirupsen/logrus"
+	"github.com/spf13/cobra"
+
+	"github.com/onyx-dot-app/onyx/tools/ods/internal/paths"
+)
+
+func newDevUpCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "up",
+		Short: "Start the devcontainer",
+		Long: `Start the devcontainer, pulling the image if needed.
+
+Examples:
+  ods dev up`,
+		Run: func(cmd *cobra.Command, args []string) {
+			runDevcontainer("up", nil)
+		},
+	}
+
+	return cmd
+}
+
+// devcontainerImage reads the image field from .devcontainer/devcontainer.json.
+func devcontainerImage() string {
+	root, err := paths.GitRoot()
+	if err != nil {
+		log.Fatalf("Failed to find git root: %v", err)
+	}
+
+	data, err := os.ReadFile(filepath.Join(root, ".devcontainer", "devcontainer.json"))
+	if err != nil {
+		log.Fatalf("Failed to read devcontainer.json: %v", err)
+	}
+
+	var cfg struct {
+		Image string `json:"image"`
+	}
+	if err := json.Unmarshal(data, &cfg); err != nil {
+		log.Fatalf("Failed to parse devcontainer.json: %v", err)
+	}
+	if cfg.Image == "" {
+		log.Fatal("No image field in devcontainer.json")
+	}
+	return cfg.Image
+}
+
+// checkDevcontainerCLI ensures the devcontainer CLI is installed.
+func checkDevcontainerCLI() {
+	if _, err := exec.LookPath("devcontainer"); err != nil {
+		log.Fatal("devcontainer CLI is not installed. Install it with: npm install -g @devcontainers/cli")
+	}
+}
+
+// ensureDockerSock sets the DOCKER_SOCK environment variable if not already set.
+// devcontainer.json references ${localEnv:DOCKER_SOCK} for the socket mount.
+func ensureDockerSock() {
+	if os.Getenv("DOCKER_SOCK") != "" {
+		return
+	}
+
+	sock := detectDockerSock()
+	if err := os.Setenv("DOCKER_SOCK", sock); err != nil {
+		log.Fatalf("Failed to set DOCKER_SOCK: %v", err)
+	}
+}
+
+// detectDockerSock returns the path to the Docker socket on the host.
+func detectDockerSock() string {
+	// Prefer explicit DOCKER_HOST (strip unix:// prefix if present).
+	if dh := os.Getenv("DOCKER_HOST"); dh != "" {
+		const prefix = "unix://"
+		if len(dh) > len(prefix) && dh[:len(prefix)] == prefix {
+			return dh[len(prefix):]
+		}
+		// Only bare paths (starting with /) are valid socket paths.
+		// Non-unix schemes (e.g. tcp://) can't be bind-mounted.
+		if len(dh) > 0 && dh[0] == '/' {
+			return dh
+		}
+		log.Warnf("DOCKER_HOST=%q is not a unix socket path; falling back to local socket detection", dh)
+	}
+
+	// Linux rootless Docker: $XDG_RUNTIME_DIR/docker.sock
+	if runtime.GOOS == "linux" {
+		if xdg := os.Getenv("XDG_RUNTIME_DIR"); xdg != "" {
+			sock := filepath.Join(xdg, "docker.sock")
+			if _, err := os.Stat(sock); err == nil {
+				return sock
+			}
+		}
+	}
+
+	// macOS Docker Desktop: ~/.docker/run/docker.sock
+	if runtime.GOOS == "darwin" {
+		if home, err := os.UserHomeDir(); err == nil {
+			sock := filepath.Join(home, ".docker", "run", "docker.sock")
+			if _, err := os.Stat(sock); err == nil {
+				return sock
+			}
+		}
+	}
+
+	// Fallback: standard socket path (Linux with standard Docker, macOS symlink)
+	return "/var/run/docker.sock"
+}
+
+// worktreeGitMount returns a --mount flag value that makes a git worktree's
+// .git reference resolve inside the container. In a worktree, .git is a file
+// containing "gitdir: /path/to/main/.git/worktrees/<name>", so we need the
+// main repo's .git directory to exist at the same absolute host path inside
+// the container.
+//
+// Returns ("", false) when the workspace is not a worktree.
+func worktreeGitMount(root string) (string, bool) {
+	dotgit := filepath.Join(root, ".git")
+	info, err := os.Lstat(dotgit)
+	if err != nil || info.IsDir() {
+		return "", false // regular repo or no .git
+	}
+
+	// .git is a file — parse the gitdir path.
+	out, err := exec.Command("git", "-C", root, "rev-parse", "--git-common-dir").Output()
+	if err != nil {
+		log.Warnf("Failed to detect git common dir: %v", err)
+		return "", false
+	}
+	commonDir := strings.TrimSpace(string(out))
+
+	// Resolve to absolute path.
+	if !filepath.IsAbs(commonDir) {
+		commonDir = filepath.Join(root, commonDir)
+	}
+	commonDir, _ = filepath.EvalSymlinks(commonDir)
+
+	mount := fmt.Sprintf("type=bind,source=%s,target=%s", commonDir, commonDir)
+	log.Debugf("Worktree detected — mounting main .git: %s", commonDir)
+	return mount, true
+}
+
+// sshAgentMount returns a --mount flag value that forwards the host's SSH agent
+// socket into the container.  Returns ("", false) when SSH_AUTH_SOCK is unset or
+// the socket is not accessible.
+func sshAgentMount() (string, bool) {
+	sock := os.Getenv("SSH_AUTH_SOCK")
+	if sock == "" {
+		log.Debug("SSH_AUTH_SOCK not set — skipping SSH agent forwarding")
+		return "", false
+	}
+	if _, err := os.Stat(sock); err != nil {
+		log.Debugf("SSH_AUTH_SOCK=%s not accessible: %v", sock, err)
+		return "", false
+	}
+	mount := fmt.Sprintf("type=bind,source=%s,target=/tmp/ssh-agent.sock", sock)
+	log.Debugf("Forwarding SSH agent: %s", sock)
+	return mount, true
+}
+
+// ensureRemoteUser sets DEVCONTAINER_REMOTE_USER when rootless Docker is
+// detected.  Container root maps to the host user in rootless mode, so running
+// as root inside the container avoids the UID mismatch on new files.
+// Must be called after ensureDockerSock.
+func ensureRemoteUser() {
+	if os.Getenv("DEVCONTAINER_REMOTE_USER") != "" {
+		return
+	}
+
+	if runtime.GOOS == "linux" {
+		sock := os.Getenv("DOCKER_SOCK")
+		xdg := os.Getenv("XDG_RUNTIME_DIR")
+		// Heuristic: rootless Docker on Linux typically places its socket
+		// under $XDG_RUNTIME_DIR. If DOCKER_SOCK was set to a custom path
+		// outside XDG_RUNTIME_DIR, set DEVCONTAINER_REMOTE_USER=root manually.
+		if xdg != "" && strings.HasPrefix(sock, xdg) {
+			log.Debug("Rootless Docker detected — setting DEVCONTAINER_REMOTE_USER=root")
+			if err := os.Setenv("DEVCONTAINER_REMOTE_USER", "root"); err != nil {
+				log.Warnf("Failed to set DEVCONTAINER_REMOTE_USER: %v", err)
+			}
+		}
+	}
+}
+
+// runDevcontainer executes "devcontainer <action> --workspace-folder <root> [extraArgs...]".
+func runDevcontainer(action string, extraArgs []string) {
+	checkDevcontainerCLI()
+	ensureDockerSock()
+	ensureRemoteUser()
+
+	root, err := paths.GitRoot()
+	if err != nil {
+		log.Fatalf("Failed to find git root: %v", err)
+	}
+
+	args := []string{action, "--workspace-folder", root}
+	if mount, ok := worktreeGitMount(root); ok {
+		args = append(args, "--mount", mount)
+	}
+	if mount, ok := sshAgentMount(); ok {
+		args = append(args, "--mount", mount)
+	}
+	args = append(args, extraArgs...)
+
+	log.Debugf("Running: devcontainer %v", args)
+
+	c := exec.Command("devcontainer", args...)
+	c.Stdout = os.Stdout
+	c.Stderr = os.Stderr
+	c.Stdin = os.Stdin
+
+	if err := c.Run(); err != nil {
+		log.Fatalf("devcontainer %s failed: %v", action, err)
+	}
+}

tools/ods/cmd/root.go

@@ -53,6 +53,7 @@ func NewRootCommand() *cobra.Command {
 	cmd.AddCommand(NewRunCICommand())
 	cmd.AddCommand(NewScreenshotDiffCommand())
 	cmd.AddCommand(NewDesktopCommand())
+	cmd.AddCommand(NewDevCommand())
 	cmd.AddCommand(NewWebCommand())
 	cmd.AddCommand(NewLatestStableTagCommand())
 	cmd.AddCommand(NewWhoisCommand())

uv.lock

@@ -4511,7 +4511,7 @@ dev = [
     { name = "matplotlib", specifier = "==3.10.8" },
     { name = "mypy", specifier = "==1.13.0" },
     { name = "mypy-extensions", specifier = "==1.0.0" },
-    { name = "onyx-devtools", specifier = "==0.7.3" },
+    { name = "onyx-devtools", specifier = "==0.7.5" },
     { name = "openapi-generator-cli", specifier = "==7.17.0" },
     { name = "pandas-stubs", specifier = "~=2.3.3" },
     { name = "pre-commit", specifier = "==3.2.2" },
@@ -4554,19 +4554,19 @@ model-server = [
 
 [[package]]
 name = "onyx-devtools"
-version = "0.7.3"
+version = "0.7.5"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "fastapi" },
     { name = "openapi-generator-cli" },
 ]
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/72/64/c75be8ab325896cc64bccd0e1e139a03ce305bf05598967922d380fc4694/onyx_devtools-0.7.3-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:675e2fdbd8d291fba4b8a6dfcf2bc94c56d22d11f395a9f0d0c3c0e5b39d7f9b", size = 4220613, upload-time = "2026-04-09T00:04:36.624Z" },
-    { url = "https://files.pythonhosted.org/packages/ae/1f/589ff6bd446c4498f5bcdfd2a315709e91fc15edf5440c91ff64cbf0800f/onyx_devtools-0.7.3-py3-none-macosx_11_0_arm64.whl", hash = "sha256:bf3993de8ba02d6c2f1ab12b5b9b965e005040b37502f97db8a7d88d9b0cde4b", size = 3897867, upload-time = "2026-04-09T00:04:40.781Z" },
-    { url = "https://files.pythonhosted.org/packages/10/c0/53c9173eefc13218707282c5b99753960d039684994c3b3caf90ce286094/onyx_devtools-0.7.3-py3-none-manylinux_2_17_aarch64.whl", hash = "sha256:6138a94084bed05c674ad210a0bc4006c43bc4384e8eb54d469233de85c72bd7", size = 3762408, upload-time = "2026-04-09T00:04:41.592Z" },
-    { url = "https://files.pythonhosted.org/packages/d2/37/69fadb65112854a596d200f704da94b837817d4dd0f46cb4482dc0309c94/onyx_devtools-0.7.3-py3-none-manylinux_2_17_x86_64.whl", hash = "sha256:90dac91b0cdc32eb8861f6e83545009a34c439fd3c41fc7dd499acd0105b660e", size = 4184427, upload-time = "2026-04-09T00:04:41.525Z" },
-    { url = "https://files.pythonhosted.org/packages/bd/45/91c829ccb45f1a15e7c9641eccc6dd154adb540e03c7dee2a8f28cea24d0/onyx_devtools-0.7.3-py3-none-win_amd64.whl", hash = "sha256:abc68d70bec06e349481beec4b212de28a1a8b7ed6ef3b41daf7093ee10b44f3", size = 4299935, upload-time = "2026-04-09T00:04:40.262Z" },
-    { url = "https://files.pythonhosted.org/packages/cc/30/c5adcb8e3b46b71d8d92c3f9ee0c1d0bc5e2adc9f46e93931f21b36a3ee4/onyx_devtools-0.7.3-py3-none-win_arm64.whl", hash = "sha256:9e4411cadc5e81fabc9ed991402e3b4b40f02800681299c277b2142e5af0dcee", size = 3840228, upload-time = "2026-04-09T00:04:39.708Z" },
+    { url = "https://files.pythonhosted.org/packages/cb/f8/844e34f5126ae40fff0d012bba0b28f031f8871062759bb3789eae4f5e0a/onyx_devtools-0.7.5-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:b3cd434c722ae48a1f651748a9f094711b29d1a9f37fbbadef3144f2cdb0f16d", size = 4238900, upload-time = "2026-04-10T07:02:16.382Z" },
+    { url = "https://files.pythonhosted.org/packages/2d/97/d1db725f900b199fa3f7a7a7c9b51ae75d4b18755c924f00f06a7703e552/onyx_devtools-0.7.5-py3-none-macosx_11_0_arm64.whl", hash = "sha256:c50e3d76d4f8cc4faa6250e758d42f0249067f0e17bc82b99c6c00dd48114393", size = 3913672, upload-time = "2026-04-10T07:02:17.46Z" },
+    { url = "https://files.pythonhosted.org/packages/31/83/e11bedb0a1321b63c844a418be1990c172ed363c6ee612978c3a38df71f1/onyx_devtools-0.7.5-py3-none-manylinux_2_17_aarch64.whl", hash = "sha256:ec01aeaaa14854b0933bb85bbfc51184599d3dbf1c0097ff59c1c72db8222a5a", size = 3779585, upload-time = "2026-04-10T07:02:16.31Z" },
+    { url = "https://files.pythonhosted.org/packages/b3/85/128d25cd35c1adc436dcff9ab4f2c20cf29528d09415280c1230ff0ca993/onyx_devtools-0.7.5-py3-none-manylinux_2_17_x86_64.whl", hash = "sha256:586d50ecb6dcea95611135e4cd4529ebedd8ab84a41b1adf3be1280a48dc52af", size = 4201962, upload-time = "2026-04-10T07:02:14.466Z" },
+    { url = "https://files.pythonhosted.org/packages/99/5d/83c80f918b399fea998cd41bfe90bda733eda77e133ca4dc1e9ce18a9b4a/onyx_devtools-0.7.5-py3-none-win_amd64.whl", hash = "sha256:c45d80f0093ba738120b77c4c0bde13843e33d786ae8608eb10490f06183d89b", size = 4320088, upload-time = "2026-04-10T07:02:17.09Z" },
+    { url = "https://files.pythonhosted.org/packages/26/bf/b9c85cc61981bd71c0f1cbb50192763b11788a7c8636b1e01f750251c92c/onyx_devtools-0.7.5-py3-none-win_arm64.whl", hash = "sha256:9852a7cc29939371e016b794f2cffdb88680280d857d24c191c5188884416a3d", size = 3858839, upload-time = "2026-04-10T07:02:20.098Z" },
 ]
 
 [[package]]

web/lib/opal/src/logos/anthropic.tsx

@@ -10,7 +10,7 @@ const SvgAnthropic = ({ size, ...props }: IconProps) => (
   >
     <path
       d="M36.1779 9.78003H29.1432L41.9653 42.2095H49L36.1779 9.78003ZM15.8221 9.78003L3 42.2095H10.1844L12.8286 35.4243H26.2495L28.8438 42.2095H36.0282L23.2061 9.78003H15.8221ZM15.1236 29.3874L19.5141 18.0121L23.9046 29.3874H15.1236Z"
-      fill="currentColor"
+      fill="var(--text-05)"
     />
   </svg>
 );

web/lib/opal/src/logos/aws.tsx

@@ -12,7 +12,7 @@ const SvgAws = ({ size, ...props }: IconProps) => (
     <title>AWS</title>
     <path
       d="M14.6195 23.2934C14.6195 23.9333 14.7233 24.4522 14.8443 24.8326C14.9827 25.2131 15.1556 25.6282 15.3978 26.0778C15.4842 26.2162 15.5188 26.3546 15.5188 26.4756C15.5188 26.6486 15.4151 26.8215 15.1902 26.9945L14.1007 27.7208C13.945 27.8246 13.7894 27.8765 13.651 27.8765C13.4781 27.8765 13.3051 27.79 13.1322 27.6344C12.89 27.3749 12.6825 27.0982 12.5096 26.8215C12.3366 26.5275 12.1637 26.1989 11.9734 25.8011C10.6245 27.3922 8.92958 28.1878 6.88881 28.1878C5.43606 28.1878 4.27731 27.7727 3.42988 26.9426C2.58244 26.1124 2.15007 25.0056 2.15007 23.622C2.15007 22.152 2.66891 20.9586 3.72389 20.0593C4.77886 19.16 6.17973 18.7103 7.96108 18.7103C8.54909 18.7103 9.15441 18.7622 9.79431 18.8487C10.4342 18.9352 11.0914 19.0735 11.7832 19.2292V17.9667C11.7832 16.6523 11.5065 15.7356 10.9703 15.1995C10.4169 14.6634 9.483 14.404 8.15132 14.404C7.546 14.404 6.9234 14.4731 6.28349 14.6288C5.64359 14.7844 5.02098 14.9747 4.41567 15.2168C4.13896 15.3379 3.93142 15.407 3.81036 15.4416C3.6893 15.4762 3.60282 15.4935 3.53364 15.4935C3.29152 15.4935 3.17046 15.3206 3.17046 14.9574V14.1099C3.17046 13.8332 3.20505 13.6257 3.29152 13.5046C3.37799 13.3836 3.53364 13.2625 3.77577 13.1414C4.38108 12.8301 5.10746 12.5707 5.9549 12.3632C6.80233 12.1384 7.70165 12.0346 8.65286 12.0346C10.7109 12.0346 12.2156 12.5015 13.1841 13.4355C14.1353 14.3694 14.6195 15.7875 14.6195 17.6899V23.2934ZM7.63248 25.9222C8.2032 25.9222 8.79122 25.8184 9.41383 25.6109C10.0364 25.4034 10.5899 25.0229 11.0568 24.504C11.3335 24.1754 11.5411 23.8122 11.6448 23.3972C11.7486 22.9821 11.8178 22.4806 11.8178 21.8925V21.1662C11.3162 21.0451 10.7801 20.9413 10.2267 20.8722C9.67325 20.803 9.13711 20.7684 8.60098 20.7684C7.44224 20.7684 6.5948 20.9932 6.02407 21.4602C5.45335 21.9271 5.17664 22.5843 5.17664 23.4491C5.17664 24.2619 5.38417 24.8672 5.81654 25.2823C6.23161 25.7147 6.83692 25.9222 7.63248 25.9222ZM21.5201 27.79C21.2088 27.79 21.0012 27.7381 20.8629 27.6171C20.7245 27.5133 20.6035 27.2712 20.4997 26.9426L16.4355 13.5738C16.3317 13.2279 16.2798 13.0031 16.2798 12.882C16.2798 12.6053 16.4182 12.4497 16.6949 12.4497H18.3897C18.7183 12.4497 18.9432 12.5015 19.0642 12.6226C19.2026 12.7264 19.3064 12.9685 19.4101 13.2971L22.3156 24.7462L25.0136 13.2971C25.1001 12.9512 25.2038 12.7264 25.3422 12.6226C25.4806 12.5188 25.7227 12.4497 26.034 12.4497H27.4176C27.7462 12.4497 27.971 12.5015 28.1093 12.6226C28.2477 12.7264 28.3688 12.9685 28.4379 13.2971L31.1705 24.8845L34.1625 13.2971C34.2662 12.9512 34.3873 12.7264 34.5084 12.6226C34.6467 12.5188 34.8716 12.4497 35.1829 12.4497H36.7913C37.068 12.4497 37.2236 12.588 37.2236 12.882C37.2236 12.9685 37.2063 13.055 37.189 13.1587C37.1717 13.2625 37.1372 13.4009 37.068 13.5911L32.9 26.9599C32.7962 27.3058 32.6751 27.5306 32.5368 27.6344C32.3984 27.7381 32.1736 27.8073 31.8796 27.8073H30.3922C30.0636 27.8073 29.8388 27.7554 29.7004 27.6344C29.5621 27.5133 29.441 27.2885 29.3719 26.9426L26.6912 15.7875L24.0278 26.9253C23.9413 27.2712 23.8376 27.496 23.6992 27.6171C23.5609 27.7381 23.3187 27.79 23.0074 27.79H21.5201ZM43.7437 28.257C42.8444 28.257 41.9451 28.1532 41.0803 27.9457C40.2156 27.7381 39.5411 27.5133 39.0914 27.2539C38.8147 27.0982 38.6245 26.9253 38.5553 26.7696C38.4861 26.614 38.4515 26.441 38.4515 26.2854V25.4034C38.4515 25.0402 38.5899 24.8672 38.8493 24.8672C38.9531 24.8672 39.0569 24.8845 39.1606 24.9191C39.2644 24.9537 39.42 25.0229 39.593 25.0921C40.181 25.3515 40.8209 25.559 41.4954 25.6974C42.1872 25.8357 42.8617 25.9049 43.5535 25.9049C44.643 25.9049 45.4905 25.7147 46.0785 25.3342C46.6665 24.9537 46.9778 24.4003 46.9778 23.6912C46.9778 23.2069 46.8222 22.8092 46.5109 22.4806C46.1996 22.152 45.6115 21.858 44.7641 21.5812L42.2564 20.803C40.9939 20.4052 40.0599 19.8172 39.4892 19.0389C38.9185 18.278 38.6245 17.4305 38.6245 16.5312C38.6245 15.8048 38.7801 15.1649 39.0914 14.6115C39.4027 14.0581 39.8178 13.5738 40.3367 13.1933C40.8555 12.7956 41.4435 12.5015 42.1353 12.294C42.8271 12.0865 43.5535 12 44.3144 12C44.6949 12 45.0927 12.0173 45.4732 12.0692C45.871 12.1211 46.2341 12.1902 46.5973 12.2594C46.9432 12.3459 47.2718 12.4324 47.5831 12.5361C47.8944 12.6399 48.1366 12.7437 48.3095 12.8474C48.5516 12.9858 48.7246 13.1242 48.8283 13.2798C48.9321 13.4182 48.984 13.6084 48.984 13.8505V14.6634C48.984 15.0266 48.8456 15.2168 48.5862 15.2168C48.4479 15.2168 48.223 15.1476 47.929 15.0093C46.9432 14.5596 45.8364 14.3348 44.6084 14.3348C43.6227 14.3348 42.8444 14.4904 42.3083 14.819C41.7721 15.1476 41.4954 15.6492 41.4954 16.3583C41.4954 16.8425 41.6684 17.2576 42.0142 17.5862C42.3601 17.9148 43 18.2434 43.9167 18.5374L46.3725 19.3156C47.6177 19.7134 48.517 20.2668 49.0532 20.9759C49.5893 21.685 49.8487 22.4979 49.8487 23.3972C49.8487 24.1408 49.6931 24.8153 49.3991 25.4034C49.0878 25.9914 48.6727 26.5102 48.1366 26.9253C47.6004 27.3577 46.9605 27.669 46.2168 27.8938C45.4386 28.1359 44.6257 28.257 43.7437 28.257Z"
-      fill="#252F3E"
+      className="fill-[#252F3E] dark:fill-text-05"
     />
     <path
       fillRule="evenodd"

web/lib/opal/src/logos/cohere.tsx

@@ -0,0 +1,25 @@
+import type { IconProps } from "@opal/types";
+const SvgCohere = ({ size, ...props }: IconProps) => (
+  <svg
+    width={size}
+    height={size}
+    viewBox="0 0 52 52"
+    fill="none"
+    xmlns="http://www.w3.org/2000/svg"
+    {...props}
+  >
+    <path
+      d="M18.256 30.224C19.4293 30.224 21.776 30.1653 25.0613 28.816C28.8747 27.232 36.384 24.416 41.84 21.4827C45.6533 19.4293 47.296 16.7307 47.296 13.0933C47.296 8.10667 43.248 4 38.2027 4H17.0827C9.86667 4 4 9.86667 4 17.0827C4 24.2987 9.51467 30.224 18.256 30.224Z"
+      fill="#39594D"
+    />
+    <path
+      d="M21.8347 39.2C21.8347 35.68 23.9467 32.4533 27.232 31.104L33.8613 28.3467C40.608 25.5893 48 30.5173 48 37.792C48 43.424 43.424 48 37.792 48H30.576C25.7653 48 21.8347 44.0693 21.8347 39.2Z"
+      fill="#D18EE2"
+    />
+    <path
+      d="M11.568 31.9253C7.40267 31.9253 4 35.328 4 39.4933V40.4907C4 44.5973 7.40267 48 11.568 48C15.7333 48 19.136 44.5973 19.136 40.432V39.4347C19.0773 35.328 15.7333 31.9253 11.568 31.9253Z"
+      fill="#FF7759"
+    />
+  </svg>
+);
+export default SvgCohere;

web/lib/opal/src/logos/index.ts

@@ -3,6 +3,7 @@ export { default as SvgAws } from "@opal/logos/aws";
 export { default as SvgAzure } from "@opal/logos/azure";
 export { default as SvgBifrost } from "@opal/logos/bifrost";
 export { default as SvgClaude } from "@opal/logos/claude";
+export { default as SvgCohere } from "@opal/logos/cohere";
 export { default as SvgDeepseek } from "@opal/logos/deepseek";
 export { default as SvgDiscord } from "@opal/logos/discord";
 export { default as SvgGemini } from "@opal/logos/gemini";
@@ -11,6 +12,7 @@ export { default as SvgLitellm } from "@opal/logos/litellm";
 export { default as SvgLmStudio } from "@opal/logos/lm-studio";
 export { default as SvgMicrosoft } from "@opal/logos/microsoft";
 export { default as SvgMistral } from "@opal/logos/mistral";
+export { default as SvgNomic } from "@opal/logos/nomic";
 export { default as SvgOllama } from "@opal/logos/ollama";
 export { default as SvgOnyxLogo } from "@opal/logos/onyx-logo";
 export { default as SvgOnyxLogoTyped } from "@opal/logos/onyx-logo-typed";
@@ -19,3 +21,4 @@ export { default as SvgOpenai } from "@opal/logos/openai";
 export { default as SvgOpenrouter } from "@opal/logos/openrouter";
 export { default as SvgQwen } from "@opal/logos/qwen";
 export { default as SvgSlack } from "@opal/logos/slack";
+export { default as SvgVoyage } from "@opal/logos/voyage";

web/lib/opal/src/logos/nomic.tsx

@@ -0,0 +1,21 @@
+import type { IconProps } from "@opal/types";
+const SvgNomic = ({ size, ...props }: IconProps) => (
+  <svg
+    width={size}
+    height={size}
+    viewBox="0 0 52 52"
+    fill="none"
+    xmlns="http://www.w3.org/2000/svg"
+    {...props}
+  >
+    <path
+      d="M35.858 6.31995H46V45.6709H35.6146C32.0852 36.8676 25.1481 27.7804 15.7363 24.8189V6.31995H25.4726C26.5274 12.7296 30.1618 18.3744 35.858 21.6546V6.31995Z"
+      fill="var(--text-05)"
+    />
+    <path
+      d="M15.7363 24.8189V45.6709H6L6 30.0927C9.05968 27.6167 11.9635 25.8737 15.7363 24.8189Z"
+      fill="var(--text-05)"
+    />
+  </svg>
+);
+export default SvgNomic;

web/lib/opal/src/logos/openai.tsx

@@ -1,7 +1,7 @@
 import React from "react";
 import type { IconProps } from "@opal/types";
 
-const SvgOpenAI = ({ size, ...props }: IconProps) => {
+const SvgOpenai = ({ size, ...props }: IconProps) => {
   const clipId = React.useId();
   return (
     <svg
@@ -15,16 +15,16 @@ const SvgOpenAI = ({ size, ...props }: IconProps) => {
       <g clipPath={`url(#${clipId})`}>
         <path
           d="M6.27989 5.99136V4.58828C6.27989 4.4701 6.32383 4.38143 6.42625 4.32242L9.22206 2.69783C9.60266 2.4763 10.0564 2.37296 10.5247 2.37296C12.2813 2.37296 13.3937 3.74654 13.3937 5.20864C13.3937 5.31199 13.3937 5.43016 13.379 5.54833L10.4808 3.83506C10.3052 3.73172 10.1295 3.73172 9.95386 3.83506L6.27989 5.99136ZM12.8082 11.4561V8.10334C12.8082 7.89651 12.7203 7.74883 12.5447 7.64548L8.87071 5.48918L10.071 4.79498C10.1734 4.73597 10.2613 4.73597 10.3637 4.79498L13.1595 6.41959C13.9647 6.89226 14.5061 7.89651 14.5061 8.87124C14.5061 9.99365 13.8476 11.0277 12.8082 11.4561ZM5.41629 8.50218L4.21603 7.7933C4.11361 7.73429 4.06967 7.64563 4.06967 7.52745V4.27824C4.06967 2.69797 5.26993 1.50157 6.89473 1.50157C7.50955 1.50157 8.08029 1.70841 8.56345 2.07761L5.67991 3.76136C5.5043 3.86471 5.41643 4.01239 5.41643 4.21923L5.41629 8.50218ZM7.99984 10.0086L6.27988 9.03389V6.96624L7.99984 5.99151L9.71963 6.96624V9.03389L7.99984 10.0086ZM9.10494 14.4985C8.49012 14.4985 7.91938 14.2917 7.43622 13.9226L10.3197 12.2387C10.4953 12.1354 10.5832 11.9878 10.5832 11.7809V7.4978L11.7982 8.20668C11.9006 8.2657 11.9445 8.35436 11.9445 8.47254V11.7218C11.9445 13.302 10.7296 14.4985 9.10494 14.4985ZM5.63583 11.205L2.84002 9.58041C2.03489 9.10771 1.4934 8.10348 1.4934 7.12875C1.4934 5.99151 2.16672 4.97244 3.20591 4.5441V7.91148C3.20591 8.11831 3.29379 8.26599 3.46939 8.36934L7.12882 10.5108L5.92856 11.205C5.82613 11.264 5.73825 11.264 5.63583 11.205ZM5.47491 13.6272C3.82088 13.6272 2.60592 12.3717 2.60592 10.821C2.60592 10.7028 2.62061 10.5846 2.63517 10.4665L5.51871 12.1502C5.69432 12.2535 5.87006 12.2535 6.04567 12.1502L9.71964 10.0088V11.4119C9.71964 11.53 9.67571 11.6186 9.57328 11.6777L6.77746 13.3023C6.39688 13.5238 5.94323 13.6272 5.47491 13.6272ZM9.10494 15.3846C10.8761 15.3846 12.3544 14.1145 12.6912 12.4307C14.3305 12.0024 15.3845 10.4516 15.3845 8.87139C15.3845 7.8375 14.9453 6.83326 14.1549 6.10955C14.2281 5.79937 14.2721 5.48918 14.2721 5.17914C14.2721 3.06718 12.5741 1.48677 10.6126 1.48677C10.2175 1.48677 9.83689 1.54578 9.4563 1.67878C8.79753 1.02891 7.88999 0.615387 6.89473 0.615387C5.12357 0.615387 3.64528 1.88548 3.30848 3.56923C1.66914 3.99756 0.615234 5.54834 0.615234 7.1286C0.615234 8.1625 1.05431 9.16673 1.84474 9.89044C1.77155 10.2006 1.72762 10.5108 1.72762 10.8209C1.72762 12.9328 3.42558 14.5132 5.38704 14.5132C5.78218 14.5132 6.16278 14.4542 6.54336 14.3213C7.20198 14.9711 8.10953 15.3846 9.10494 15.3846Z"
-          fill="currentColor"
+          fill="var(--text-05)"
         />
       </g>
       <defs>
         <clipPath id={clipId}>
-          <rect width="16" height="16" fill="white" />
+          <rect width="16" height="16" fill="var(--text-05)" />
         </clipPath>
       </defs>
     </svg>
   );
 };
 
-export default SvgOpenAI;
+export default SvgOpenai;

web/lib/opal/src/logos/openrouter.tsx

@@ -14,7 +14,7 @@ const SvgOpenrouter = ({ size, ...props }: IconProps) => (
       fillRule="evenodd"
       clipRule="evenodd"
       d="M33.6 0L48 8.19239V8.36602L33.6 16.4V12.2457L31.8202 12.1858C29.7043 12.1299 28.6014 12.1898 27.2887 12.4053C25.1628 12.7546 23.2168 13.5569 21.001 15.1035L16.6733 18.1071C16.1059 18.4962 15.6843 18.7776 15.3147 19.0151L14.2857 19.6577L13.4925 20.1247L14.2617 20.5837L15.3207 21.2583C16.2717 21.8849 17.6583 22.8469 20.7173 24.9823C22.9351 26.529 24.8791 27.3312 27.005 27.6805L27.6044 27.7703C28.991 27.9519 30.7029 27.9579 33.6 27.8362V23.6L48 31.7198V31.8934L33.6 40V36.284L31.9041 36.3279C29.1349 36.4117 27.6344 36.3319 25.6344 36.0046C22.2498 35.4458 19.1209 34.1566 15.8821 31.8954L11.5704 28.9019C11.0745 28.5603 10.5715 28.2289 10.0619 27.908L9.12887 27.3492C8.62495 27.0592 8.11878 26.7731 7.61039 26.491C5.81019 25.4912 1.12488 24.2658 0 24.2658V15.836C1.12687 15.822 6.09391 14.5946 7.89011 13.5928L9.92008 12.4353L10.7952 11.8884C11.6503 11.3296 12.9371 10.4396 16.1618 8.19039C19.4006 5.92925 22.5275 4.63803 25.9141 4.08123C28.2158 3.70204 29.9237 3.65614 33.6 3.80582V0Z"
-      fill="currentColor"
+      fill="var(--text-05)"
     />
   </svg>
 );

web/lib/opal/src/logos/voyage.tsx

@@ -0,0 +1,17 @@
+import type { IconProps } from "@opal/types";
+const SvgVoyage = ({ size, ...props }: IconProps) => (
+  <svg
+    width={size}
+    height={size}
+    viewBox="0 0 52 52"
+    fill="none"
+    xmlns="http://www.w3.org/2000/svg"
+    {...props}
+  >
+    <path
+      d="M14.1848 8V8.11C14.1408 8.24212 14.1139 8.37935 14.1048 8.51833C14.0865 8.70167 14.0782 8.865 14.0782 9.01C14.0782 9.575 14.1498 10.2017 14.2915 10.8933C14.4532 11.5683 14.7482 12.4133 15.1765 13.4333L27.0515 40.71L38.5248 13.65C38.7932 12.9767 39.0798 12.24 39.3832 11.4383C39.6865 10.6383 39.8382 9.82833 39.8382 9.00833C39.8444 8.70074 39.7901 8.39492 39.6782 8.10833V8H45.1732V8.11C44.8332 8.455 44.4232 9.07333 43.9398 9.96667C43.4565 10.8583 42.9298 11.9583 42.3582 13.27L26.9982 48H24.8532L10.2982 14.6083C9.95818 13.825 9.60151 13.07 9.22484 12.3417C8.86818 11.6133 8.52818 10.9583 8.20818 10.375C7.88484 9.775 7.59984 9.275 7.34984 8.875C7.19246 8.61074 7.02226 8.35432 6.83984 8.10667V8H14.1848Z"
+      className="fill-[#012E33] dark:fill-text-05"
+    />
+  </svg>
+);
+export default SvgVoyage;

web/src/app/admin/bots/[bot-id]/channels/SlackChannelConfigFormFields.tsx

@@ -17,6 +17,7 @@ import DocumentSetCard from "@/sections/cards/DocumentSetCard";
 import CollapsibleSection from "@/app/admin/agents/CollapsibleSection";
 import { StandardAnswerCategoryResponse } from "@/components/standardAnswers/getStandardAnswerCategoriesIfEE";
 import { StandardAnswerCategoryDropdownField } from "@/components/standardAnswers/StandardAnswerCategoryDropdown";
+import InputComboBox from "@/refresh-components/inputs/InputComboBox";
 import { RadioGroup } from "@/components/ui/radio-group";
 import { RadioGroupItemField } from "@/components/ui/RadioGroupItemField";
 import { AlertCircle } from "lucide-react";
@@ -126,6 +127,24 @@ export function SlackChannelConfigFormFields({
     return documentSets.filter((ds) => !documentSetContainsSync(ds));
   }, [documentSets]);
 
+  const searchAgentOptions = useMemo(
+    () =>
+      availableAgents.map((persona) => ({
+        label: persona.name,
+        value: String(persona.id),
+      })),
+    [availableAgents]
+  );
+
+  const nonSearchAgentOptions = useMemo(
+    () =>
+      nonSearchAgents.map((persona) => ({
+        label: persona.name,
+        value: String(persona.id),
+      })),
+    [nonSearchAgents]
+  );
+
   useEffect(() => {
     const invalidSelected = values.document_sets.filter((dsId: number) =>
       unselectableSets.some((us) => us.id === dsId)
@@ -355,12 +374,14 @@ export function SlackChannelConfigFormFields({
               </>
             </SubLabel>
 
-            <SelectorFormField
-              name="persona_id"
-              options={availableAgents.map((persona) => ({
-                name: persona.name,
-                value: persona.id,
-              }))}
+            <InputComboBox
+              placeholder="Search for an agent..."
+              value={String(values.persona_id ?? "")}
+              onValueChange={(val) =>
+                setFieldValue("persona_id", val ? Number(val) : null)
+              }
+              options={searchAgentOptions}
+              strict
             />
             {viewSyncEnabledAgents && syncEnabledAgents.length > 0 && (
               <div className="mt-4">
@@ -419,12 +440,14 @@ export function SlackChannelConfigFormFields({
               </>
             </SubLabel>
 
-            <SelectorFormField
-              name="persona_id"
-              options={nonSearchAgents.map((persona) => ({
-                name: persona.name,
-                value: persona.id,
-              }))}
+            <InputComboBox
+              placeholder="Search for an agent..."
+              value={String(values.persona_id ?? "")}
+              onValueChange={(val) =>
+                setFieldValue("persona_id", val ? Number(val) : null)
+              }
+              options={nonSearchAgentOptions}
+              strict
             />
           </div>
         )}

web/src/app/ee/admin/performance/query-history/QueryHistoryTable.tsx

@@ -61,7 +61,7 @@ function QueryHistoryTableRow({
       key={chatSessionMinimal.id}
       className="hover:bg-accent-background cursor-pointer relative select-none"
     >
-      <TableCell>
+      <TableCell className="max-w-xs">
         <Text className="whitespace-normal line-clamp-5">
           {chatSessionMinimal.first_user_message ||
             chatSessionMinimal.name ||

web/src/refresh-pages/AppPage.tsx

@@ -871,15 +871,20 @@ export default function AppPage({ firstMessage }: ChatPageProps) {
                         agent={liveAgent}
                         isDefaultAgent={isDefaultAgent}
                       />
-                      {liveAgent && !llmManager.isLoadingProviders && (
-                        <ModelSelector
-                          llmManager={llmManager}
-                          selectedModels={multiModel.selectedModels}
-                          onAdd={multiModel.addModel}
-                          onRemove={multiModel.removeModel}
-                          onReplace={multiModel.replaceModel}
-                        />
-                      )}
+                      {!isSearch &&
+                        !(
+                          state.phase === "idle" && state.appMode === "search"
+                        ) &&
+                        liveAgent &&
+                        !llmManager.isLoadingProviders && (
+                          <ModelSelector
+                            llmManager={llmManager}
+                            selectedModels={multiModel.selectedModels}
+                            onAdd={multiModel.addModel}
+                            onRemove={multiModel.removeModel}
+                            onReplace={multiModel.replaceModel}
+                          />
+                        )}
                     </Section>
                     <Spacer rem={1.5} />
                   </Fade>

web/src/sections/modals/llmConfig/BifrostModal.tsx

@@ -50,7 +50,7 @@ function BifrostModalInternals({
     const { models, error } = await fetchBifrostModels({
       api_base: formikProps.values.api_base,
       api_key: formikProps.values.api_key || undefined,
-      provider_name: LLMProviderName.BIFROST,
+      provider_name: existingLlmProvider?.name,
     });
     if (error) {
       throw new Error(error);

web/src/sections/modals/llmConfig/LiteLLMProxyModal.tsx

@@ -52,7 +52,7 @@ function LiteLLMProxyModalInternals({
     const { models, error } = await fetchLiteLLMProxyModels({
       api_base: formikProps.values.api_base,
       api_key: formikProps.values.api_key,
-      provider_name: LLMProviderName.LITELLM_PROXY,
+      provider_name: existingLlmProvider?.name,
     });
     if (error) {
       throw new Error(error);

web/src/sections/modals/llmConfig/OpenRouterModal.tsx

@@ -52,7 +52,7 @@ function OpenRouterModalInternals({
     const { models, error } = await fetchOpenRouterModels({
       api_base: formikProps.values.api_base,
       api_key: formikProps.values.api_key,
-      provider_name: LLMProviderName.OPENROUTER,
+      provider_name: existingLlmProvider?.name,
     });
     if (error) {
       throw new Error(error);