.

.github/workflows/helm-chart-releases.yml

@@ -47,8 +47,7 @@ jobs:
           done
 
       - name: Publish Helm charts to gh-pages
-        # NOTE: HEAD of https://github.com/stefanprodan/helm-gh-pages/pull/43
-        uses: stefanprodan/helm-gh-pages@ad32ad3b8720abfeaac83532fd1e9bdfca5bbe27 # zizmor: ignore[impostor-commit]
+        uses: stefanprodan/helm-gh-pages@0ad2bb377311d61ac04ad9eb6f252fb68e207260 # ratchet:stefanprodan/helm-gh-pages@v1.7.0
         with:
           token: ${{ secrets.GITHUB_TOKEN }}
           charts_dir: deployment/helm/charts

README.md

@@ -35,7 +35,7 @@ Onyx comes loaded with advanced features like Agents, Web Search, RAG, MCP, Deep
 > [!TIP]
 > Run Onyx with one command (or see deployment section below):
 > ```
-> curl -fsSL https://onyx.app/install_onyx.sh | bash
+> curl -fsSL https://raw.githubusercontent.com/onyx-dot-app/onyx/main/deployment/docker_compose/install.sh > install.sh && chmod +x install.sh && ./install.sh
 > ```
 
 ****

backend/ee/onyx/background/celery/tasks/doc_permission_syncing/tasks.py

@@ -28,7 +28,6 @@ from onyx.access.models import DocExternalAccess
 from onyx.access.models import ElementExternalAccess
 from onyx.background.celery.apps.app_base import task_logger
 from onyx.background.celery.celery_redis import celery_find_task
-from onyx.background.celery.celery_redis import celery_get_broker_client
 from onyx.background.celery.celery_redis import celery_get_queue_length
 from onyx.background.celery.celery_redis import celery_get_queued_task_ids
 from onyx.background.celery.celery_redis import celery_get_unacked_task_ids
@@ -188,6 +187,7 @@ def check_for_doc_permissions_sync(self: Task, *, tenant_id: str) -> bool | None
     # (which lives on a different db number)
     r = get_redis_client()
     r_replica = get_redis_replica_client()
+    r_celery: Redis = self.app.broker_connection().channel().client  # type: ignore
 
     lock_beat: RedisLock = r.lock(
         OnyxRedisLocks.CHECK_CONNECTOR_DOC_PERMISSIONS_SYNC_BEAT_LOCK,
@@ -227,7 +227,6 @@ def check_for_doc_permissions_sync(self: Task, *, tenant_id: str) -> bool | None
             # tasks can be in the queue in redis, in reserved tasks (prefetched by the worker),
             # or be currently executing
             try:
-                r_celery = celery_get_broker_client(self.app)
                 validate_permission_sync_fences(
                     tenant_id, r, r_replica, r_celery, lock_beat
                 )

backend/ee/onyx/background/celery/tasks/external_group_syncing/tasks.py

@@ -29,7 +29,6 @@ from ee.onyx.external_permissions.sync_params import (
 from ee.onyx.external_permissions.sync_params import get_source_perm_sync_config
 from onyx.background.celery.apps.app_base import task_logger
 from onyx.background.celery.celery_redis import celery_find_task
-from onyx.background.celery.celery_redis import celery_get_broker_client
 from onyx.background.celery.celery_redis import celery_get_unacked_task_ids
 from onyx.background.celery.tasks.beat_schedule import CLOUD_BEAT_MULTIPLIER_DEFAULT
 from onyx.background.error_logging import emit_background_error
@@ -163,6 +162,7 @@ def check_for_external_group_sync(self: Task, *, tenant_id: str) -> bool | None:
     # (which lives on a different db number)
     r = get_redis_client()
     r_replica = get_redis_replica_client()
+    r_celery: Redis = self.app.broker_connection().channel().client  # type: ignore
 
     lock_beat: RedisLock = r.lock(
         OnyxRedisLocks.CHECK_CONNECTOR_EXTERNAL_GROUP_SYNC_BEAT_LOCK,
@@ -221,7 +221,6 @@ def check_for_external_group_sync(self: Task, *, tenant_id: str) -> bool | None:
             # tasks can be in the queue in redis, in reserved tasks (prefetched by the worker),
             # or be currently executing
             try:
-                r_celery = celery_get_broker_client(self.app)
                 validate_external_group_sync_fences(
                     tenant_id, self.app, r, r_replica, r_celery, lock_beat
                 )

backend/onyx/background/celery/celery_redis.py

@@ -1,6 +1,5 @@
 # These are helper objects for tracking the keys we need to write in redis
 import json
-import threading
 from typing import Any
 from typing import cast
 
@@ -8,59 +7,7 @@ from celery import Celery
 from redis import Redis
 
 from onyx.background.celery.configs.base import CELERY_SEPARATOR
-from onyx.configs.app_configs import REDIS_HEALTH_CHECK_INTERVAL
 from onyx.configs.constants import OnyxCeleryPriority
-from onyx.configs.constants import REDIS_SOCKET_KEEPALIVE_OPTIONS
-
-
-_broker_client: Redis | None = None
-_broker_url: str | None = None
-_broker_client_lock = threading.Lock()
-
-
-def celery_get_broker_client(app: Celery) -> Redis:
-    """Return a shared Redis client connected to the Celery broker DB.
-
-    Uses a module-level singleton so all tasks on a worker share one
-    connection instead of creating a new one per call. The client
-    connects directly to the broker Redis DB (parsed from the broker URL).
-
-    Thread-safe via lock — safe for use in Celery thread-pool workers.
-
-    Usage:
-        r_celery = celery_get_broker_client(self.app)
-        length = celery_get_queue_length(queue, r_celery)
-    """
-    global _broker_client, _broker_url
-    with _broker_client_lock:
-        url = app.conf.broker_url
-        if _broker_client is not None and _broker_url == url:
-            try:
-                _broker_client.ping()
-                return _broker_client
-            except Exception:
-                try:
-                    _broker_client.close()
-                except Exception:
-                    pass
-                _broker_client = None
-        elif _broker_client is not None:
-            try:
-                _broker_client.close()
-            except Exception:
-                pass
-            _broker_client = None
-
-        _broker_url = url
-        _broker_client = Redis.from_url(
-            url,
-            decode_responses=False,
-            health_check_interval=REDIS_HEALTH_CHECK_INTERVAL,
-            socket_keepalive=True,
-            socket_keepalive_options=REDIS_SOCKET_KEEPALIVE_OPTIONS,
-            retry_on_timeout=True,
-        )
-        return _broker_client
 
 
 def celery_get_unacked_length(r: Redis) -> int:

backend/onyx/background/celery/tasks/connector_deletion/tasks.py

@@ -14,7 +14,6 @@ from redis.lock import Lock as RedisLock
 from sqlalchemy.orm import Session
 
 from onyx.background.celery.apps.app_base import task_logger
-from onyx.background.celery.celery_redis import celery_get_broker_client
 from onyx.background.celery.celery_redis import celery_get_queue_length
 from onyx.background.celery.celery_redis import celery_get_queued_task_ids
 from onyx.configs.app_configs import JOB_TIMEOUT
@@ -133,6 +132,7 @@ def revoke_tasks_blocking_deletion(
 def check_for_connector_deletion_task(self: Task, *, tenant_id: str) -> bool | None:
     r = get_redis_client()
     r_replica = get_redis_replica_client()
+    r_celery: Redis = self.app.broker_connection().channel().client  # type: ignore
 
     lock_beat: RedisLock = r.lock(
         OnyxRedisLocks.CHECK_CONNECTOR_DELETION_BEAT_LOCK,
@@ -149,7 +149,6 @@ def check_for_connector_deletion_task(self: Task, *, tenant_id: str) -> bool | N
         if not r.exists(OnyxRedisSignals.BLOCK_VALIDATE_CONNECTOR_DELETION_FENCES):
             # clear fences that don't have associated celery tasks in progress
             try:
-                r_celery = celery_get_broker_client(self.app)
                 validate_connector_deletion_fences(
                     tenant_id, r, r_replica, r_celery, lock_beat
                 )

backend/onyx/background/celery/tasks/docprocessing/tasks.py

@@ -22,7 +22,6 @@ from sqlalchemy.orm import Session
 
 from onyx.background.celery.apps.app_base import task_logger
 from onyx.background.celery.celery_redis import celery_find_task
-from onyx.background.celery.celery_redis import celery_get_broker_client
 from onyx.background.celery.celery_redis import celery_get_unacked_task_ids
 from onyx.background.celery.celery_utils import httpx_init_vespa_pool
 from onyx.background.celery.memory_monitoring import emit_process_memory
@@ -450,7 +449,7 @@ def check_indexing_completion(
             ):
                 # Check if the task exists in the celery queue
                 # This handles the case where Redis dies after task creation but before task execution
-                redis_celery = celery_get_broker_client(task.app)
+                redis_celery = task.app.broker_connection().channel().client  # type: ignore
                 task_exists = celery_find_task(
                     attempt.celery_task_id,
                     OnyxCeleryQueues.CONNECTOR_DOC_FETCHING,

backend/onyx/background/celery/tasks/monitoring/tasks.py

@@ -1,5 +1,6 @@
 import json
 import time
+from collections.abc import Callable
 from datetime import timedelta
 from itertools import islice
 from typing import Any
@@ -18,7 +19,6 @@ from sqlalchemy import text
 from sqlalchemy.orm import Session
 
 from onyx.background.celery.apps.app_base import task_logger
-from onyx.background.celery.celery_redis import celery_get_broker_client
 from onyx.background.celery.celery_redis import celery_get_queue_length
 from onyx.background.celery.celery_redis import celery_get_unacked_task_ids
 from onyx.background.celery.memory_monitoring import emit_process_memory
@@ -698,27 +698,31 @@ def monitor_background_processes(self: Task, *, tenant_id: str) -> None:
         return None
 
     try:
+        # Get Redis client for Celery broker
+        redis_celery = self.app.broker_connection().channel().client  # type: ignore
         redis_std = get_redis_client()
 
-        # Collect queue metrics with broker connection
-        r_celery = celery_get_broker_client(self.app)
-        queue_metrics = _collect_queue_metrics(r_celery)
+        # Define metric collection functions and their dependencies
+        metric_functions: list[Callable[[], list[Metric]]] = [
+            lambda: _collect_queue_metrics(redis_celery),
+            lambda: _collect_connector_metrics(db_session, redis_std),
+            lambda: _collect_sync_metrics(db_session, redis_std),
+        ]
 
-        # Collect remaining metrics (no broker connection needed)
+        # Collect and log each metric
         with get_session_with_current_tenant() as db_session:
-            all_metrics: list[Metric] = queue_metrics
-            all_metrics.extend(_collect_connector_metrics(db_session, redis_std))
-            all_metrics.extend(_collect_sync_metrics(db_session, redis_std))
+            for metric_fn in metric_functions:
+                metrics = metric_fn()
+                for metric in metrics:
+                    # double check to make sure we aren't double-emitting metrics
+                    if metric.key is None or not _has_metric_been_emitted(
+                        redis_std, metric.key
+                    ):
+                        metric.log()
+                        metric.emit(tenant_id)
 
-            for metric in all_metrics:
-                if metric.key is None or not _has_metric_been_emitted(
-                    redis_std, metric.key
-                ):
-                    metric.log()
-                    metric.emit(tenant_id)
-
-                if metric.key is not None:
-                    _mark_metric_as_emitted(redis_std, metric.key)
+                    if metric.key is not None:
+                        _mark_metric_as_emitted(redis_std, metric.key)
 
         task_logger.info("Successfully collected background metrics")
     except SoftTimeLimitExceeded:
@@ -886,7 +890,7 @@ def monitor_celery_queues_helper(
 ) -> None:
     """A task to monitor all celery queue lengths."""
 
-    r_celery = celery_get_broker_client(task.app)
+    r_celery = task.app.broker_connection().channel().client  # type: ignore
     n_celery = celery_get_queue_length(OnyxCeleryQueues.PRIMARY, r_celery)
     n_docfetching = celery_get_queue_length(
         OnyxCeleryQueues.CONNECTOR_DOC_FETCHING, r_celery
@@ -1076,7 +1080,7 @@ def cloud_monitor_celery_pidbox(
     num_deleted = 0
 
     MAX_PIDBOX_IDLE = 24 * 3600  # 1 day in seconds
-    r_celery = celery_get_broker_client(self.app)
+    r_celery: Redis = self.app.broker_connection().channel().client  # type: ignore
     for key in r_celery.scan_iter("*.reply.celery.pidbox"):
         key_bytes = cast(bytes, key)
         key_str = key_bytes.decode("utf-8")

backend/onyx/background/celery/tasks/pruning/tasks.py

@@ -17,7 +17,6 @@ from sqlalchemy.orm import Session
 
 from onyx.background.celery.apps.app_base import task_logger
 from onyx.background.celery.celery_redis import celery_find_task
-from onyx.background.celery.celery_redis import celery_get_broker_client
 from onyx.background.celery.celery_redis import celery_get_queue_length
 from onyx.background.celery.celery_redis import celery_get_queued_task_ids
 from onyx.background.celery.celery_redis import celery_get_unacked_task_ids
@@ -204,6 +203,7 @@ def _is_pruning_due(cc_pair: ConnectorCredentialPair) -> bool:
 def check_for_pruning(self: Task, *, tenant_id: str) -> bool | None:
     r = get_redis_client()
     r_replica = get_redis_replica_client()
+    r_celery: Redis = self.app.broker_connection().channel().client  # type: ignore
 
     lock_beat: RedisLock = r.lock(
         OnyxRedisLocks.CHECK_PRUNE_BEAT_LOCK,
@@ -261,7 +261,6 @@ def check_for_pruning(self: Task, *, tenant_id: str) -> bool | None:
             # tasks can be in the queue in redis, in reserved tasks (prefetched by the worker),
             # or be currently executing
             try:
-                r_celery = celery_get_broker_client(self.app)
                 validate_pruning_fences(tenant_id, r, r_replica, r_celery, lock_beat)
             except Exception:
                 task_logger.exception("Exception while validating pruning fences")

backend/onyx/background/celery/tasks/user_file_processing/tasks.py

@@ -16,7 +16,6 @@ from sqlalchemy.orm import Session
 
 from onyx.access.access import build_access_for_user_files
 from onyx.background.celery.apps.app_base import task_logger
-from onyx.background.celery.celery_redis import celery_get_broker_client
 from onyx.background.celery.celery_redis import celery_get_queue_length
 from onyx.background.celery.celery_utils import httpx_init_vespa_pool
 from onyx.background.celery.tasks.shared.RetryDocumentIndex import RetryDocumentIndex
@@ -106,7 +105,7 @@ def _user_file_delete_queued_key(user_file_id: str | UUID) -> str:
 
 
 def get_user_file_project_sync_queue_depth(celery_app: Celery) -> int:
-    redis_celery = celery_get_broker_client(celery_app)
+    redis_celery: Redis = celery_app.broker_connection().channel().client  # type: ignore
     return celery_get_queue_length(
         OnyxCeleryQueues.USER_FILE_PROJECT_SYNC, redis_celery
     )
@@ -239,7 +238,7 @@ def check_user_file_processing(self: Task, *, tenant_id: str) -> None:
     skipped_guard = 0
     try:
         # --- Protection 1: queue depth backpressure ---
-        r_celery = celery_get_broker_client(self.app)
+        r_celery = self.app.broker_connection().channel().client  # type: ignore
         queue_len = celery_get_queue_length(
             OnyxCeleryQueues.USER_FILE_PROCESSING, r_celery
         )
@@ -592,7 +591,7 @@ def check_for_user_file_delete(self: Task, *, tenant_id: str) -> None:
         # --- Protection 1: queue depth backpressure ---
         # NOTE: must use the broker's Redis client (not redis_client) because
         # Celery queues live on a separate Redis DB with CELERY_SEPARATOR keys.
-        r_celery = celery_get_broker_client(self.app)
+        r_celery: Redis = self.app.broker_connection().channel().client  # type: ignore
         queue_len = celery_get_queue_length(OnyxCeleryQueues.USER_FILE_DELETE, r_celery)
         if queue_len > USER_FILE_DELETE_MAX_QUEUE_DEPTH:
             task_logger.warning(

backend/onyx/db/projects.py

@@ -7,6 +7,7 @@ from fastapi import HTTPException
 from fastapi import UploadFile
 from pydantic import BaseModel
 from pydantic import ConfigDict
+from pydantic import Field
 from sqlalchemy import func
 from sqlalchemy.orm import Session
 from starlette.background import BackgroundTasks
@@ -34,9 +35,19 @@ class CategorizedFilesResult(BaseModel):
     user_files: list[UserFile]
     rejected_files: list[RejectedFile]
     id_to_temp_id: dict[str, str]
+    # Filenames that should be stored but not indexed.
+    skip_indexing_filenames: set[str] = Field(default_factory=set)
     # Allow SQLAlchemy ORM models inside this result container
     model_config = ConfigDict(arbitrary_types_allowed=True)
 
+    @property
+    def indexable_files(self) -> list[UserFile]:
+        return [
+            uf
+            for uf in self.user_files
+            if (uf.name or "") not in self.skip_indexing_filenames
+        ]
+
 
 def build_hashed_file_key(file: UploadFile) -> str:
     name_prefix = (file.filename or "")[:50]
@@ -98,6 +109,7 @@ def create_user_files(
         user_files=user_files,
         rejected_files=rejected_files,
         id_to_temp_id=id_to_temp_id,
+        skip_indexing_filenames=categorized_files.skip_indexing,
     )
 
 
@@ -123,6 +135,7 @@ def upload_files_to_user_files_with_indexing(
     user_files = categorized_files_result.user_files
     rejected_files = categorized_files_result.rejected_files
     id_to_temp_id = categorized_files_result.id_to_temp_id
+    indexable_files = categorized_files_result.indexable_files
     # Trigger per-file processing immediately for the current tenant
     tenant_id = get_current_tenant_id()
     for rejected_file in rejected_files:
@@ -134,12 +147,12 @@ def upload_files_to_user_files_with_indexing(
         from onyx.background.task_utils import drain_processing_loop
 
         background_tasks.add_task(drain_processing_loop, tenant_id)
-        for user_file in user_files:
+        for user_file in indexable_files:
             logger.info(f"Queued in-process processing for user_file_id={user_file.id}")
     else:
         from onyx.background.celery.versioned_apps.client import app as client_app
 
-        for user_file in user_files:
+        for user_file in indexable_files:
             task = client_app.send_task(
                 OnyxCeleryTask.PROCESS_SINGLE_USER_FILE,
                 kwargs={"user_file_id": user_file.id, "tenant_id": tenant_id},
@@ -155,6 +168,7 @@ def upload_files_to_user_files_with_indexing(
         user_files=user_files,
         rejected_files=rejected_files,
         id_to_temp_id=id_to_temp_id,
+        skip_indexing_filenames=categorized_files_result.skip_indexing_filenames,
     )
 
 

backend/onyx/document_index/disabled.py

@@ -5,7 +5,6 @@ accidentally reaches the vector DB layer will fail loudly instead of timing
 out against a nonexistent Vespa/OpenSearch instance.
 """
 
-from collections.abc import Iterable
 from typing import Any
 
 from onyx.context.search.models import IndexFilters
@@ -67,7 +66,7 @@ class DisabledDocumentIndex(DocumentIndex):
     # ------------------------------------------------------------------
     def index(
         self,
-        chunks: Iterable[DocMetadataAwareIndexChunk],  # noqa: ARG002
+        chunks: list[DocMetadataAwareIndexChunk],  # noqa: ARG002
         index_batch_params: IndexBatchParams,  # noqa: ARG002
     ) -> set[DocumentInsertionRecord]:
         raise RuntimeError(VECTOR_DB_DISABLED_ERROR)

backend/onyx/document_index/interfaces.py

@@ -1,5 +1,4 @@
 import abc
-from collections.abc import Iterable
 from dataclasses import dataclass
 from datetime import datetime
 from typing import Any
@@ -207,7 +206,7 @@ class Indexable(abc.ABC):
     @abc.abstractmethod
     def index(
         self,
-        chunks: Iterable[DocMetadataAwareIndexChunk],
+        chunks: list[DocMetadataAwareIndexChunk],
         index_batch_params: IndexBatchParams,
     ) -> set[DocumentInsertionRecord]:
         """
@@ -227,8 +226,8 @@ class Indexable(abc.ABC):
         it is done automatically outside of this code.
 
         Parameters:
-        - chunks: Document chunks with all of the information needed for
-                indexing to the document index.
+        - chunks: Document chunks with all of the information needed for indexing to the document
+                index.
         - tenant_id: The tenant id of the user whose chunks are being indexed
         - large_chunks_enabled: Whether large chunks are enabled
 

backend/onyx/document_index/interfaces_new.py

@@ -1,5 +1,4 @@
 import abc
-from collections.abc import Iterable
 from typing import Self
 
 from pydantic import BaseModel
@@ -210,10 +209,10 @@ class Indexable(abc.ABC):
     @abc.abstractmethod
     def index(
         self,
-        chunks: Iterable[DocMetadataAwareIndexChunk],
+        chunks: list[DocMetadataAwareIndexChunk],
         indexing_metadata: IndexingMetadata,
     ) -> list[DocumentInsertionRecord]:
-        """Indexes an iterable of document chunks into the document index.
+        """Indexes a list of document chunks into the document index.
 
         This is often a batch operation including chunks from multiple
         documents.

backend/onyx/document_index/opensearch/opensearch_document_index.py

@@ -1,5 +1,5 @@
 import json
-from collections.abc import Iterable
+from collections import defaultdict
 from typing import Any
 
 import httpx
@@ -351,7 +351,7 @@ class OpenSearchOldDocumentIndex(OldDocumentIndex):
 
     def index(
         self,
-        chunks: Iterable[DocMetadataAwareIndexChunk],
+        chunks: list[DocMetadataAwareIndexChunk],
         index_batch_params: IndexBatchParams,
     ) -> set[OldDocumentInsertionRecord]:
         """
@@ -647,10 +647,10 @@ class OpenSearchDocumentIndex(DocumentIndex):
 
     def index(
         self,
-        chunks: Iterable[DocMetadataAwareIndexChunk],
-        indexing_metadata: IndexingMetadata,
+        chunks: list[DocMetadataAwareIndexChunk],
+        indexing_metadata: IndexingMetadata,  # noqa: ARG002
     ) -> list[DocumentInsertionRecord]:
-        """Indexes an iterable of document chunks into the document index.
+        """Indexes a list of document chunks into the document index.
 
         Groups chunks by document ID and for each document, deletes existing
         chunks and indexes the new chunks in bulk.
@@ -673,34 +673,29 @@ class OpenSearchDocumentIndex(DocumentIndex):
                 document is newly indexed or had already existed and was just
                 updated.
         """
-        total_chunks = sum(
-            cc.new_chunk_cnt
-            for cc in indexing_metadata.doc_id_to_chunk_cnt_diff.values()
+        # Group chunks by document ID.
+        doc_id_to_chunks: dict[str, list[DocMetadataAwareIndexChunk]] = defaultdict(
+            list
         )
+        for chunk in chunks:
+            doc_id_to_chunks[chunk.source_document.id].append(chunk)
         logger.debug(
-            f"[OpenSearchDocumentIndex] Indexing {total_chunks} chunks from {len(indexing_metadata.doc_id_to_chunk_cnt_diff)} "
+            f"[OpenSearchDocumentIndex] Indexing {len(chunks)} chunks from {len(doc_id_to_chunks)} "
             f"documents for index {self._index_name}."
         )
 
         document_indexing_results: list[DocumentInsertionRecord] = []
-        deleted_doc_ids: set[str] = set()
-        # Buffer chunks per document as they arrive from the iterable.
-        # When the document ID changes flush the buffered chunks.
-        current_doc_id: str | None = None
-        current_chunks: list[DocMetadataAwareIndexChunk] = []
-
-        def _flush_chunks(doc_chunks: list[DocMetadataAwareIndexChunk]) -> None:
-            assert len(doc_chunks) > 0, "doc_chunks is empty"
-
+        # Try to index per-document.
+        for _, chunks in doc_id_to_chunks.items():
             # Create a batch of OpenSearch-formatted chunks for bulk insertion.
-            # Since we are doing this in batches, an error occurring midway
-            # can result in a state where chunks are deleted and not all the
-            # new chunks have been indexed.
+            # Do this before deleting existing chunks to reduce the amount of
+            # time the document index has no content for a given document, and
+            # to reduce the chance of entering a state where we delete chunks,
+            # then some error happens, and never successfully index new chunks.
             chunk_batch: list[DocumentChunk] = [
-                _convert_onyx_chunk_to_opensearch_document(chunk)
-                for chunk in doc_chunks
+                _convert_onyx_chunk_to_opensearch_document(chunk) for chunk in chunks
             ]
-            onyx_document: Document = doc_chunks[0].source_document
+            onyx_document: Document = chunks[0].source_document
             # First delete the doc's chunks from the index. This is so that
             # there are no dangling chunks in the index, in the event that the
             # new document's content contains fewer chunks than the previous
@@ -709,40 +704,22 @@ class OpenSearchDocumentIndex(DocumentIndex):
             # if the chunk count has actually decreased. This assumes that
             # overlapping chunks are perfectly overwritten. If we can't
             # guarantee that then we need the code as-is.
-            if onyx_document.id not in deleted_doc_ids:
-                num_chunks_deleted = self.delete(
-                    onyx_document.id, onyx_document.chunk_count
-                )
-                deleted_doc_ids.add(onyx_document.id)
-                # If we see that chunks were deleted we assume the doc already
-                # existed. We record the result before bulk_index_documents
-                # runs. If indexing raises, this entire result list is discarded
-                # by the caller's retry logic, so early recording is safe.
-                document_indexing_results.append(
-                    DocumentInsertionRecord(
-                        document_id=onyx_document.id,
-                        already_existed=num_chunks_deleted > 0,
-                    )
-                )
+            num_chunks_deleted = self.delete(
+                onyx_document.id, onyx_document.chunk_count
+            )
+            # If we see that chunks were deleted we assume the doc already
+            # existed.
+            document_insertion_record = DocumentInsertionRecord(
+                document_id=onyx_document.id,
+                already_existed=num_chunks_deleted > 0,
+            )
             # Now index. This will raise if a chunk of the same ID exists, which
             # we do not expect because we should have deleted all chunks.
             self._client.bulk_index_documents(
                 documents=chunk_batch,
                 tenant_state=self._tenant_state,
             )
-
-        for chunk in chunks:
-            doc_id = chunk.source_document.id
-            if doc_id != current_doc_id:
-                if current_chunks:
-                    _flush_chunks(current_chunks)
-                current_doc_id = doc_id
-                current_chunks = [chunk]
-            else:
-                current_chunks.append(chunk)
-
-        if current_chunks:
-            _flush_chunks(current_chunks)
+            document_indexing_results.append(document_insertion_record)
 
         return document_indexing_results
 

backend/onyx/document_index/vespa/index.py

@@ -6,7 +6,6 @@ import re
 import time
 import urllib
 import zipfile
-from collections.abc import Iterable
 from dataclasses import dataclass
 from datetime import datetime
 from datetime import timedelta
@@ -462,7 +461,7 @@ class VespaIndex(DocumentIndex):
 
     def index(
         self,
-        chunks: Iterable[DocMetadataAwareIndexChunk],
+        chunks: list[DocMetadataAwareIndexChunk],
         index_batch_params: IndexBatchParams,
     ) -> set[OldDocumentInsertionRecord]:
         """

backend/onyx/document_index/vespa/vespa_document_index.py

@@ -1,8 +1,6 @@
 import concurrent.futures
 import logging
 import random
-from collections.abc import Generator
-from collections.abc import Iterable
 from typing import Any
 from uuid import UUID
 
@@ -320,7 +318,7 @@ class VespaDocumentIndex(DocumentIndex):
 
     def index(
         self,
-        chunks: Iterable[DocMetadataAwareIndexChunk],
+        chunks: list[DocMetadataAwareIndexChunk],
         indexing_metadata: IndexingMetadata,
     ) -> list[DocumentInsertionRecord]:
         doc_id_to_chunk_cnt_diff = indexing_metadata.doc_id_to_chunk_cnt_diff
@@ -340,31 +338,22 @@ class VespaDocumentIndex(DocumentIndex):
 
         # Vespa has restrictions on valid characters, yet document IDs come from
         # external w.r.t. this class. We need to sanitize them.
-        #
-        # Instead of materializing all cleaned chunks upfront, we stream them
-        # through a generator that cleans IDs and builds the original-ID mapping
-        # incrementally as chunks flow into Vespa.
-        def _clean_and_track(
-            chunks_iter: Iterable[DocMetadataAwareIndexChunk],
-            id_map: dict[str, str],
-            seen_ids: set[str],
-        ) -> Generator[DocMetadataAwareIndexChunk, None, None]:
-            """Cleans chunk IDs and builds the original-ID mapping
-            incrementally as chunks flow through, avoiding a separate
-            materialization pass."""
-            for chunk in chunks_iter:
-                original_id = chunk.source_document.id
-                cleaned = clean_chunk_id_copy(chunk)
-                cleaned_id = cleaned.source_document.id
-                # Needed so the final DocumentInsertionRecord returned can have
-                # the original document ID. cleaned_chunks might not contain IDs
-                # exactly as callers supplied them.
-                id_map[cleaned_id] = original_id
-                seen_ids.add(cleaned_id)
-                yield cleaned
+        cleaned_chunks: list[DocMetadataAwareIndexChunk] = [
+            clean_chunk_id_copy(chunk) for chunk in chunks
+        ]
+        assert len(cleaned_chunks) == len(
+            chunks
+        ), "Bug: Cleaned chunks and input chunks have different lengths."
 
-        new_document_id_to_original_document_id: dict[str, str] = {}
-        all_cleaned_doc_ids: set[str] = set()
+        # Needed so the final DocumentInsertionRecord returned can have the
+        # original document ID. cleaned_chunks might not contain IDs exactly as
+        # callers supplied them.
+        new_document_id_to_original_document_id: dict[str, str] = dict()
+        for i, cleaned_chunk in enumerate(cleaned_chunks):
+            old_chunk = chunks[i]
+            new_document_id_to_original_document_id[
+                cleaned_chunk.source_document.id
+            ] = old_chunk.source_document.id
 
         existing_docs: set[str] = set()
 
@@ -420,13 +409,7 @@ class VespaDocumentIndex(DocumentIndex):
                     executor=executor,
                 )
 
-            # Insert new Vespa documents, streaming through the cleaning
-            # pipeline so chunks are never fully materialized.
-            cleaned_chunks = _clean_and_track(
-                chunks,
-                new_document_id_to_original_document_id,
-                all_cleaned_doc_ids,
-            )
+            # Insert new Vespa documents.
             for chunk_batch in batch_generator(cleaned_chunks, BATCH_SIZE):
                 batch_index_vespa_chunks(
                     chunks=chunk_batch,
@@ -436,6 +419,10 @@ class VespaDocumentIndex(DocumentIndex):
                     executor=executor,
                 )
 
+        all_cleaned_doc_ids: set[str] = {
+            chunk.source_document.id for chunk in cleaned_chunks
+        }
+
         return [
             DocumentInsertionRecord(
                 document_id=new_document_id_to_original_document_id[cleaned_doc_id],

backend/onyx/file_processing/file_types.py

@@ -15,6 +15,7 @@ PLAIN_TEXT_MIME_TYPE = "text/plain"
 class OnyxMimeTypes:
     IMAGE_MIME_TYPES = {"image/jpg", "image/jpeg", "image/png", "image/webp"}
     CSV_MIME_TYPES = {"text/csv"}
+    TABULAR_MIME_TYPES = CSV_MIME_TYPES | {SPREADSHEET_MIME_TYPE}
     TEXT_MIME_TYPES = {
         PLAIN_TEXT_MIME_TYPE,
         "text/markdown",
@@ -34,13 +35,12 @@ class OnyxMimeTypes:
         PDF_MIME_TYPE,
         WORD_PROCESSING_MIME_TYPE,
         PRESENTATION_MIME_TYPE,
-        SPREADSHEET_MIME_TYPE,
         "message/rfc822",
         "application/epub+zip",
     }
 
     ALLOWED_MIME_TYPES = IMAGE_MIME_TYPES.union(
-        TEXT_MIME_TYPES, DOCUMENT_MIME_TYPES, CSV_MIME_TYPES
+        TEXT_MIME_TYPES, DOCUMENT_MIME_TYPES, TABULAR_MIME_TYPES
     )
 
     EXCLUDED_IMAGE_TYPES = {

backend/onyx/file_store/models.py

@@ -13,13 +13,14 @@ class ChatFileType(str, Enum):
     DOC = "document"
     # Plain text only contain the text
     PLAIN_TEXT = "plain_text"
-    CSV = "csv"
+    # Tabular data files (CSV, TSV, XLSX) — metadata-only injection
+    TABULAR = "tabular"
 
     def is_text_file(self) -> bool:
         return self in (
             ChatFileType.PLAIN_TEXT,
             ChatFileType.DOC,
-            ChatFileType.CSV,
+            ChatFileType.TABULAR,
         )
 
 

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

@@ -3844,9 +3844,9 @@
       }
     },
     "node_modules/@ts-morph/common/node_modules/brace-expansion": {
-      "version": "5.0.5",
-      "resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-5.0.5.tgz",
-      "integrity": "sha512-VZznLgtwhn+Mact9tfiwx64fA9erHH/MCXEUfB/0bX/6Fz6ny5EGTXYltMocqg4xFAQZtnO3DHWWXi8RiuN7cQ==",
+      "version": "5.0.3",
+      "resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-5.0.3.tgz",
+      "integrity": "sha512-fy6KJm2RawA5RcHkLa1z/ScpBeA762UF9KmZQxwIbDtRJrgLzM10depAiEQ+CXYcoiqW1/m96OAAoke2nE9EeA==",
       "license": "MIT",
       "dependencies": {
         "balanced-match": "^4.0.2"
@@ -4224,9 +4224,9 @@
       }
     },
     "node_modules/@typescript-eslint/typescript-estree/node_modules/brace-expansion": {
-      "version": "2.0.3",
-      "resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-2.0.3.tgz",
-      "integrity": "sha512-MCV/fYJEbqx68aE58kv2cA/kiky1G8vux3OR6/jbS+jIMe/6fJWa0DTzJU7dqijOWYwHi1t29FlfYI9uytqlpA==",
+      "version": "2.0.2",
+      "resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-2.0.2.tgz",
+      "integrity": "sha512-Jt0vHyM+jmUBqojB7E1NIYadt0vI0Qxjxd2TErW94wDz+E2LAm5vKMXXwg6ZZBTHPuUlDgQHKXvjGBdfcF1ZDQ==",
       "dev": true,
       "license": "MIT",
       "dependencies": {
@@ -5007,9 +5007,9 @@
       }
     },
     "node_modules/brace-expansion": {
-      "version": "1.1.13",
-      "resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-1.1.13.tgz",
-      "integrity": "sha512-9ZLprWS6EENmhEOpjCYW2c8VkmOvckIJZfkr7rBW6dObmfgJ/L1GpSYW5Hpo9lDz4D1+n0Ckz8rU7FwHDQiG/w==",
+      "version": "1.1.12",
+      "resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-1.1.12.tgz",
+      "integrity": "sha512-9T9UjW3r0UW5c1Q7GTwllptXwhvYmEzFhzMfZ9H7FQWt+uZePjZPjBP/W1ZEyZ1twGWom5/56TF4lPcqjnDHcg==",
       "dev": true,
       "license": "MIT",
       "dependencies": {

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

@@ -123,8 +123,9 @@ def _validate_endpoint(
     (not reachable — indicates the api_key is invalid).
 
     Timeout handling:
-    - Any httpx.TimeoutException (ConnectTimeout, ReadTimeout, WriteTimeout, PoolTimeout) →
-      timeout (operator should consider increasing timeout_seconds).
+    - ConnectTimeout: TCP handshake never completed → cannot_connect.
+    - ReadTimeout / WriteTimeout: TCP was established, server responded slowly → timeout
+      (operator should consider increasing timeout_seconds).
     - All other exceptions → cannot_connect.
     """
     _check_ssrf_safety(endpoint_url)

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

@@ -76,11 +76,26 @@ class CategorizedFiles(BaseModel):
     acceptable: list[UploadFile] = Field(default_factory=list)
     rejected: list[RejectedFile] = Field(default_factory=list)
     acceptable_file_to_token_count: dict[str, int] = Field(default_factory=dict)
+    # Filenames within `acceptable` that should be stored but not indexed.
+    skip_indexing: set[str] = Field(default_factory=set)
 
     # Allow FastAPI UploadFile instances
     model_config = ConfigDict(arbitrary_types_allowed=True)
 
 
+# Extensions that bypass the token-count threshold on upload.
+_TOKEN_THRESHOLD_EXEMPT_EXTENSIONS: set[str] = {
+    ".csv",
+    ".tsv",
+    ".xlsx",
+}
+
+
+def _skip_token_threshold(extension: str) -> bool:
+    """Return True if this file extension should bypass the token limit."""
+    return extension.lower() in _TOKEN_THRESHOLD_EXEMPT_EXTENSIONS
+
+
 def _apply_long_side_cap(width: int, height: int, cap: int) -> tuple[int, int]:
     if max(width, height) <= cap:
         return width, height
@@ -264,7 +279,17 @@ def categorize_uploaded_files(
                 token_count = count_tokens(
                     text_content, tokenizer, token_limit=token_threshold
                 )
-                if token_threshold is not None and token_count > token_threshold:
+                exceeds_threshold = (
+                    token_threshold is not None and token_count > token_threshold
+                )
+                if exceeds_threshold and _skip_token_threshold(extension):
+                    # Exempt extensions (e.g. spreadsheets) are accepted
+                    # but flagged to skip indexing — only metadata is
+                    # injected into the LLM context.
+                    results.acceptable.append(upload)
+                    results.acceptable_file_to_token_count[filename] = token_count
+                    results.skip_indexing.add(filename)
+                elif exceeds_threshold:
                     results.rejected.append(
                         RejectedFile(
                             filename=filename,

backend/onyx/server/metrics/indexing_pipeline.py

@@ -12,6 +12,7 @@ stale, which is fine for monitoring dashboards.
 import json
 import threading
 import time
+from collections.abc import Callable
 from datetime import datetime
 from datetime import timezone
 from typing import Any
@@ -103,23 +104,25 @@ class _CachedCollector(Collector):
 
 
 class QueueDepthCollector(_CachedCollector):
-    """Reads Celery queue lengths from the broker Redis on each scrape."""
+    """Reads Celery queue lengths from the broker Redis on each scrape.
+
+    Uses a Redis client factory (callable) rather than a stored client
+    reference so the connection is always fresh from Celery's pool.
+    """
 
     def __init__(self, cache_ttl: float = _DEFAULT_CACHE_TTL) -> None:
         super().__init__(cache_ttl)
-        self._celery_app: Any | None = None
+        self._get_redis: Callable[[], Redis] | None = None
 
-    def set_celery_app(self, app: Any) -> None:
-        """Set the Celery app for broker Redis access."""
-        self._celery_app = app
+    def set_redis_factory(self, factory: Callable[[], Redis]) -> None:
+        """Set a callable that returns a broker Redis client on demand."""
+        self._get_redis = factory
 
     def _collect_fresh(self) -> list[GaugeMetricFamily]:
-        if self._celery_app is None:
+        if self._get_redis is None:
             return []
 
-        from onyx.background.celery.celery_redis import celery_get_broker_client
-
-        redis_client = celery_get_broker_client(self._celery_app)
+        redis_client = self._get_redis()
 
         depth = GaugeMetricFamily(
             "onyx_queue_depth",
@@ -401,19 +404,17 @@ class RedisHealthCollector(_CachedCollector):
 
     def __init__(self, cache_ttl: float = _DEFAULT_CACHE_TTL) -> None:
         super().__init__(cache_ttl)
-        self._celery_app: Any | None = None
+        self._get_redis: Callable[[], Redis] | None = None
 
-    def set_celery_app(self, app: Any) -> None:
-        """Set the Celery app for broker Redis access."""
-        self._celery_app = app
+    def set_redis_factory(self, factory: Callable[[], Redis]) -> None:
+        """Set a callable that returns a broker Redis client on demand."""
+        self._get_redis = factory
 
     def _collect_fresh(self) -> list[GaugeMetricFamily]:
-        if self._celery_app is None:
+        if self._get_redis is None:
             return []
 
-        from onyx.background.celery.celery_redis import celery_get_broker_client
-
-        redis_client = celery_get_broker_client(self._celery_app)
+        redis_client = self._get_redis()
 
         memory_used = GaugeMetricFamily(
             "onyx_redis_memory_used_bytes",

backend/onyx/server/metrics/indexing_pipeline_setup.py

@@ -3,8 +3,12 @@
 Called once by the monitoring celery worker after Redis and DB are ready.
 """
 
+from collections.abc import Callable
+from typing import Any
+
 from celery import Celery
 from prometheus_client.registry import REGISTRY
+from redis import Redis
 
 from onyx.server.metrics.indexing_pipeline import ConnectorHealthCollector
 from onyx.server.metrics.indexing_pipeline import IndexAttemptCollector
@@ -17,7 +21,7 @@ from onyx.utils.logger import setup_logger
 logger = setup_logger()
 
 # Module-level singletons — these are lightweight objects (no connections or DB
-# state) until configure() / set_celery_app() is called. Keeping them at
+# state) until configure() / set_redis_factory() is called. Keeping them at
 # module level ensures they survive the lifetime of the worker process and are
 # only registered with the Prometheus registry once.
 _queue_collector = QueueDepthCollector()
@@ -28,15 +32,72 @@ _worker_health_collector = WorkerHealthCollector()
 _heartbeat_monitor: WorkerHeartbeatMonitor | None = None
 
 
+def _make_broker_redis_factory(celery_app: Celery) -> Callable[[], Redis]:
+    """Create a factory that returns a cached broker Redis client.
+
+    Reuses a single connection across scrapes to avoid leaking connections.
+    Reconnects automatically if the cached connection becomes stale.
+    """
+    _cached_client: list[Redis | None] = [None]
+    # Keep a reference to the Kombu Connection so we can close it on
+    # reconnect (the raw Redis client outlives the Kombu wrapper).
+    _cached_kombu_conn: list[Any] = [None]
+
+    def _close_client(client: Redis) -> None:
+        """Best-effort close of a Redis client."""
+        try:
+            client.close()
+        except Exception:
+            logger.debug("Failed to close stale Redis client", exc_info=True)
+
+    def _close_kombu_conn() -> None:
+        """Best-effort close of the cached Kombu Connection."""
+        conn = _cached_kombu_conn[0]
+        if conn is not None:
+            try:
+                conn.close()
+            except Exception:
+                logger.debug("Failed to close Kombu connection", exc_info=True)
+            _cached_kombu_conn[0] = None
+
+    def _get_broker_redis() -> Redis:
+        client = _cached_client[0]
+        if client is not None:
+            try:
+                client.ping()
+                return client
+            except Exception:
+                logger.debug("Cached Redis client stale, reconnecting")
+                _close_client(client)
+                _cached_client[0] = None
+                _close_kombu_conn()
+
+        # Get a fresh Redis client from the broker connection.
+        # We hold this client long-term (cached above) rather than using a
+        # context manager, because we need it to persist across scrapes.
+        # The caching logic above ensures we only ever hold one connection,
+        # and we close it explicitly on reconnect.
+        conn = celery_app.broker_connection()
+        # kombu's Channel exposes .client at runtime (the underlying Redis
+        # client) but the type stubs don't declare it.
+        new_client: Redis = conn.channel().client  # type: ignore[attr-defined]
+        _cached_client[0] = new_client
+        _cached_kombu_conn[0] = conn
+        return new_client
+
+    return _get_broker_redis
+
+
 def setup_indexing_pipeline_metrics(celery_app: Celery) -> None:
     """Register all indexing pipeline collectors with the default registry.
 
     Args:
-        celery_app: The Celery application instance. Used to obtain a
+        celery_app: The Celery application instance. Used to obtain a fresh
             broker Redis client on each scrape for queue depth metrics.
     """
-    _queue_collector.set_celery_app(celery_app)
-    _redis_health_collector.set_celery_app(celery_app)
+    redis_factory = _make_broker_redis_factory(celery_app)
+    _queue_collector.set_redis_factory(redis_factory)
+    _redis_health_collector.set_redis_factory(redis_factory)
 
     # Start the heartbeat monitor daemon thread — uses a single persistent
     # connection to receive worker-heartbeat events.

backend/onyx/server/query_and_chat/chat_utils.py

@@ -9,8 +9,8 @@ def mime_type_to_chat_file_type(mime_type: str | None) -> ChatFileType:
     if mime_type in OnyxMimeTypes.IMAGE_MIME_TYPES:
         return ChatFileType.IMAGE
 
-    if mime_type in OnyxMimeTypes.CSV_MIME_TYPES:
-        return ChatFileType.CSV
+    if mime_type in OnyxMimeTypes.TABULAR_MIME_TYPES:
+        return ChatFileType.TABULAR
 
     if mime_type in OnyxMimeTypes.DOCUMENT_MIME_TYPES:
         return ChatFileType.DOC

backend/onyx/tools/tool_implementations/file_reader/file_reader_tool.py

@@ -169,10 +169,10 @@ class FileReaderTool(Tool[FileReaderToolOverrideKwargs]):
 
         chat_file = self._load_file(file_id)
 
-        # Only PLAIN_TEXT and CSV are guaranteed to contain actual text bytes.
+        # Only PLAIN_TEXT and TABULAR are guaranteed to contain actual text bytes.
         # DOC type in a loaded file means plaintext extraction failed and the
         # content is the original binary (e.g. raw PDF/DOCX bytes).
-        if chat_file.file_type not in (ChatFileType.PLAIN_TEXT, ChatFileType.CSV):
+        if chat_file.file_type not in (ChatFileType.PLAIN_TEXT, ChatFileType.TABULAR):
             raise ToolCallException(
                 message=f"File {file_id} is not a text file (type={chat_file.file_type})",
                 llm_facing_message=(

backend/requirements/dev.txt

@@ -263,7 +263,7 @@ oauthlib==3.2.2
     # via
     #   kubernetes
     #   requests-oauthlib
-onyx-devtools==0.7.2
+onyx-devtools==0.7.1
     # via onyx
 openai==2.14.0
     # via

backend/tests/external_dependency_unit/celery/test_persona_file_sync.py

@@ -129,10 +129,6 @@ def _patch_task_app(task: Any, mock_app: MagicMock) -> Generator[None, None, Non
             return_value=mock_app,
         ),
         patch(_PATCH_QUEUE_DEPTH, return_value=0),
-        patch(
-            "onyx.background.celery.tasks.user_file_processing.tasks.celery_get_broker_client",
-            return_value=MagicMock(),
-        ),
     ):
         yield
 

backend/tests/external_dependency_unit/celery/test_user_file_delete_queue.py

@@ -88,22 +88,10 @@ def _patch_task_app(task: Any, mock_app: MagicMock) -> Generator[None, None, Non
     the actual task instance.  We patch ``app`` on that instance's class
     (a unique Celery-generated Task subclass) so the mock is scoped to this
     task only.
-
-    Also patches ``celery_get_broker_client`` so the mock app doesn't need
-    a real broker URL.
     """
     task_instance = task.run.__self__
-    with (
-        patch.object(
-            type(task_instance),
-            "app",
-            new_callable=PropertyMock,
-            return_value=mock_app,
-        ),
-        patch(
-            "onyx.background.celery.tasks.user_file_processing.tasks.celery_get_broker_client",
-            return_value=MagicMock(),
-        ),
+    with patch.object(
+        type(task_instance), "app", new_callable=PropertyMock, return_value=mock_app
     ):
         yield
 

backend/tests/external_dependency_unit/celery/test_user_file_processing_queue.py

@@ -90,17 +90,8 @@ def _patch_task_app(task: Any, mock_app: MagicMock) -> Generator[None, None, Non
     task only.
     """
     task_instance = task.run.__self__
-    with (
-        patch.object(
-            type(task_instance),
-            "app",
-            new_callable=PropertyMock,
-            return_value=mock_app,
-        ),
-        patch(
-            "onyx.background.celery.tasks.user_file_processing.tasks.celery_get_broker_client",
-            return_value=MagicMock(),
-        ),
+    with patch.object(
+        type(task_instance), "app", new_callable=PropertyMock, return_value=mock_app
     ):
         yield
 

backend/tests/external_dependency_unit/document_index/test_document_index.py

@@ -6,7 +6,6 @@ These tests assume Vespa and OpenSearch are running.
 import time
 import uuid
 from collections.abc import Generator
-from collections.abc import Iterator
 
 import httpx
 import pytest
@@ -22,7 +21,6 @@ from onyx.document_index.opensearch.opensearch_document_index import (
 )
 from onyx.document_index.vespa.index import VespaIndex
 from onyx.document_index.vespa.vespa_document_index import VespaDocumentIndex
-from onyx.indexing.models import DocMetadataAwareIndexChunk
 from tests.external_dependency_unit.constants import TEST_TENANT_ID
 from tests.external_dependency_unit.document_index.conftest import EMBEDDING_DIM
 from tests.external_dependency_unit.document_index.conftest import make_chunk
@@ -203,25 +201,3 @@ class TestDocumentIndexNew:
             assert len(result_map) == 2
             assert result_map[existing_doc] is True
             assert result_map[new_doc] is False
-
-    def test_index_accepts_generator(
-        self,
-        document_indices: list[DocumentIndexNew],
-        tenant_context: None,  # noqa: ARG002
-    ) -> None:
-        """index() accepts a generator (any iterable), not just a list."""
-        for document_index in document_indices:
-            doc_id = f"test_gen_{uuid.uuid4().hex[:8]}"
-            metadata = make_indexing_metadata([doc_id], old_counts=[0], new_counts=[3])
-
-            def chunk_gen() -> Iterator[DocMetadataAwareIndexChunk]:
-                for i in range(3):
-                    yield make_chunk(doc_id, chunk_id=i)
-
-            results = document_index.index(
-                chunks=chunk_gen(), indexing_metadata=metadata
-            )
-
-            assert len(results) == 1
-            assert results[0].document_id == doc_id
-            assert results[0].already_existed is False

backend/tests/external_dependency_unit/document_index/test_document_index_old.py

@@ -5,7 +5,6 @@ These tests assume Vespa and OpenSearch are running.
 
 import time
 from collections.abc import Generator
-from collections.abc import Iterator
 
 import pytest
 
@@ -167,29 +166,3 @@ class TestDocumentIndexOld:
                 batch_retrieval=True,
             )
             assert len(inference_chunks) == 0
-
-    def test_index_accepts_generator(
-        self,
-        document_indices: list[DocumentIndex],
-        tenant_context: None,  # noqa: ARG002
-    ) -> None:
-        """index() accepts a generator (any iterable), not just a list."""
-        for document_index in document_indices:
-
-            def chunk_gen() -> Iterator[DocMetadataAwareIndexChunk]:
-                for i in range(3):
-                    yield make_chunk("test_doc_gen", chunk_id=i)
-
-            index_batch_params = IndexBatchParams(
-                doc_id_to_previous_chunk_cnt={"test_doc_gen": 0},
-                doc_id_to_new_chunk_cnt={"test_doc_gen": 3},
-                tenant_id=get_current_tenant_id(),
-                large_chunks_enabled=False,
-            )
-
-            results = document_index.index(chunk_gen(), index_batch_params)
-
-            assert len(results) == 1
-            record = results.pop()
-            assert record.document_id == "test_doc_gen"
-            assert record.already_existed is False

backend/tests/external_dependency_unit/tools/test_python_tool.py

@@ -1175,7 +1175,7 @@ def test_code_interpreter_receives_chat_files(
 
     file_descriptor: FileDescriptor = {
         "id": user_file.file_id,
-        "type": ChatFileType.CSV,
+        "type": ChatFileType.TABULAR,
         "name": "data.csv",
         "user_file_id": str(user_file.id),
     }

backend/tests/unit/onyx/background/celery/test_celery_redis.py

@@ -1,87 +0,0 @@
-"""Tests for celery_get_broker_client singleton."""
-
-from collections.abc import Iterator
-from unittest.mock import MagicMock
-from unittest.mock import patch
-
-import pytest
-
-from onyx.background.celery import celery_redis
-
-
-@pytest.fixture(autouse=True)
-def reset_singleton() -> Iterator[None]:
-    """Reset the module-level singleton between tests."""
-    celery_redis._broker_client = None
-    celery_redis._broker_url = None
-    yield
-    celery_redis._broker_client = None
-    celery_redis._broker_url = None
-
-
-def _make_mock_app(broker_url: str = "redis://localhost:6379/15") -> MagicMock:
-    app = MagicMock()
-    app.conf.broker_url = broker_url
-    return app
-
-
-class TestCeleryGetBrokerClient:
-    @patch("onyx.background.celery.celery_redis.Redis")
-    def test_creates_client_on_first_call(self, mock_redis_cls: MagicMock) -> None:
-        mock_client = MagicMock()
-        mock_redis_cls.from_url.return_value = mock_client
-
-        app = _make_mock_app()
-        result = celery_redis.celery_get_broker_client(app)
-
-        assert result is mock_client
-        call_args = mock_redis_cls.from_url.call_args
-        assert call_args[0][0] == "redis://localhost:6379/15"
-        assert call_args[1]["decode_responses"] is False
-        assert call_args[1]["socket_keepalive"] is True
-        assert call_args[1]["retry_on_timeout"] is True
-
-    @patch("onyx.background.celery.celery_redis.Redis")
-    def test_reuses_cached_client(self, mock_redis_cls: MagicMock) -> None:
-        mock_client = MagicMock()
-        mock_client.ping.return_value = True
-        mock_redis_cls.from_url.return_value = mock_client
-
-        app = _make_mock_app()
-        client1 = celery_redis.celery_get_broker_client(app)
-        client2 = celery_redis.celery_get_broker_client(app)
-
-        assert client1 is client2
-        # from_url called only once
-        assert mock_redis_cls.from_url.call_count == 1
-
-    @patch("onyx.background.celery.celery_redis.Redis")
-    def test_reconnects_on_ping_failure(self, mock_redis_cls: MagicMock) -> None:
-        stale_client = MagicMock()
-        stale_client.ping.side_effect = ConnectionError("disconnected")
-
-        fresh_client = MagicMock()
-        fresh_client.ping.return_value = True
-
-        mock_redis_cls.from_url.side_effect = [stale_client, fresh_client]
-
-        app = _make_mock_app()
-
-        # First call creates stale_client
-        client1 = celery_redis.celery_get_broker_client(app)
-        assert client1 is stale_client
-
-        # Second call: ping fails, creates fresh_client
-        client2 = celery_redis.celery_get_broker_client(app)
-        assert client2 is fresh_client
-        assert mock_redis_cls.from_url.call_count == 2
-
-    @patch("onyx.background.celery.celery_redis.Redis")
-    def test_uses_broker_url_from_app_config(self, mock_redis_cls: MagicMock) -> None:
-        mock_redis_cls.from_url.return_value = MagicMock()
-
-        app = _make_mock_app("redis://custom-host:6380/3")
-        celery_redis.celery_get_broker_client(app)
-
-        call_args = mock_redis_cls.from_url.call_args
-        assert call_args[0][0] == "redis://custom-host:6380/3"

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

@@ -139,7 +139,7 @@ def test_csv_file_type() -> None:
     result = _extract_referenced_file_descriptors([tool_call], message)
 
     assert len(result) == 1
-    assert result[0]["type"] == ChatFileType.CSV
+    assert result[0]["type"] == ChatFileType.TABULAR
 
 
 def test_unknown_extension_defaults_to_plain_text() -> None:

backend/tests/unit/onyx/tools/test_tool_runner_chat_files.py

@@ -82,7 +82,7 @@ class TestChatFileConversion:
             ChatLoadedFile(
                 file_id="file-2",
                 content=b"csv,data\n1,2",
-                file_type=ChatFileType.CSV,
+                file_type=ChatFileType.TABULAR,
                 filename="data.csv",
                 content_text="csv,data\n1,2",
                 token_count=5,

backend/tests/unit/server/metrics/test_indexing_pipeline_collectors.py

@@ -1,6 +1,5 @@
 """Tests for indexing pipeline Prometheus collectors."""
 
-from collections.abc import Iterator
 from datetime import datetime
 from datetime import timedelta
 from datetime import timezone
@@ -14,16 +13,6 @@ from onyx.server.metrics.indexing_pipeline import IndexAttemptCollector
 from onyx.server.metrics.indexing_pipeline import QueueDepthCollector
 
 
-@pytest.fixture(autouse=True)
-def _mock_broker_client() -> Iterator[None]:
-    """Patch celery_get_broker_client for all collector tests."""
-    with patch(
-        "onyx.background.celery.celery_redis.celery_get_broker_client",
-        return_value=MagicMock(),
-    ):
-        yield
-
-
 class TestQueueDepthCollector:
     def test_returns_empty_when_factory_not_set(self) -> None:
         collector = QueueDepthCollector()
@@ -35,7 +24,8 @@ class TestQueueDepthCollector:
 
     def test_collects_queue_depths(self) -> None:
         collector = QueueDepthCollector(cache_ttl=0)
-        collector.set_celery_app(MagicMock())
+        mock_redis = MagicMock()
+        collector.set_redis_factory(lambda: mock_redis)
 
         with (
             patch(
@@ -70,8 +60,8 @@ class TestQueueDepthCollector:
 
     def test_handles_redis_error_gracefully(self) -> None:
         collector = QueueDepthCollector(cache_ttl=0)
-        MagicMock()
-        collector.set_celery_app(MagicMock())
+        mock_redis = MagicMock()
+        collector.set_redis_factory(lambda: mock_redis)
 
         with patch(
             "onyx.server.metrics.indexing_pipeline.celery_get_queue_length",
@@ -84,8 +74,8 @@ class TestQueueDepthCollector:
 
     def test_caching_returns_stale_within_ttl(self) -> None:
         collector = QueueDepthCollector(cache_ttl=60)
-        MagicMock()
-        collector.set_celery_app(MagicMock())
+        mock_redis = MagicMock()
+        collector.set_redis_factory(lambda: mock_redis)
 
         with (
             patch(
@@ -108,10 +98,31 @@ class TestQueueDepthCollector:
 
         assert first is second  # Same object, from cache
 
+    def test_factory_called_each_scrape(self) -> None:
+        """Verify the Redis factory is called on each fresh collect, not cached."""
+        collector = QueueDepthCollector(cache_ttl=0)
+        factory = MagicMock(return_value=MagicMock())
+        collector.set_redis_factory(factory)
+
+        with (
+            patch(
+                "onyx.server.metrics.indexing_pipeline.celery_get_queue_length",
+                return_value=0,
+            ),
+            patch(
+                "onyx.server.metrics.indexing_pipeline.celery_get_unacked_task_ids",
+                return_value=set(),
+            ),
+        ):
+            collector.collect()
+            collector.collect()
+
+        assert factory.call_count == 2
+
     def test_error_returns_stale_cache(self) -> None:
         collector = QueueDepthCollector(cache_ttl=0)
-        MagicMock()
-        collector.set_celery_app(MagicMock())
+        mock_redis = MagicMock()
+        collector.set_redis_factory(lambda: mock_redis)
 
         # First call succeeds
         with (

backend/tests/unit/server/metrics/test_indexing_pipeline_setup.py

@@ -1,22 +1,96 @@
-"""Tests for indexing pipeline setup."""
+"""Tests for indexing pipeline setup (Redis factory caching)."""
 
 from unittest.mock import MagicMock
 
-from onyx.server.metrics.indexing_pipeline import QueueDepthCollector
-from onyx.server.metrics.indexing_pipeline import RedisHealthCollector
+from onyx.server.metrics.indexing_pipeline_setup import _make_broker_redis_factory
 
 
-class TestCollectorCeleryAppSetup:
-    def test_queue_depth_collector_uses_celery_app(self) -> None:
-        """QueueDepthCollector.set_celery_app stores the app for broker access."""
-        collector = QueueDepthCollector()
-        mock_app = MagicMock()
-        collector.set_celery_app(mock_app)
-        assert collector._celery_app is mock_app
+def _make_mock_app(client: MagicMock) -> MagicMock:
+    """Create a mock Celery app whose broker_connection().channel().client
+    returns the given client."""
+    mock_app = MagicMock()
+    mock_conn = MagicMock()
+    mock_conn.channel.return_value.client = client
 
-    def test_redis_health_collector_uses_celery_app(self) -> None:
-        """RedisHealthCollector.set_celery_app stores the app for broker access."""
-        collector = RedisHealthCollector()
-        mock_app = MagicMock()
-        collector.set_celery_app(mock_app)
-        assert collector._celery_app is mock_app
+    mock_app.broker_connection.return_value = mock_conn
+
+    return mock_app
+
+
+class TestMakeBrokerRedisFactory:
+    def test_caches_redis_client_across_calls(self) -> None:
+        """Factory should reuse the same client on subsequent calls."""
+        mock_client = MagicMock()
+        mock_client.ping.return_value = True
+        mock_app = _make_mock_app(mock_client)
+
+        factory = _make_broker_redis_factory(mock_app)
+
+        client1 = factory()
+        client2 = factory()
+
+        assert client1 is client2
+        # broker_connection should only be called once
+        assert mock_app.broker_connection.call_count == 1
+
+    def test_reconnects_when_ping_fails(self) -> None:
+        """Factory should create a new client if ping fails (stale connection)."""
+        mock_client_stale = MagicMock()
+        mock_client_stale.ping.side_effect = ConnectionError("disconnected")
+
+        mock_client_fresh = MagicMock()
+        mock_client_fresh.ping.return_value = True
+
+        mock_app = _make_mock_app(mock_client_stale)
+
+        factory = _make_broker_redis_factory(mock_app)
+
+        # First call — creates and caches
+        client1 = factory()
+        assert client1 is mock_client_stale
+        assert mock_app.broker_connection.call_count == 1
+
+        # Switch to fresh client for next connection
+        mock_conn_fresh = MagicMock()
+        mock_conn_fresh.channel.return_value.client = mock_client_fresh
+        mock_app.broker_connection.return_value = mock_conn_fresh
+
+        # Second call — ping fails on stale, reconnects
+        client2 = factory()
+        assert client2 is mock_client_fresh
+        assert mock_app.broker_connection.call_count == 2
+
+    def test_reconnect_closes_stale_client(self) -> None:
+        """When ping fails, the old client should be closed before reconnecting."""
+        mock_client_stale = MagicMock()
+        mock_client_stale.ping.side_effect = ConnectionError("disconnected")
+
+        mock_client_fresh = MagicMock()
+        mock_client_fresh.ping.return_value = True
+
+        mock_app = _make_mock_app(mock_client_stale)
+
+        factory = _make_broker_redis_factory(mock_app)
+
+        # First call — creates and caches
+        factory()
+
+        # Switch to fresh client
+        mock_conn_fresh = MagicMock()
+        mock_conn_fresh.channel.return_value.client = mock_client_fresh
+        mock_app.broker_connection.return_value = mock_conn_fresh
+
+        # Second call — ping fails, should close stale client
+        factory()
+        mock_client_stale.close.assert_called_once()
+
+    def test_first_call_creates_connection(self) -> None:
+        """First call should always create a new connection."""
+        mock_client = MagicMock()
+        mock_app = _make_mock_app(mock_client)
+
+        factory = _make_broker_redis_factory(mock_app)
+        client = factory()
+
+        assert client is mock_client
+        mock_app.broker_connection.assert_called_once()

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: 0.4.38
+version: 0.4.37
 appVersion: latest
 annotations:
   category: Productivity

deployment/helm/charts/onyx/templates/celery-worker-docfetching-metrics-service.yaml

@@ -1,26 +0,0 @@
-{{- /* Metrics port must match the default in metrics_server.py (_DEFAULT_PORTS).
-       Do NOT use PROMETHEUS_METRICS_PORT env var in Helm — each worker needs its own port. */ -}}
-{{- if and .Values.vectorDB.enabled (gt (int .Values.celery_worker_docfetching.replicaCount) 0) }}
-apiVersion: v1
-kind: Service
-metadata:
-  name: {{ include "onyx.fullname" . }}-celery-worker-docfetching-metrics
-  labels:
-    {{- include "onyx.labels" . | nindent 4 }}
-    {{- if .Values.celery_worker_docfetching.deploymentLabels }}
-    {{- toYaml .Values.celery_worker_docfetching.deploymentLabels | nindent 4 }}
-    {{- end }}
-    metrics: "true"
-spec:
-  type: ClusterIP
-  ports:
-    - port: 9092
-      targetPort: metrics
-      protocol: TCP
-      name: metrics
-  selector:
-    {{- include "onyx.selectorLabels" . | nindent 4 }}
-    {{- if .Values.celery_worker_docfetching.deploymentLabels }}
-    {{- toYaml .Values.celery_worker_docfetching.deploymentLabels | nindent 4 }}
-    {{- end }}
-{{- end }}

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

@@ -73,10 +73,6 @@ spec:
               "-Q",
               "connector_doc_fetching",
             ]
-          ports:
-            - name: metrics
-              containerPort: 9092
-              protocol: TCP
           resources:
             {{- toYaml .Values.celery_worker_docfetching.resources | nindent 12 }}
           envFrom:

deployment/helm/charts/onyx/templates/celery-worker-docprocessing-metrics-service.yaml

@@ -1,26 +0,0 @@
-{{- /* Metrics port must match the default in metrics_server.py (_DEFAULT_PORTS).
-       Do NOT use PROMETHEUS_METRICS_PORT env var in Helm — each worker needs its own port. */ -}}
-{{- if and .Values.vectorDB.enabled (gt (int .Values.celery_worker_docprocessing.replicaCount) 0) }}
-apiVersion: v1
-kind: Service
-metadata:
-  name: {{ include "onyx.fullname" . }}-celery-worker-docprocessing-metrics
-  labels:
-    {{- include "onyx.labels" . | nindent 4 }}
-    {{- if .Values.celery_worker_docprocessing.deploymentLabels }}
-    {{- toYaml .Values.celery_worker_docprocessing.deploymentLabels | nindent 4 }}
-    {{- end }}
-    metrics: "true"
-spec:
-  type: ClusterIP
-  ports:
-    - port: 9093
-      targetPort: metrics
-      protocol: TCP
-      name: metrics
-  selector:
-    {{- include "onyx.selectorLabels" . | nindent 4 }}
-    {{- if .Values.celery_worker_docprocessing.deploymentLabels }}
-    {{- toYaml .Values.celery_worker_docprocessing.deploymentLabels | nindent 4 }}
-    {{- end }}
-{{- end }}

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

@@ -73,10 +73,6 @@ spec:
               "-Q",
               "docprocessing",
             ]
-          ports:
-            - name: metrics
-              containerPort: 9093
-              protocol: TCP
           resources:
             {{- toYaml .Values.celery_worker_docprocessing.resources | nindent 12 }}
           envFrom:

deployment/helm/charts/onyx/templates/celery-worker-monitoring-metrics-service.yaml

@@ -1,26 +0,0 @@
-{{- /* Metrics port must match the default in metrics_server.py (_DEFAULT_PORTS).
-       Do NOT use PROMETHEUS_METRICS_PORT env var in Helm — each worker needs its own port. */ -}}
-{{- if and .Values.vectorDB.enabled (gt (int .Values.celery_worker_monitoring.replicaCount) 0) }}
-apiVersion: v1
-kind: Service
-metadata:
-  name: {{ include "onyx.fullname" . }}-celery-worker-monitoring-metrics
-  labels:
-    {{- include "onyx.labels" . | nindent 4 }}
-    {{- if .Values.celery_worker_monitoring.deploymentLabels }}
-    {{- toYaml .Values.celery_worker_monitoring.deploymentLabels | nindent 4 }}
-    {{- end }}
-    metrics: "true"
-spec:
-  type: ClusterIP
-  ports:
-    - port: 9096
-      targetPort: metrics
-      protocol: TCP
-      name: metrics
-  selector:
-    {{- include "onyx.selectorLabels" . | nindent 4 }}
-    {{- if .Values.celery_worker_monitoring.deploymentLabels }}
-    {{- toYaml .Values.celery_worker_monitoring.deploymentLabels | nindent 4 }}
-    {{- end }}
-{{- end }}

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

@@ -70,10 +70,6 @@ spec:
               "-Q",
               "monitoring",
             ]
-          ports:
-            - name: metrics
-              containerPort: 9096
-              protocol: TCP
           resources:
             {{- toYaml .Values.celery_worker_monitoring.resources | nindent 12 }}
           envFrom:

pyproject.toml

@@ -144,7 +144,7 @@ dev = [
     "matplotlib==3.10.8",
     "mypy-extensions==1.0.0",
     "mypy==1.13.0",
-    "onyx-devtools==0.7.2",
+    "onyx-devtools==0.7.1",
     "openapi-generator-cli==7.17.0",
     "pandas-stubs~=2.3.3",
     "pre-commit==3.2.2",

uv.lock

@@ -4458,7 +4458,7 @@ requires-dist = [
     { name = "numpy", marker = "extra == 'model-server'", specifier = "==2.4.1" },
     { name = "oauthlib", marker = "extra == 'backend'", specifier = "==3.2.2" },
     { name = "office365-rest-python-client", marker = "extra == 'backend'", specifier = "==2.6.2" },
-    { name = "onyx-devtools", marker = "extra == 'dev'", specifier = "==0.7.2" },
+    { name = "onyx-devtools", marker = "extra == 'dev'", specifier = "==0.7.1" },
     { name = "openai", specifier = "==2.14.0" },
     { name = "openapi-generator-cli", marker = "extra == 'dev'", specifier = "==7.17.0" },
     { name = "openinference-instrumentation", marker = "extra == 'backend'", specifier = "==0.1.42" },
@@ -4563,19 +4563,19 @@ requires-dist = [{ name = "onyx", extras = ["backend", "dev", "ee"], editable =
 
 [[package]]
 name = "onyx-devtools"
-version = "0.7.2"
+version = "0.7.1"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "fastapi" },
     { name = "openapi-generator-cli" },
 ]
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/22/b0/765ed49157470e8ccc8ab89e6a896ade50cde3aa2a494662ad4db92a48c4/onyx_devtools-0.7.2-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:553a2b5e61b29b7913c991c8d5aed78f930f0f81a0f42229c6a8de2b1e8ff57e", size = 4203859, upload-time = "2026-03-27T15:09:49.63Z" },
-    { url = "https://files.pythonhosted.org/packages/f7/9d/bba0a44a16d2fc27e5441aaf10727e10514e7a49bce70eca02bced566eb9/onyx_devtools-0.7.2-py3-none-macosx_11_0_arm64.whl", hash = "sha256:5cf0782dca8b3d861de9e18e65e990cfce5161cd559df44d8fabd3fefd54fdcd", size = 3879750, upload-time = "2026-03-27T15:09:42.413Z" },
-    { url = "https://files.pythonhosted.org/packages/4d/d8/c5725e8af14c74fe0aeed29e4746400bb3c0a078fd1240df729dc6432b84/onyx_devtools-0.7.2-py3-none-manylinux_2_17_aarch64.whl", hash = "sha256:9a0d67373e16b4fbb38a5290c0d9dfd4cfa837e5da0c165b32841b9d37f7455b", size = 3743529, upload-time = "2026-03-27T15:09:44.546Z" },
-    { url = "https://files.pythonhosted.org/packages/1a/82/b7c398a21dbc3e14fd7a29e49caa86b1bc0f8d7c75c051514785441ab779/onyx_devtools-0.7.2-py3-none-manylinux_2_17_x86_64.whl", hash = "sha256:794af14b2de575d0ae41b94551399eca8f8ba9b950c5db7acb7612767fd228f9", size = 4166562, upload-time = "2026-03-27T15:09:49.471Z" },
-    { url = "https://files.pythonhosted.org/packages/26/76/be129e2baafc91fe792d919b1f4d73fc943ba9c2b728a60f1fb98e0c115a/onyx_devtools-0.7.2-py3-none-win_amd64.whl", hash = "sha256:83b3eb84df58d865e4f714222a5fab3ea464836e2c8690569454a940bbb651ff", size = 4282270, upload-time = "2026-03-27T15:09:44.676Z" },
-    { url = "https://files.pythonhosted.org/packages/3b/72/29b8c8dbcf069c56475f00511f04c4aaa5ba3faba1dfc8276107d4b3ef7f/onyx_devtools-0.7.2-py3-none-win_arm64.whl", hash = "sha256:62f0836624ee6a5b31e64fd93162e7fce142ac8a4f959607e411824bc2b88174", size = 3823053, upload-time = "2026-03-27T15:09:43.546Z" },
+    { url = "https://files.pythonhosted.org/packages/65/9d/74bcd02583706bdf90c8ac9084eb60bd71d0671392152410ab21b7b68ea1/onyx_devtools-0.7.1-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:178385dce0b413fd2a1f761055a99f556ec536ef5c32963fc273e751813621eb", size = 4007974, upload-time = "2026-03-17T21:10:39.267Z" },
+    { url = "https://files.pythonhosted.org/packages/f0/f8/d8ddb32120428c083c60eb07244479da6e07eaebd31847658a049ab33815/onyx_devtools-0.7.1-py3-none-macosx_11_0_arm64.whl", hash = "sha256:7960ae6e440ebf1584e02d9e1d0c9ef543b1d54c2584cdcace15695aec3121b2", size = 3696924, upload-time = "2026-03-17T21:10:50.716Z" },
+    { url = "https://files.pythonhosted.org/packages/87/21/1e427280066db42ff9dd5f34c70b9dca5d9781f96d0d9a88aaa454fdb432/onyx_devtools-0.7.1-py3-none-manylinux_2_17_aarch64.whl", hash = "sha256:6785dda88ca0a3d8464a9bfab76a253ed90da89d53a9c4a67227980f37df1ccf", size = 3568300, upload-time = "2026-03-17T21:10:41.997Z" },
+    { url = "https://files.pythonhosted.org/packages/0e/0e/afbbe1164b3d016ddb5352353cb2541eef5a8b2c04e8f02d5d1319cb8b8c/onyx_devtools-0.7.1-py3-none-manylinux_2_17_x86_64.whl", hash = "sha256:9e77f2b725c0c00061a3dda5eba199404b51638cec0bf54fc7611fee1f26db34", size = 3974668, upload-time = "2026-03-17T21:10:43.879Z" },
+    { url = "https://files.pythonhosted.org/packages/8a/a5/22840643289ef4ca83931b7a79fba8f1db7e626b4b870d4b4f8206c4ff5f/onyx_devtools-0.7.1-py3-none-win_amd64.whl", hash = "sha256:de37daa0e4db9b5dccf94408a3422be4f821e380ab70081bd1032cec1e3c91e6", size = 4078640, upload-time = "2026-03-17T21:10:40.275Z" },
+    { url = "https://files.pythonhosted.org/packages/1e/c1/a0295506a521d9942b0f55523781a113e4555420d800a386d5a2eb46a7ad/onyx_devtools-0.7.1-py3-none-win_arm64.whl", hash = "sha256:ab88c53ebda6dff27350316b4ac9bd5f258cd586c2109971a9d976411e1e22ea", size = 3636787, upload-time = "2026-03-17T21:10:37.492Z" },
 ]
 
 [[package]]

web/lib/opal/src/layouts/content/ContentMd.tsx

@@ -73,15 +73,6 @@ interface ContentMdProps {
   /** When `true`, the title color hooks into `Interactive`'s `--interactive-foreground` variable. */
   withInteractive?: boolean;
 
-  /** Optional class name applied to the title element. */
-  titleClassName?: string;
-
-  /** Optional class name applied to the icon element. */
-  iconClassName?: string;
-
-  /** Content rendered below the description, indented to align with it. */
-  bottomChildren?: React.ReactNode;
-
   /** Ref forwarded to the root `<div>`. */
   ref?: React.Ref<HTMLDivElement>;
 }
@@ -155,9 +146,6 @@ function ContentMd({
   tag,
   sizePreset = "main-ui",
   withInteractive,
-  titleClassName,
-  iconClassName,
-  bottomChildren,
   ref,
 }: ContentMdProps) {
   const [editing, setEditing] = useState(false);
@@ -196,11 +184,7 @@ function ContentMd({
             style={{ minHeight: config.lineHeight }}
           >
             <Icon
-              className={cn(
-                "opal-content-md-icon",
-                config.iconColorClass,
-                iconClassName
-              )}
+              className={cn("opal-content-md-icon", config.iconColorClass)}
               style={{ width: config.iconSize, height: config.iconSize }}
             />
           </div>
@@ -243,8 +227,7 @@ function ContentMd({
                 "opal-content-md-title",
                 config.titleFont,
                 "text-text-04",
-                editable && "cursor-pointer",
-                titleClassName
+                editable && "cursor-pointer"
               )}
               title={toPlainString(title)}
               onClick={editable ? startEditing : undefined}
@@ -312,13 +295,6 @@ function ContentMd({
           {resolveStr(description)}
         </div>
       )}
-      {bottomChildren && (
-        <div
-          style={Icon ? { paddingLeft: config.descriptionIndent } : undefined}
-        >
-          {bottomChildren}
-        </div>
-      )}
     </div>
   );
 }

web/lib/opal/src/layouts/content/components.tsx

@@ -138,12 +138,6 @@ type MdContentProps = ContentBaseProps & {
   auxIcon?: "info-gray" | "info-blue" | "warning" | "error";
   /** Tag rendered beside the title. */
   tag?: TagProps;
-  /** Optional class name applied to the title element. */
-  titleClassName?: string;
-  /** Optional class name applied to the icon element. */
-  iconClassName?: string;
-  /** Content rendered below the description, indented to align with it. */
-  bottomChildren?: React.ReactNode;
 };
 
 /** ContentSm does not support descriptions or inline editing. */

web/src/app/admin/configuration/image-generation/ImageGenerationContent.tsx

@@ -2,6 +2,7 @@
 
 import { useState, useMemo, useEffect } from "react";
 import useSWR from "swr";
+import Text from "@/refresh-components/texts/Text";
 import { Select } from "@/refresh-components/cards";
 import { useCreateModal } from "@/refresh-components/contexts/ModalContext";
 import { toast } from "@/hooks/useToast";
@@ -23,9 +24,8 @@ import { ProviderIcon } from "@/app/admin/configuration/llm/ProviderIcon";
 import Message from "@/refresh-components/messages/Message";
 import ConfirmationModalLayout from "@/refresh-components/layouts/ConfirmationModalLayout";
 import InputSelect from "@/refresh-components/inputs/InputSelect";
-import { Button, Text } from "@opal/components";
+import { Button } from "@opal/components";
 import { SvgSlash, SvgUnplug } from "@opal/icons";
-import { markdown } from "@opal/utils";
 
 const NO_DEFAULT_VALUE = "__none__";
 
@@ -201,10 +201,10 @@ export default function ImageGenerationContent() {
       <div className="flex flex-col gap-6">
         {/* Section Header */}
         <div className="flex flex-col gap-0.5">
-          <Text font="main-content-emphasis" color="text-05">
+          <Text mainContentEmphasis text05>
             Image Generation Model
           </Text>
-          <Text font="secondary-body" color="text-03">
+          <Text secondaryBody text03>
             Select a model to generate images in chat.
           </Text>
         </div>
@@ -223,7 +223,7 @@ export default function ImageGenerationContent() {
         {/* Provider Groups */}
         {IMAGE_PROVIDER_GROUPS.map((group) => (
           <div key={group.name} className="flex flex-col gap-2">
-            <Text font="secondary-body" color="text-03">
+            <Text secondaryBody text03>
               {group.name}
             </Text>
             <div className="flex flex-col gap-2">
@@ -277,13 +277,12 @@ export default function ImageGenerationContent() {
           {needsReplacement ? (
             hasReplacements ? (
               <Section alignItems="start">
-                <Text as="p" color="text-03">
-                  {markdown(
-                    `**${disconnectProvider.title}** is currently the default image generation model. Session history will be preserved.`
-                  )}
+                <Text as="p" text03>
+                  <b>{disconnectProvider.title}</b> is currently the default
+                  image generation model. Session history will be preserved.
                 </Text>
                 <Section alignItems="start" gap={0.25}>
-                  <Text as="p" color="text-04">
+                  <Text as="p" text04>
                     Set New Default
                   </Text>
                   <InputSelect
@@ -330,24 +329,22 @@ export default function ImageGenerationContent() {
               </Section>
             ) : (
               <>
-                <Text as="p" color="text-03">
-                  {markdown(
-                    `**${disconnectProvider.title}** is currently the default image generation model.`
-                  )}
+                <Text as="p" text03>
+                  <b>{disconnectProvider.title}</b> is currently the default
+                  image generation model.
                 </Text>
-                <Text as="p" color="text-03">
+                <Text as="p" text03>
                   Connect another provider to continue using image generation.
                 </Text>
               </>
             )
           ) : (
             <>
-              <Text as="p" color="text-03">
-                {markdown(
-                  `**${disconnectProvider.title}** models will no longer be used to generate images.`
-                )}
+              <Text as="p" text03>
+                <b>{disconnectProvider.title}</b> models will no longer be used
+                to generate images.
               </Text>
-              <Text as="p" color="text-03">
+              <Text as="p" text03>
                 Session history will be preserved.
               </Text>
             </>

web/src/app/admin/configuration/web-search/page.tsx

@@ -15,7 +15,7 @@ import { Callout } from "@/components/ui/callout";
 import { cn } from "@/lib/utils";
 import { toast } from "@/hooks/useToast";
 import { SvgGlobe, SvgOnyxLogo, SvgSlash, SvgUnplug } from "@opal/icons";
-import { Button } from "@opal/components";
+import { Button as OpalButton } from "@opal/components";
 import { ADMIN_ROUTES } from "@/lib/admin-routes";
 import { WebProviderSetupModal } from "@/app/admin/configuration/web-search/WebProviderSetupModal";
 import ConfirmationModalLayout from "@/refresh-components/layouts/ConfirmationModalLayout";
@@ -151,7 +151,7 @@ function WebSearchDisconnectModal({
       description="This will remove the stored credentials for this provider."
       onClose={onClose}
       submit={
-        <Button
+        <OpalButton
           variant="danger"
           onClick={onDisconnect}
           disabled={
@@ -159,7 +159,7 @@ function WebSearchDisconnectModal({
           }
         >
           Disconnect
-        </Button>
+        </OpalButton>
       }
     >
       {needsReplacement ? (

web/src/app/app/interfaces.ts

@@ -75,14 +75,14 @@ export enum ChatFileType {
   IMAGE = "image",
   DOCUMENT = "document",
   PLAIN_TEXT = "plain_text",
-  CSV = "csv",
+  TABULAR = "tabular",
   USER_KNOWLEDGE = "user_knowledge",
 }
 
 export const isTextFile = (fileType: ChatFileType) =>
   [
     ChatFileType.PLAIN_TEXT,
-    ChatFileType.CSV,
+    ChatFileType.TABULAR,
     ChatFileType.USER_KNOWLEDGE,
     ChatFileType.DOCUMENT,
   ].includes(fileType);

web/src/app/app/message/FileDisplay.tsx

@@ -42,7 +42,7 @@ export default function FileDisplay({ files }: FileDisplayProps) {
       file.type === ChatFileType.DOCUMENT
   );
   const imageFiles = files.filter((file) => file.type === ChatFileType.IMAGE);
-  const csvFiles = files.filter((file) => file.type === ChatFileType.CSV);
+  const csvFiles = files.filter((file) => file.type === ChatFileType.TABULAR);
 
   const presentingDocument: MinimalOnyxDocument = {
     document_id: previewingFile?.id ?? "",

web/src/refresh-components/Popover.tsx

@@ -142,7 +142,6 @@ function PopoverContent({
         collisionPadding={8}
         className={cn(
           "bg-background-neutral-00 p-1 z-popover rounded-12 border shadow-md data-[state=open]:animate-in data-[state=closed]:animate-out data-[state=closed]:fade-out-0 data-[state=open]:fade-in-0 data-[state=closed]:zoom-out-95 data-[state=open]:zoom-in-95 data-[side=bottom]:slide-in-from-top-2 data-[side=left]:slide-in-from-right-2 data-[side=right]:slide-in-from-left-2 data-[side=top]:slide-in-from-bottom-2",
-          "flex flex-col",
           "max-h-[var(--radix-popover-content-available-height)]",
           "overflow-hidden",
           widthClasses[width]
@@ -227,7 +226,7 @@ export function PopoverMenu({
   });
 
   return (
-    <Section alignItems="stretch" height="auto" className="flex-1 min-h-0">
+    <Section alignItems="stretch">
       <ShadowDiv
         scrollContainerRef={scrollContainerRef}
         className="flex flex-col gap-1 max-h-[20rem] w-full"

web/src/refresh-components/ShadowDiv.tsx

@@ -105,7 +105,7 @@ export default function ShadowDiv({
   }, [containerRef, checkScroll]);
 
   return (
-    <div className="relative min-h-0 flex flex-col">
+    <div className="relative min-h-0">
       <div
         ref={containerRef}
         className={cn("overflow-y-auto", className)}

web/src/refresh-pages/admin/HooksPage/ConnectedHookCard.tsx

@@ -1,420 +0,0 @@
-"use client";
-
-import { useState } from "react";
-import { toast } from "@/hooks/useToast";
-import { Button } from "@opal/components";
-import { Disabled } from "@opal/core";
-import { cn } from "@/lib/utils";
-import { ContentAction } from "@opal/layouts";
-import Card from "@/refresh-components/cards/Card";
-import Text from "@/refresh-components/texts/Text";
-import { Section } from "@/layouts/general-layouts";
-import {
-  SvgCheckCircle,
-  SvgExternalLink,
-  SvgPlug,
-  SvgRefreshCw,
-  SvgSettings,
-  SvgTrash,
-  SvgUnplug,
-} from "@opal/icons";
-import Modal, { BasicModalFooter } from "@/refresh-components/Modal";
-import type {
-  HookPointMeta,
-  HookResponse,
-} from "@/refresh-pages/admin/HooksPage/interfaces";
-import {
-  activateHook,
-  deactivateHook,
-  deleteHook,
-  validateHook,
-} from "@/refresh-pages/admin/HooksPage/svc";
-import { getHookPointIcon } from "@/refresh-pages/admin/HooksPage/hookPointIcons";
-
-// ---------------------------------------------------------------------------
-// Sub-component: disconnect confirmation modal
-// ---------------------------------------------------------------------------
-
-interface DisconnectConfirmModalProps {
-  open: boolean;
-  onOpenChange: (open: boolean) => void;
-  hook: HookResponse;
-  onDisconnect: () => void;
-  onDisconnectAndDelete: () => void;
-}
-
-function DisconnectConfirmModal({
-  open,
-  onOpenChange,
-  hook,
-  onDisconnect,
-  onDisconnectAndDelete,
-}: DisconnectConfirmModalProps) {
-  return (
-    <Modal open={open} onOpenChange={onOpenChange}>
-      <Modal.Content width="md" height="fit">
-        <Modal.Header
-          icon={(props) => (
-            <SvgUnplug {...props} className="text-action-danger-05" />
-          )}
-          title={`Disconnect ${hook.name}`}
-          onClose={() => onOpenChange(false)}
-        />
-        <Modal.Body>
-          <div className="flex flex-col gap-4">
-            <Text mainUiBody text03>
-              Onyx will stop calling this endpoint for hook{" "}
-              <strong>
-                <em>{hook.name}</em>
-              </strong>
-              . In-flight requests will continue to run. The external endpoint
-              may still retain data previously sent to it. You can reconnect
-              this hook later if needed.
-            </Text>
-            <Text mainUiBody text03>
-              You can also delete this hook. Deletion cannot be undone.
-            </Text>
-          </div>
-        </Modal.Body>
-        <Modal.Footer>
-          <BasicModalFooter
-            cancel={
-              <Button
-                prominence="secondary"
-                onClick={() => onOpenChange(false)}
-              >
-                Cancel
-              </Button>
-            }
-            submit={
-              <div className="flex items-center gap-2">
-                <Button
-                  variant="danger"
-                  prominence="secondary"
-                  onClick={onDisconnectAndDelete}
-                >
-                  Disconnect &amp; Delete
-                </Button>
-                <Button
-                  variant="danger"
-                  prominence="primary"
-                  onClick={onDisconnect}
-                >
-                  Disconnect
-                </Button>
-              </div>
-            }
-          />
-        </Modal.Footer>
-      </Modal.Content>
-    </Modal>
-  );
-}
-
-// ---------------------------------------------------------------------------
-// Sub-component: delete confirmation modal
-// ---------------------------------------------------------------------------
-
-interface DeleteConfirmModalProps {
-  open: boolean;
-  onOpenChange: (open: boolean) => void;
-  hook: HookResponse;
-  onDelete: () => void;
-}
-
-function DeleteConfirmModal({
-  open,
-  onOpenChange,
-  hook,
-  onDelete,
-}: DeleteConfirmModalProps) {
-  return (
-    <Modal open={open} onOpenChange={onOpenChange}>
-      <Modal.Content width="md" height="fit">
-        <Modal.Header
-          icon={(props) => (
-            <SvgTrash {...props} className="text-action-danger-05" />
-          )}
-          title={`Delete ${hook.name}`}
-          onClose={() => onOpenChange(false)}
-        />
-        <Modal.Body>
-          <div className="flex flex-col gap-4">
-            <Text mainUiBody text03>
-              Hook{" "}
-              <strong>
-                <em>{hook.name}</em>
-              </strong>{" "}
-              will be permanently removed from this hook point. The external
-              endpoint may still retain data previously sent to it.
-            </Text>
-            <Text mainUiBody text03>
-              Deletion cannot be undone.
-            </Text>
-          </div>
-        </Modal.Body>
-        <Modal.Footer>
-          <BasicModalFooter
-            cancel={
-              <Button
-                prominence="secondary"
-                onClick={() => onOpenChange(false)}
-              >
-                Cancel
-              </Button>
-            }
-            submit={
-              <Button variant="danger" prominence="primary" onClick={onDelete}>
-                Delete
-              </Button>
-            }
-          />
-        </Modal.Footer>
-      </Modal.Content>
-    </Modal>
-  );
-}
-
-// ---------------------------------------------------------------------------
-// ConnectedHookCard
-// ---------------------------------------------------------------------------
-
-export interface ConnectedHookCardProps {
-  hook: HookResponse;
-  spec: HookPointMeta | undefined;
-  onEdit: () => void;
-  onDeleted: () => void;
-  onToggled: (updated: HookResponse) => void;
-}
-
-export default function ConnectedHookCard({
-  hook,
-  spec,
-  onEdit,
-  onDeleted,
-  onToggled,
-}: ConnectedHookCardProps) {
-  const [isBusy, setIsBusy] = useState(false);
-  const [disconnectConfirmOpen, setDisconnectConfirmOpen] = useState(false);
-  const [deleteConfirmOpen, setDeleteConfirmOpen] = useState(false);
-
-  async function handleDelete() {
-    setDeleteConfirmOpen(false);
-    setIsBusy(true);
-    try {
-      await deleteHook(hook.id);
-      onDeleted();
-    } catch (err) {
-      console.error("Failed to delete hook:", err);
-      toast.error(
-        err instanceof Error ? err.message : "Failed to delete hook."
-      );
-    } finally {
-      setIsBusy(false);
-    }
-  }
-
-  async function handleActivate() {
-    setIsBusy(true);
-    try {
-      const updated = await activateHook(hook.id);
-      onToggled(updated);
-    } catch (err) {
-      console.error("Failed to reconnect hook:", err);
-      toast.error(
-        err instanceof Error ? err.message : "Failed to reconnect hook."
-      );
-    } finally {
-      setIsBusy(false);
-    }
-  }
-
-  async function handleDeactivate() {
-    setDisconnectConfirmOpen(false);
-    setIsBusy(true);
-    try {
-      const updated = await deactivateHook(hook.id);
-      onToggled(updated);
-    } catch (err) {
-      console.error("Failed to deactivate hook:", err);
-      toast.error(
-        err instanceof Error ? err.message : "Failed to deactivate hook."
-      );
-    } finally {
-      setIsBusy(false);
-    }
-  }
-
-  async function handleDisconnectAndDelete() {
-    setDisconnectConfirmOpen(false);
-    setIsBusy(true);
-    try {
-      const deactivated = await deactivateHook(hook.id);
-      onToggled(deactivated);
-      await deleteHook(hook.id);
-      onDeleted();
-    } catch (err) {
-      console.error("Failed to disconnect hook:", err);
-      toast.error(
-        err instanceof Error ? err.message : "Failed to disconnect hook."
-      );
-    } finally {
-      setIsBusy(false);
-    }
-  }
-
-  async function handleValidate() {
-    setIsBusy(true);
-    try {
-      const result = await validateHook(hook.id);
-      if (result.status === "passed") {
-        toast.success("Hook validated successfully.");
-      } else {
-        toast.error(
-          result.error_message ?? `Validation failed: ${result.status}`
-        );
-      }
-    } catch (err) {
-      console.error("Failed to validate hook:", err);
-      toast.error(
-        err instanceof Error ? err.message : "Failed to validate hook."
-      );
-    } finally {
-      setIsBusy(false);
-    }
-  }
-
-  const HookIcon = getHookPointIcon(hook.hook_point);
-
-  return (
-    <>
-      <DisconnectConfirmModal
-        open={disconnectConfirmOpen}
-        onOpenChange={setDisconnectConfirmOpen}
-        hook={hook}
-        onDisconnect={handleDeactivate}
-        onDisconnectAndDelete={handleDisconnectAndDelete}
-      />
-      <DeleteConfirmModal
-        open={deleteConfirmOpen}
-        onOpenChange={setDeleteConfirmOpen}
-        hook={hook}
-        onDelete={handleDelete}
-      />
-      <Card
-        variant="primary"
-        padding={0.5}
-        gap={0}
-        className={cn(
-          "hover:border-border-02",
-          !hook.is_active && "!bg-background-neutral-02"
-        )}
-      >
-        <ContentAction
-          sizePreset="main-ui"
-          variant="section"
-          paddingVariant="sm"
-          icon={HookIcon}
-          title={hook.name}
-          titleClassName={!hook.is_active ? "line-through" : undefined}
-          iconClassName="text-text-04"
-          description={`Hook Point: ${spec?.display_name ?? hook.hook_point}`}
-          bottomChildren={
-            spec?.docs_url ? (
-              <a
-                href={spec.docs_url}
-                target="_blank"
-                rel="noopener noreferrer"
-                className="flex items-center gap-1 w-fit font-secondary-body text-text-03"
-              >
-                <span className="underline">Documentation</span>
-                <SvgExternalLink size={12} className="shrink-0" />
-              </a>
-            ) : undefined
-          }
-          rightChildren={
-            <Section
-              flexDirection="column"
-              alignItems="end"
-              width="fit"
-              height="fit"
-              gap={0}
-            >
-              <div className="flex items-center gap-1 p-2">
-                {hook.is_active ? (
-                  <>
-                    <Text mainUiAction text03>
-                      Connected
-                    </Text>
-                    <SvgCheckCircle
-                      size={16}
-                      className="text-status-success-05"
-                    />
-                  </>
-                ) : (
-                  <div
-                    className={cn(
-                      "flex items-center gap-1",
-                      isBusy
-                        ? "opacity-50 pointer-events-none"
-                        : "cursor-pointer"
-                    )}
-                    onClick={handleActivate}
-                  >
-                    <Text mainUiAction text03>
-                      Reconnect
-                    </Text>
-                    <SvgPlug size={16} className="text-text-03 shrink-0" />
-                  </div>
-                )}
-              </div>
-              <Disabled disabled={isBusy}>
-                {/* Plain div instead of Section: Section applies style={{ padding }} inline which
-                    overrides Tailwind padding classes, making per-side padding (pl/pr/pb) ineffective. */}
-                <div className="flex items-center gap-0.5 pl-1 pr-1 pb-1">
-                  {hook.is_active ? (
-                    <>
-                      <Button
-                        prominence="tertiary"
-                        size="sm"
-                        icon={SvgUnplug}
-                        onClick={() => setDisconnectConfirmOpen(true)}
-                        tooltip="Disconnect Hook"
-                        aria-label="Deactivate hook"
-                      />
-                      <Button
-                        prominence="tertiary"
-                        size="sm"
-                        icon={SvgRefreshCw}
-                        onClick={handleValidate}
-                        tooltip="Test Connection"
-                        aria-label="Re-validate hook"
-                      />
-                    </>
-                  ) : (
-                    <Button
-                      prominence="tertiary"
-                      size="sm"
-                      icon={SvgTrash}
-                      onClick={() => setDeleteConfirmOpen(true)}
-                      tooltip="Delete"
-                      aria-label="Delete hook"
-                    />
-                  )}
-                  <Button
-                    prominence="tertiary"
-                    size="sm"
-                    icon={SvgSettings}
-                    onClick={onEdit}
-                    tooltip="Manage"
-                    aria-label="Configure hook"
-                  />
-                </div>
-              </Disabled>
-            </Section>
-          }
-        />
-      </Card>
-    </>
-  );
-}

web/src/refresh-pages/admin/HooksPage/HooksContent.tsx

@@ -1,211 +1,117 @@
 "use client";
 
-import { useState } from "react";
+import { useState, useEffect } from "react";
+import { toast } from "@/hooks/useToast";
 import { useHookSpecs } from "@/hooks/useHookSpecs";
-import { useHooks } from "@/hooks/useHooks";
 import SimpleLoader from "@/refresh-components/loaders/SimpleLoader";
-import { Button } from "@opal/components";
 import { ContentAction } from "@opal/layouts";
+import { Button } from "@opal/components";
 import InputSearch from "@/refresh-components/inputs/InputSearch";
 import Card from "@/refresh-components/cards/Card";
 import Text from "@/refresh-components/texts/Text";
-import { SvgArrowExchange, SvgExternalLink } from "@opal/icons";
-import HookFormModal from "@/refresh-pages/admin/HooksPage/HookFormModal";
-import ConnectedHookCard from "@/refresh-pages/admin/HooksPage/ConnectedHookCard";
-import { getHookPointIcon } from "@/refresh-pages/admin/HooksPage/hookPointIcons";
-import type {
-  HookPointMeta,
-  HookResponse,
-} from "@/refresh-pages/admin/HooksPage/interfaces";
+import {
+  SvgArrowExchange,
+  SvgBubbleText,
+  SvgExternalLink,
+  SvgFileBroadcast,
+  SvgHookNodes,
+} from "@opal/icons";
+import { IconFunctionComponent } from "@opal/types";
 
-// ---------------------------------------------------------------------------
-// Main component
-// ---------------------------------------------------------------------------
+const HOOK_POINT_ICONS: Record<string, IconFunctionComponent> = {
+  document_ingestion: SvgFileBroadcast,
+  query_processing: SvgBubbleText,
+};
+
+function getHookPointIcon(hookPoint: string): IconFunctionComponent {
+  return HOOK_POINT_ICONS[hookPoint] ?? SvgHookNodes;
+}
 
 export default function HooksContent() {
   const [search, setSearch] = useState("");
-  const [connectSpec, setConnectSpec] = useState<HookPointMeta | null>(null);
-  const [editHook, setEditHook] = useState<HookResponse | null>(null);
 
-  const { specs, isLoading: specsLoading, error: specsError } = useHookSpecs();
-  const {
-    hooks,
-    isLoading: hooksLoading,
-    error: hooksError,
-    mutate,
-  } = useHooks();
+  const { specs, isLoading, error } = useHookSpecs();
 
-  if (specsLoading || hooksLoading) {
+  useEffect(() => {
+    if (error) {
+      toast.error("Failed to load hook specifications.");
+    }
+  }, [error]);
+
+  if (isLoading) {
     return <SimpleLoader />;
   }
 
-  if (specsError || hooksError) {
+  if (error) {
     return (
       <Text text03 secondaryBody>
-        Failed to load{specsError ? " hook specifications" : " hooks"}. Please
-        refresh the page.
+        Failed to load hook specifications. Please refresh the page.
       </Text>
     );
   }
 
-  const hooksByPoint: Record<string, HookResponse[]> = {};
-  for (const hook of hooks ?? []) {
-    (hooksByPoint[hook.hook_point] ??= []).push(hook);
-  }
-
-  const searchLower = search.toLowerCase();
-
-  // Connected hooks sorted alphabetically by hook name
-  const connectedHooks = (hooks ?? [])
-    .filter(
-      (hook) =>
-        !searchLower ||
-        hook.name.toLowerCase().includes(searchLower) ||
-        specs
-          ?.find((s) => s.hook_point === hook.hook_point)
-          ?.display_name.toLowerCase()
-          .includes(searchLower)
-    )
-    .sort((a, b) => a.name.localeCompare(b.name));
-
-  // Unconnected hook point specs sorted alphabetically
-  const unconnectedSpecs = (specs ?? [])
-    .filter(
-      (spec) =>
-        (hooksByPoint[spec.hook_point]?.length ?? 0) === 0 &&
-        (!searchLower ||
-          spec.display_name.toLowerCase().includes(searchLower) ||
-          spec.description.toLowerCase().includes(searchLower))
-    )
-    .sort((a, b) => a.display_name.localeCompare(b.display_name));
-
-  function handleHookSuccess(updated: HookResponse) {
-    mutate((prev) => {
-      if (!prev) return [updated];
-      const idx = prev.findIndex((h) => h.id === updated.id);
-      if (idx >= 0) {
-        const next = [...prev];
-        next[idx] = updated;
-        return next;
-      }
-      return [...prev, updated];
-    });
-  }
-
-  function handleHookDeleted(id: number) {
-    mutate((prev) => prev?.filter((h) => h.id !== id));
-  }
-
-  const connectSpec_ =
-    connectSpec ??
-    (editHook
-      ? specs?.find((s) => s.hook_point === editHook.hook_point)
-      : undefined);
+  const filtered = (specs ?? []).filter(
+    (spec) =>
+      spec.display_name.toLowerCase().includes(search.toLowerCase()) ||
+      spec.description.toLowerCase().includes(search.toLowerCase())
+  );
 
   return (
-    <>
-      <div className="flex flex-col gap-6">
-        <InputSearch
-          placeholder="Search hooks..."
-          value={search}
-          onChange={(e) => setSearch(e.target.value)}
-        />
+    <div className="flex flex-col gap-6">
+      <InputSearch
+        placeholder="Search hooks..."
+        value={search}
+        onChange={(e) => setSearch(e.target.value)}
+      />
 
-        <div className="flex flex-col gap-2">
-          {connectedHooks.length === 0 && unconnectedSpecs.length === 0 ? (
-            <Text text03 secondaryBody>
-              {search
-                ? "No hooks match your search."
-                : "No hook points are available."}
-            </Text>
-          ) : (
-            <>
-              {connectedHooks.map((hook) => {
-                const spec = specs?.find(
-                  (s) => s.hook_point === hook.hook_point
-                );
-                return (
-                  <ConnectedHookCard
-                    key={hook.id}
-                    hook={hook}
-                    spec={spec}
-                    onEdit={() => setEditHook(hook)}
-                    onDeleted={() => handleHookDeleted(hook.id)}
-                    onToggled={handleHookSuccess}
-                  />
-                );
-              })}
-              {unconnectedSpecs.map((spec) => {
-                const UnconnectedIcon = getHookPointIcon(spec.hook_point);
-                return (
-                  <Card
-                    key={spec.hook_point}
-                    variant="secondary"
-                    padding={0.5}
-                    gap={0}
-                    className="hover:border-border-02"
+      <div className="flex flex-col gap-2">
+        {filtered.length === 0 ? (
+          <Text text03 secondaryBody>
+            {search
+              ? "No hooks match your search."
+              : "No hook points are available."}
+          </Text>
+        ) : (
+          filtered.map((spec) => (
+            <Card
+              key={spec.hook_point}
+              variant="secondary"
+              padding={0.5}
+              gap={0}
+            >
+              <ContentAction
+                icon={getHookPointIcon(spec.hook_point)}
+                title={spec.display_name}
+                description={spec.description}
+                sizePreset="main-content"
+                variant="section"
+                paddingVariant="fit"
+                rightChildren={
+                  // TODO(Bo-Onyx): wire up Connect — open modal to create/edit hook
+                  <Button prominence="tertiary" rightIcon={SvgArrowExchange}>
+                    Connect
+                  </Button>
+                }
+              />
+              {spec.docs_url && (
+                <div className="pl-7 pt-1">
+                  <a
+                    href={spec.docs_url}
+                    target="_blank"
+                    rel="noopener noreferrer"
+                    className="flex items-center gap-1 w-fit text-text-03"
                   >
-                    <ContentAction
-                      sizePreset="main-ui"
-                      variant="section"
-                      paddingVariant="sm"
-                      icon={UnconnectedIcon}
-                      title={spec.display_name}
-                      iconClassName="text-text-04"
-                      description={spec.description}
-                      bottomChildren={
-                        spec.docs_url ? (
-                          <a
-                            href={spec.docs_url}
-                            target="_blank"
-                            rel="noopener noreferrer"
-                            className="flex items-center gap-1 w-fit font-secondary-body text-text-03"
-                          >
-                            <span className="underline">Documentation</span>
-                            <SvgExternalLink size={12} className="shrink-0" />
-                          </a>
-                        ) : undefined
-                      }
-                      rightChildren={
-                        <Button
-                          prominence="tertiary"
-                          rightIcon={SvgArrowExchange}
-                          onClick={() => setConnectSpec(spec)}
-                        >
-                          Connect
-                        </Button>
-                      }
-                    />
-                  </Card>
-                );
-              })}
-            </>
-          )}
-        </div>
+                    <Text as="span" secondaryBody text03 className="underline">
+                      Documentation
+                    </Text>
+                    <SvgExternalLink size={16} className="text-text-02" />
+                  </a>
+                </div>
+              )}
+            </Card>
+          ))
+        )}
       </div>
-
-      {/* Create modal */}
-      <HookFormModal
-        key={connectSpec?.hook_point ?? "create"}
-        open={!!connectSpec}
-        onOpenChange={(open) => {
-          if (!open) setConnectSpec(null);
-        }}
-        spec={connectSpec ?? undefined}
-        onSuccess={handleHookSuccess}
-      />
-
-      {/* Edit modal */}
-      <HookFormModal
-        key={editHook?.id ?? "edit"}
-        open={!!editHook}
-        onOpenChange={(open) => {
-          if (!open) setEditHook(null);
-        }}
-        hook={editHook ?? undefined}
-        spec={connectSpec_ ?? undefined}
-        onSuccess={handleHookSuccess}
-      />
-    </>
+    </div>
   );
 }

web/src/refresh-pages/admin/HooksPage/hookPointIcons.ts

@@ -1,13 +0,0 @@
-import { SvgBubbleText, SvgFileBroadcast, SvgHookNodes } from "@opal/icons";
-import type { IconFunctionComponent } from "@opal/types";
-
-const HOOK_POINT_ICONS: Record<string, IconFunctionComponent> = {
-  document_ingestion: SvgFileBroadcast,
-  query_processing: SvgBubbleText,
-};
-
-function getHookPointIcon(hookPoint: string): IconFunctionComponent {
-  return HOOK_POINT_ICONS[hookPoint] ?? SvgHookNodes;
-}
-
-export { HOOK_POINT_ICONS, getHookPointIcon };

web/src/refresh-pages/admin/VoiceConfigurationPage.tsx

@@ -7,6 +7,7 @@ import {
   IconProps,
   OpenAIIcon,
 } from "@/components/icons/icons";
+import Text from "@/refresh-components/texts/Text";
 import { Select } from "@/refresh-components/cards";
 import Message from "@/refresh-components/messages/Message";
 import * as SettingsLayouts from "@/layouts/settings-layouts";
@@ -25,8 +26,7 @@ import { toast } from "@/hooks/useToast";
 import { Callout } from "@/components/ui/callout";
 import { Content } from "@opal/layouts";
 import { SvgMicrophone, SvgSlash, SvgUnplug } from "@opal/icons";
-import { Button, Text } from "@opal/components";
-import { markdown } from "@opal/utils";
+import { Button as OpalButton } from "@opal/components";
 import ConfirmationModalLayout from "@/refresh-components/layouts/ConfirmationModalLayout";
 import { Section } from "@/layouts/general-layouts";
 import { ADMIN_ROUTES } from "@/lib/admin-routes";
@@ -205,7 +205,7 @@ function VoiceDisconnectModal({
       description="Voice models"
       onClose={onClose}
       submit={
-        <Button
+        <OpalButton
           variant="danger"
           onClick={onDisconnect}
           disabled={
@@ -213,19 +213,19 @@ function VoiceDisconnectModal({
           }
         >
           Disconnect
-        </Button>
+        </OpalButton>
       }
     >
       {needsReplacement ? (
         hasReplacements ? (
           <Section alignItems="start">
-            <Text as="p" color="text-03">
-              {markdown(
-                `**${disconnectTarget.providerLabel}** models will no longer be used for speech-to-text or text-to-speech, and it will no longer be your default. Session history will be preserved.`
-              )}
+            <Text as="p" text03>
+              <b>{disconnectTarget.providerLabel}</b> models will no longer be
+              used for speech-to-text or text-to-speech, and it will no longer
+              be your default. Session history will be preserved.
             </Text>
             <Section alignItems="start" gap={0.25}>
-              <Text as="p" color="text-04">
+              <Text as="p" text04>
                 Set New Default
               </Text>
               <InputSelect
@@ -256,24 +256,23 @@ function VoiceDisconnectModal({
           </Section>
         ) : (
           <>
-            <Text as="p" color="text-03">
-              {markdown(
-                `**${disconnectTarget.providerLabel}** models will no longer be used for speech-to-text or text-to-speech, and it will no longer be your default.`
-              )}
+            <Text as="p" text03>
+              <b>{disconnectTarget.providerLabel}</b> models will no longer be
+              used for speech-to-text or text-to-speech, and it will no longer
+              be your default.
             </Text>
-            <Text as="p" color="text-03">
+            <Text as="p" text03>
               Connect another provider to continue using voice.
             </Text>
           </>
         )
       ) : (
         <>
-          <Text as="p" color="text-03">
-            {markdown(
-              `**${disconnectTarget.providerLabel}** models will no longer be available for voice.`
-            )}
+          <Text as="p" text03>
+            <b>{disconnectTarget.providerLabel}</b> models will no longer be
+            available for voice.
           </Text>
-          <Text as="p" color="text-03">
+          <Text as="p" text03>
             Session history will be preserved.
           </Text>
         </>
@@ -537,7 +536,7 @@ export default function VoiceConfigurationPage() {
           <Callout type="danger" title="Failed to load voice settings">
             {message}
             {detail && (
-              <Text as="p" font="main-content-body" color="text-03">
+              <Text as="p" mainContentBody text03>
                 {detail}
               </Text>
             )}
@@ -627,7 +626,7 @@ export default function VoiceConfigurationPage() {
 
           {TTS_PROVIDER_GROUPS.map((group) => (
             <div key={group.providerType} className="flex flex-col gap-2">
-              <Text font="secondary-body" color="text-03">
+              <Text secondaryBody text03>
                 {group.providerLabel}
               </Text>
               <div className="flex flex-col gap-2">

web/tests/e2e/settings/settings_pages.spec.ts

@@ -4,14 +4,6 @@ import { expectScreenshot } from "@tests/e2e/utils/visualRegression";
 
 test.use({ storageState: "admin_auth.json" });
 
-/** Maps each settings slug to the header title shown on that page. */
-const SLUG_TO_HEADER: Record<string, string> = {
-  general: "Profile",
-  "chat-preferences": "Chats",
-  "accounts-access": "Accounts",
-  connectors: "Connectors",
-};
-
 for (const theme of THEMES) {
   test.describe(`Settings pages (${theme} mode)`, () => {
     test.beforeEach(async ({ page }) => {
@@ -19,33 +11,21 @@ for (const theme of THEMES) {
     });
 
     test("should screenshot each settings tab", async ({ page }) => {
-      await page.goto("/app/settings/general");
-      await page
-        .getByTestId("settings-left-tab-navigation")
-        .waitFor({ state: "visible" });
+      await page.goto("/app/settings");
+      await page.waitForLoadState("networkidle");
 
       const nav = page.getByTestId("settings-left-tab-navigation");
       const tabs = nav.locator("a");
-      await expect(tabs.first()).toBeVisible({ timeout: 10_000 });
       const count = await tabs.count();
 
+      expect(count).toBeGreaterThan(0);
       for (let i = 0; i < count; i++) {
         const tab = tabs.nth(i);
         const href = await tab.getAttribute("href");
         const slug = href ? href.replace("/app/settings/", "") : `tab-${i}`;
 
         await tab.click();
-
-        const expectedHeader = SLUG_TO_HEADER[slug];
-        if (expectedHeader) {
-          await expect(
-            page
-              .locator(".opal-content-md-header")
-              .filter({ hasText: expectedHeader })
-          ).toBeVisible({ timeout: 10_000 });
-        } else {
-          await page.waitForLoadState("networkidle");
-        }
+        await page.waitForLoadState("networkidle");
 
         await expectScreenshot(page, {
           name: `settings-${theme}-${slug}`,