fix: Anthropic litellm thinking workaround (#9713)

Co-authored-by: Jamison Lahman <jamison@lahman.dev>

.greptile/rules.md

@@ -24,16 +24,6 @@ When hardcoding a boolean variable to a constant value, remove the variable enti
 
 Code changes must consider both multi-tenant and single-tenant deployments. In multi-tenant mode, preserve tenant isolation, ensure tenant context is propagated correctly, and avoid assumptions that only hold for a single shared schema or globally shared state. In single-tenant mode, avoid introducing unnecessary tenant-specific requirements or cloud-only control-plane dependencies.
 
-## Nginx Routing — New Backend Routes
-
-Whenever a new backend route is added that does NOT start with `/api`, it must also be explicitly added to ALL nginx configs:
-- `deployment/helm/charts/onyx/templates/nginx-conf.yaml` (Helm/k8s)
-- `deployment/data/nginx/app.conf.template` (docker-compose dev)
-- `deployment/data/nginx/app.conf.template.prod` (docker-compose prod)
-- `deployment/data/nginx/app.conf.template.no-letsencrypt` (docker-compose no-letsencrypt)
-
-Routes not starting with `/api` are not caught by the existing `^/(api|openapi\.json)` location block and will fall through to `location /`, which proxies to the Next.js web server and returns an HTML 404. The new location block must be placed before the `/api` block. Examples of routes that need this treatment: `/scim`, `/mcp`.
-
 ## Full vs Lite Deployments
 
 Code changes must consider both regular Onyx deployments and Onyx lite deployments. Lite deployments disable the vector DB, Redis, model servers, and background workers by default, use PostgreSQL-backed cache/auth/file storage, and rely on the API server to handle background work. Do not assume those services are available unless the code path is explicitly limited to full deployments.

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

@@ -473,8 +473,6 @@ def connector_permission_sync_generator_task(
             cc_pair = get_connector_credential_pair_from_id(
                 db_session=db_session,
                 cc_pair_id=cc_pair_id,
-                eager_load_connector=True,
-                eager_load_credential=True,
             )
             if cc_pair is None:
                 raise ValueError(

backend/ee/onyx/external_permissions/slack/doc_sync.py

@@ -8,7 +8,6 @@ from ee.onyx.external_permissions.slack.utils import fetch_user_id_to_email_map
 from onyx.access.models import DocExternalAccess
 from onyx.access.models import ExternalAccess
 from onyx.connectors.credentials_provider import OnyxDBCredentialsProvider
-from onyx.connectors.interfaces import SecondsSinceUnixEpoch
 from onyx.connectors.models import HierarchyNode
 from onyx.connectors.slack.connector import get_channels
 from onyx.connectors.slack.connector import make_paginated_slack_api_call
@@ -106,11 +105,9 @@ def _get_slack_document_access(
     slack_connector: SlackConnector,
     channel_permissions: dict[str, ExternalAccess],  # noqa: ARG001
     callback: IndexingHeartbeatInterface | None,
-    indexing_start: SecondsSinceUnixEpoch | None = None,
 ) -> Generator[DocExternalAccess, None, None]:
     slim_doc_generator = slack_connector.retrieve_all_slim_docs_perm_sync(
-        callback=callback,
-        start=indexing_start,
+        callback=callback
     )
 
     for doc_metadata_batch in slim_doc_generator:
@@ -183,15 +180,9 @@ def slack_doc_sync(
 
     slack_connector = SlackConnector(**cc_pair.connector.connector_specific_config)
     slack_connector.set_credentials_provider(provider)
-    indexing_start_ts: SecondsSinceUnixEpoch | None = (
-        cc_pair.connector.indexing_start.timestamp()
-        if cc_pair.connector.indexing_start is not None
-        else None
-    )
 
     yield from _get_slack_document_access(
-        slack_connector=slack_connector,
+        slack_connector,
         channel_permissions=channel_permissions,
         callback=callback,
-        indexing_start=indexing_start_ts,
     )

backend/ee/onyx/external_permissions/utils.py

@@ -6,7 +6,6 @@ from onyx.access.models import ElementExternalAccess
 from onyx.access.models import ExternalAccess
 from onyx.access.models import NodeExternalAccess
 from onyx.configs.constants import DocumentSource
-from onyx.connectors.interfaces import SecondsSinceUnixEpoch
 from onyx.connectors.interfaces import SlimConnectorWithPermSync
 from onyx.connectors.models import HierarchyNode
 from onyx.db.models import ConnectorCredentialPair
@@ -41,19 +40,10 @@ def generic_doc_sync(
 
     logger.info(f"Starting {doc_source} doc sync for CC Pair ID: {cc_pair.id}")
 
-    indexing_start: SecondsSinceUnixEpoch | None = (
-        cc_pair.connector.indexing_start.timestamp()
-        if cc_pair.connector.indexing_start is not None
-        else None
-    )
-
     newly_fetched_doc_ids: set[str] = set()
 
     logger.info(f"Fetching all slim documents from {doc_source}")
-    for doc_batch in slim_connector.retrieve_all_slim_docs_perm_sync(
-        start=indexing_start,
-        callback=callback,
-    ):
+    for doc_batch in slim_connector.retrieve_all_slim_docs_perm_sync(callback=callback):
         logger.info(f"Got {len(doc_batch)} slim documents from {doc_source}")
 
         if callback:

backend/onyx/configs/app_configs.py

@@ -44,31 +44,6 @@ SEND_USER_METADATA_TO_LLM_PROVIDER = (
 # User Facing Features Configs
 #####
 BLURB_SIZE = 128  # Number Encoder Tokens included in the chunk blurb
-
-# Hard ceiling for the admin-configurable file upload size (in MB).
-# Self-hosted customers can raise or lower this via the environment variable.
-_raw_max_upload_size_mb = int(os.environ.get("MAX_ALLOWED_UPLOAD_SIZE_MB", "250"))
-if _raw_max_upload_size_mb < 0:
-    logger.warning(
-        "MAX_ALLOWED_UPLOAD_SIZE_MB=%d is negative; falling back to 250",
-        _raw_max_upload_size_mb,
-    )
-    _raw_max_upload_size_mb = 250
-MAX_ALLOWED_UPLOAD_SIZE_MB = _raw_max_upload_size_mb
-
-# Default fallback for the per-user file upload size limit (in MB) when no
-# admin-configured value exists.  Clamped to MAX_ALLOWED_UPLOAD_SIZE_MB at
-# runtime so this never silently exceeds the hard ceiling.
-_raw_default_upload_size_mb = int(
-    os.environ.get("DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB", "100")
-)
-if _raw_default_upload_size_mb < 0:
-    logger.warning(
-        "DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB=%d is negative; falling back to 100",
-        _raw_default_upload_size_mb,
-    )
-    _raw_default_upload_size_mb = 100
-DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB = _raw_default_upload_size_mb
 GENERATIVE_MODEL_ACCESS_CHECK_FREQ = int(
     os.environ.get("GENERATIVE_MODEL_ACCESS_CHECK_FREQ") or 86400
 )  # 1 day
@@ -86,6 +61,17 @@ CACHE_BACKEND = CacheBackendType(
     os.environ.get("CACHE_BACKEND", CacheBackendType.REDIS)
 )
 
+# Maximum token count for a single uploaded file. Files exceeding this are rejected.
+# Defaults to 100k tokens (or 10M when vector DB is disabled).
+_DEFAULT_FILE_TOKEN_LIMIT = 10_000_000 if DISABLE_VECTOR_DB else 100_000
+FILE_TOKEN_COUNT_THRESHOLD = int(
+    os.environ.get("FILE_TOKEN_COUNT_THRESHOLD", str(_DEFAULT_FILE_TOKEN_LIMIT))
+)
+
+# Maximum upload size for a single user file (chat/projects) in MB.
+USER_FILE_MAX_UPLOAD_SIZE_MB = int(os.environ.get("USER_FILE_MAX_UPLOAD_SIZE_MB") or 50)
+USER_FILE_MAX_UPLOAD_SIZE_BYTES = USER_FILE_MAX_UPLOAD_SIZE_MB * 1024 * 1024
+
 # If set to true, will show extra/uncommon connectors in the "Other" category
 SHOW_EXTRA_CONNECTORS = os.environ.get("SHOW_EXTRA_CONNECTORS", "").lower() == "true"
 

backend/onyx/connectors/confluence/connector.py

@@ -890,8 +890,8 @@ class ConfluenceConnector(
 
     def _retrieve_all_slim_docs(
         self,
-        start: SecondsSinceUnixEpoch | None = None,
-        end: SecondsSinceUnixEpoch | None = None,
+        start: SecondsSinceUnixEpoch | None = None,  # noqa: ARG002
+        end: SecondsSinceUnixEpoch | None = None,  # noqa: ARG002
         callback: IndexingHeartbeatInterface | None = None,
         include_permissions: bool = True,
     ) -> GenerateSlimDocumentOutput:
@@ -915,8 +915,8 @@ class ConfluenceConnector(
                 self.confluence_client, doc_id, restrictions, ancestors
             ) or space_level_access_info.get(page_space_key)
 
-        # Query pages (with optional time filtering for indexing_start)
-        page_query = self._construct_page_cql_query(start, end)
+        # Query pages
+        page_query = self.base_cql_page_query + self.cql_label_filter
         for page in self.confluence_client.cql_paginate_all_expansions(
             cql=page_query,
             expand=restrictions_expand,
@@ -950,9 +950,7 @@ class ConfluenceConnector(
 
             # Query attachments for each page
             page_hierarchy_node_yielded = False
-            attachment_query = self._construct_attachment_query(
-                _get_page_id(page), start, end
-            )
+            attachment_query = self._construct_attachment_query(_get_page_id(page))
             for attachment in self.confluence_client.cql_paginate_all_expansions(
                 cql=attachment_query,
                 expand=restrictions_expand,

backend/onyx/connectors/sharepoint/connector.py

@@ -1765,11 +1765,7 @@ class SharepointConnector(
         checkpoint.current_drive_delta_next_link = None
         checkpoint.seen_document_ids.clear()
 
-    def _fetch_slim_documents_from_sharepoint(
-        self,
-        start: datetime | None = None,
-        end: datetime | None = None,
-    ) -> GenerateSlimDocumentOutput:
+    def _fetch_slim_documents_from_sharepoint(self) -> GenerateSlimDocumentOutput:
         site_descriptors = self._filter_excluded_sites(
             self.site_descriptors or self.fetch_sites()
         )
@@ -1790,9 +1786,7 @@ class SharepointConnector(
             # Process site documents if flag is True
             if self.include_site_documents:
                 for driveitem, drive_name, drive_web_url in self._fetch_driveitems(
-                    site_descriptor=site_descriptor,
-                    start=start,
-                    end=end,
+                    site_descriptor=site_descriptor
                 ):
                     if self._is_driveitem_excluded(driveitem):
                         logger.debug(f"Excluding by path denylist: {driveitem.web_url}")
@@ -1847,9 +1841,7 @@ class SharepointConnector(
 
             # Process site pages if flag is True
             if self.include_site_pages:
-                site_pages = self._fetch_site_pages(
-                    site_descriptor, start=start, end=end
-                )
+                site_pages = self._fetch_site_pages(site_descriptor)
                 for site_page in site_pages:
                     logger.debug(
                         f"Processing site page: {site_page.get('webUrl', site_page.get('name', 'Unknown'))}"
@@ -2573,22 +2565,12 @@ class SharepointConnector(
 
     def retrieve_all_slim_docs_perm_sync(
         self,
-        start: SecondsSinceUnixEpoch | None = None,
-        end: SecondsSinceUnixEpoch | None = None,
+        start: SecondsSinceUnixEpoch | None = None,  # noqa: ARG002
+        end: SecondsSinceUnixEpoch | None = None,  # noqa: ARG002
         callback: IndexingHeartbeatInterface | None = None,  # noqa: ARG002
     ) -> GenerateSlimDocumentOutput:
-        start_dt = (
-            datetime.fromtimestamp(start, tz=timezone.utc)
-            if start is not None
-            else None
-        )
-        end_dt = (
-            datetime.fromtimestamp(end, tz=timezone.utc) if end is not None else None
-        )
-        yield from self._fetch_slim_documents_from_sharepoint(
-            start=start_dt,
-            end=end_dt,
-        )
+
+        yield from self._fetch_slim_documents_from_sharepoint()
 
 
 if __name__ == "__main__":

backend/onyx/connectors/slack/connector.py

@@ -516,8 +516,6 @@ def _get_all_doc_ids(
     ] = default_msg_filter,
     callback: IndexingHeartbeatInterface | None = None,
     workspace_url: str | None = None,
-    start: SecondsSinceUnixEpoch | None = None,
-    end: SecondsSinceUnixEpoch | None = None,
 ) -> GenerateSlimDocumentOutput:
     """
     Get all document ids in the workspace, channel by channel
@@ -548,8 +546,6 @@ def _get_all_doc_ids(
             client=client,
             channel=channel,
             callback=callback,
-            oldest=str(start) if start else None,  # 0.0 -> None intentionally
-            latest=str(end) if end is not None else None,
         )
 
         for message_batch in channel_message_batches:
@@ -851,8 +847,8 @@ class SlackConnector(
 
     def retrieve_all_slim_docs_perm_sync(
         self,
-        start: SecondsSinceUnixEpoch | None = None,
-        end: SecondsSinceUnixEpoch | None = None,
+        start: SecondsSinceUnixEpoch | None = None,  # noqa: ARG002
+        end: SecondsSinceUnixEpoch | None = None,  # noqa: ARG002
         callback: IndexingHeartbeatInterface | None = None,
     ) -> GenerateSlimDocumentOutput:
         if self.client is None:
@@ -865,8 +861,6 @@ class SlackConnector(
             msg_filter_func=self.msg_filter_func,
             callback=callback,
             workspace_url=self._workspace_url,
-            start=start,
-            end=end,
         )
 
     def _load_from_checkpoint(

backend/onyx/db/projects.py

@@ -7,7 +7,6 @@ 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
@@ -35,19 +34,9 @@ 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]
@@ -109,7 +98,6 @@ 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,
     )
 
 
@@ -135,7 +123,6 @@ 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:
@@ -147,12 +134,12 @@ 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 indexable_files:
+        for user_file in user_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 indexable_files:
+        for user_file in user_files:
             task = client_app.send_task(
                 OnyxCeleryTask.PROCESS_SINGLE_USER_FILE,
                 kwargs={"user_file_id": user_file.id, "tenant_id": tenant_id},
@@ -168,7 +155,6 @@ 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/file_processing/file_types.py

@@ -15,7 +15,6 @@ 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",
@@ -35,12 +34,13 @@ 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, TABULAR_MIME_TYPES
+        TEXT_MIME_TYPES, DOCUMENT_MIME_TYPES, CSV_MIME_TYPES
     )
 
     EXCLUDED_IMAGE_TYPES = {

backend/onyx/file_store/models.py

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

backend/onyx/indexing/adapters/user_file_indexing_adapter.py

@@ -29,7 +29,6 @@ from onyx.indexing.models import DocMetadataAwareIndexChunk
 from onyx.indexing.models import IndexChunk
 from onyx.indexing.models import UpdatableChunkData
 from onyx.llm.factory import get_default_llm
-from onyx.natural_language_processing.utils import count_tokens
 from onyx.natural_language_processing.utils import get_tokenizer
 from onyx.utils.logger import setup_logger
 
@@ -174,10 +173,8 @@ class UserFileIndexingAdapter:
                     [chunk.content for chunk in user_file_chunks]
                 )
                 user_file_id_to_raw_text[str(user_file_id)] = combined_content
-                token_count: int = (
-                    count_tokens(combined_content, llm_tokenizer)
-                    if llm_tokenizer
-                    else 0
+                token_count = (
+                    len(llm_tokenizer.encode(combined_content)) if llm_tokenizer else 0
                 )
                 user_file_id_to_token_count[str(user_file_id)] = token_count
             else:

backend/onyx/llm/multi_llm.py

@@ -185,6 +185,21 @@ def _messages_contain_tool_content(messages: list[dict[str, Any]]) -> bool:
     return False
 
 
+def _prompt_contains_tool_call_history(prompt: LanguageModelInput) -> bool:
+    """Check if the prompt contains any assistant messages with tool_calls.
+
+    When Anthropic's extended thinking is enabled, the API requires every
+    assistant message to start with a thinking block before any tool_use
+    blocks.  Since we don't preserve thinking_blocks (they carry
+    cryptographic signatures that can't be reconstructed), we must skip
+    the thinking param whenever history contains prior tool-calling turns.
+    """
+    from onyx.llm.models import AssistantMessage
+
+    msgs = prompt if isinstance(prompt, list) else [prompt]
+    return any(isinstance(msg, AssistantMessage) and msg.tool_calls for msg in msgs)
+
+
 def _is_vertex_model_rejecting_output_config(model_name: str) -> bool:
     normalized_model_name = model_name.lower()
     return any(
@@ -466,7 +481,20 @@ class LitellmLLM(LLM):
                     reasoning_effort
                 )
 
-                if budget_tokens is not None:
+                # Anthropic requires every assistant message with tool_use
+                # blocks to start with a thinking block that carries a
+                # cryptographic signature.  We don't preserve those blocks
+                # across turns, so skip thinking when the history already
+                # contains tool-calling assistant messages.  LiteLLM's
+                # modify_params workaround doesn't cover all providers
+                # (notably Bedrock).
+                can_enable_thinking = (
+                    budget_tokens is not None
+                    and not _prompt_contains_tool_call_history(prompt)
+                )
+
+                if can_enable_thinking:
+                    assert budget_tokens is not None  # mypy
                     if max_tokens is not None:
                         # Anthropic has a weird rule where max token has to be at least as much as budget tokens if set
                         # and the minimum budget tokens is 1024

backend/onyx/natural_language_processing/utils.py

@@ -175,32 +175,6 @@ def get_tokenizer(
     return _check_tokenizer_cache(provider_type, model_name)
 
 
-# Max characters per encode() call.
-_ENCODE_CHUNK_SIZE = 500_000
-
-
-def count_tokens(
-    text: str,
-    tokenizer: BaseTokenizer,
-    token_limit: int | None = None,
-) -> int:
-    """Count tokens, chunking the input to avoid tiktoken stack overflow.
-
-    If token_limit is provided and the text is large enough to require
-    multiple chunks (> 500k chars), stops early once the count exceeds it.
-    When early-exiting, the returned value exceeds token_limit but may be
-    less than the true full token count.
-    """
-    if len(text) <= _ENCODE_CHUNK_SIZE:
-        return len(tokenizer.encode(text))
-    total = 0
-    for start in range(0, len(text), _ENCODE_CHUNK_SIZE):
-        total += len(tokenizer.encode(text[start : start + _ENCODE_CHUNK_SIZE]))
-        if token_limit is not None and total > token_limit:
-            return total  # Already over — skip remaining chunks
-    return total
-
-
 def tokenizer_trim_content(
     content: str, desired_length: int, tokenizer: BaseTokenizer
 ) -> str:

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

@@ -9,15 +9,20 @@ from pydantic import ConfigDict
 from pydantic import Field
 from sqlalchemy.orm import Session
 
+from onyx.configs.app_configs import FILE_TOKEN_COUNT_THRESHOLD
+from onyx.configs.app_configs import USER_FILE_MAX_UPLOAD_SIZE_BYTES
+from onyx.configs.app_configs import USER_FILE_MAX_UPLOAD_SIZE_MB
 from onyx.db.llm import fetch_default_llm_model
 from onyx.file_processing.extract_file_text import extract_file_text
 from onyx.file_processing.extract_file_text import get_file_ext
 from onyx.file_processing.file_types import OnyxFileExtensions
 from onyx.file_processing.password_validation import is_file_password_protected
-from onyx.natural_language_processing.utils import count_tokens
 from onyx.natural_language_processing.utils import get_tokenizer
-from onyx.server.settings.store import load_settings
 from onyx.utils.logger import setup_logger
+from shared_configs.configs import MULTI_TENANT
+from shared_configs.configs import SKIP_USERFILE_THRESHOLD
+from shared_configs.configs import SKIP_USERFILE_THRESHOLD_TENANT_LIST
+from shared_configs.contextvars import get_current_tenant_id
 
 
 logger = setup_logger()
@@ -76,26 +81,11 @@ 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
@@ -171,8 +161,8 @@ def categorize_uploaded_files(
       document formats (.pdf, .docx, …) and falls back to a text-detection
       heuristic for unknown extensions (.py, .js, .rs, …).
     - Uses default tokenizer to compute token length.
-    - If token length exceeds the admin-configured threshold, reject file.
-    - If extension unsupported or text cannot be extracted, reject file.
+    - If token length > threshold, reject file (unless threshold skip is enabled).
+    - If text cannot be extracted, reject file.
     - Otherwise marked as acceptable.
     """
 
@@ -183,33 +173,36 @@ def categorize_uploaded_files(
     provider_type = default_model.llm_provider.provider if default_model else None
     tokenizer = get_tokenizer(model_name=model_name, provider_type=provider_type)
 
-    # Derive limits from admin-configurable settings.
-    # For upload size: load_settings() resolves 0/None to a positive default.
-    # For token threshold: 0 means "no limit" (converted to None below).
-    settings = load_settings()
-    max_upload_size_mb = (
-        settings.user_file_max_upload_size_mb
-    )  # always positive after load_settings()
-    max_upload_size_bytes = (
-        max_upload_size_mb * 1024 * 1024 if max_upload_size_mb else None
-    )
-    token_threshold_k = settings.file_token_count_threshold_k
-    token_threshold = (
-        token_threshold_k * 1000 if token_threshold_k else None
-    )  # 0 → None = no limit
+    # Check if threshold checks should be skipped
+    skip_threshold = False
+
+    # Check global skip flag (works for both single-tenant and multi-tenant)
+    if SKIP_USERFILE_THRESHOLD:
+        skip_threshold = True
+        logger.info("Skipping userfile threshold check (global setting)")
+    # Check tenant-specific skip list (only applicable in multi-tenant)
+    elif MULTI_TENANT and SKIP_USERFILE_THRESHOLD_TENANT_LIST:
+        try:
+            current_tenant_id = get_current_tenant_id()
+            skip_threshold = current_tenant_id in SKIP_USERFILE_THRESHOLD_TENANT_LIST
+            if skip_threshold:
+                logger.info(
+                    f"Skipping userfile threshold check for tenant: {current_tenant_id}"
+                )
+        except RuntimeError as e:
+            logger.warning(f"Failed to get current tenant ID: {str(e)}")
 
     for upload in files:
         try:
             filename = get_safe_filename(upload)
 
-            # Size limit is a hard safety cap.
-            if max_upload_size_bytes is not None and is_upload_too_large(
-                upload, max_upload_size_bytes
-            ):
+            # Size limit is a hard safety cap and is enforced even when token
+            # threshold checks are skipped via SKIP_USERFILE_THRESHOLD settings.
+            if is_upload_too_large(upload, USER_FILE_MAX_UPLOAD_SIZE_BYTES):
                 results.rejected.append(
                     RejectedFile(
                         filename=filename,
-                        reason=f"Exceeds {max_upload_size_mb} MB file size limit",
+                        reason=f"Exceeds {USER_FILE_MAX_UPLOAD_SIZE_MB} MB file size limit",
                     )
                 )
                 continue
@@ -231,11 +224,11 @@ def categorize_uploaded_files(
                     )
                     continue
 
-                if token_threshold is not None and token_count > token_threshold:
+                if not skip_threshold and token_count > FILE_TOKEN_COUNT_THRESHOLD:
                     results.rejected.append(
                         RejectedFile(
                             filename=filename,
-                            reason=f"Exceeds {token_threshold_k}K token limit",
+                            reason=f"Exceeds {FILE_TOKEN_COUNT_THRESHOLD} token limit",
                         )
                     )
                 else:
@@ -276,24 +269,12 @@ def categorize_uploaded_files(
                     )
                     continue
 
-                token_count = count_tokens(
-                    text_content, tokenizer, token_limit=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:
+                token_count = len(tokenizer.encode(text_content))
+                if not skip_threshold and token_count > FILE_TOKEN_COUNT_THRESHOLD:
                     results.rejected.append(
                         RejectedFile(
                             filename=filename,
-                            reason=f"Exceeds {token_threshold_k}K token limit",
+                            reason=f"Exceeds {FILE_TOKEN_COUNT_THRESHOLD} token limit",
                         )
                     )
                 else:

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.TABULAR_MIME_TYPES:
-        return ChatFileType.TABULAR
+    if mime_type in OnyxMimeTypes.CSV_MIME_TYPES:
+        return ChatFileType.CSV
 
     if mime_type in OnyxMimeTypes.DOCUMENT_MIME_TYPES:
         return ChatFileType.DOC

backend/onyx/server/settings/api.py

@@ -9,9 +9,7 @@ from onyx import __version__ as onyx_version
 from onyx.auth.users import current_admin_user
 from onyx.auth.users import current_user
 from onyx.auth.users import is_user_admin
-from onyx.configs.app_configs import DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB
 from onyx.configs.app_configs import DISABLE_VECTOR_DB
-from onyx.configs.app_configs import MAX_ALLOWED_UPLOAD_SIZE_MB
 from onyx.configs.constants import KV_REINDEX_KEY
 from onyx.configs.constants import NotificationType
 from onyx.db.engine.sql_engine import get_session
@@ -19,16 +17,10 @@ from onyx.db.models import User
 from onyx.db.notification import dismiss_all_notifications
 from onyx.db.notification import get_notifications
 from onyx.db.notification import update_notification_last_shown
-from onyx.error_handling.error_codes import OnyxErrorCode
-from onyx.error_handling.exceptions import OnyxError
 from onyx.hooks.utils import HOOKS_AVAILABLE
 from onyx.key_value_store.factory import get_kv_store
 from onyx.key_value_store.interface import KvKeyNotFoundError
 from onyx.server.features.build.utils import is_onyx_craft_enabled
-from onyx.server.settings.models import (
-    DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_NO_VECTOR_DB,
-)
-from onyx.server.settings.models import DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_VECTOR_DB
 from onyx.server.settings.models import Notification
 from onyx.server.settings.models import Settings
 from onyx.server.settings.models import UserSettings
@@ -49,15 +41,6 @@ basic_router = APIRouter(prefix="/settings")
 def admin_put_settings(
     settings: Settings, _: User = Depends(current_admin_user)
 ) -> None:
-    if (
-        settings.user_file_max_upload_size_mb is not None
-        and settings.user_file_max_upload_size_mb > 0
-        and settings.user_file_max_upload_size_mb > MAX_ALLOWED_UPLOAD_SIZE_MB
-    ):
-        raise OnyxError(
-            OnyxErrorCode.INVALID_INPUT,
-            f"File upload size limit cannot exceed {MAX_ALLOWED_UPLOAD_SIZE_MB} MB",
-        )
     store_settings(settings)
 
 
@@ -100,16 +83,6 @@ def fetch_settings(
         vector_db_enabled=not DISABLE_VECTOR_DB,
         hooks_enabled=HOOKS_AVAILABLE,
         version=onyx_version,
-        max_allowed_upload_size_mb=MAX_ALLOWED_UPLOAD_SIZE_MB,
-        default_user_file_max_upload_size_mb=min(
-            DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB,
-            MAX_ALLOWED_UPLOAD_SIZE_MB,
-        ),
-        default_file_token_count_threshold_k=(
-            DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_NO_VECTOR_DB
-            if DISABLE_VECTOR_DB
-            else DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_VECTOR_DB
-        ),
     )
 
 

backend/onyx/server/settings/models.py

@@ -2,19 +2,12 @@ from datetime import datetime
 from enum import Enum
 
 from pydantic import BaseModel
-from pydantic import Field
 
-from onyx.configs.app_configs import DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB
-from onyx.configs.app_configs import DISABLE_VECTOR_DB
-from onyx.configs.app_configs import MAX_ALLOWED_UPLOAD_SIZE_MB
 from onyx.configs.constants import NotificationType
 from onyx.configs.constants import QueryHistoryType
 from onyx.db.models import Notification as NotificationDBModel
 from shared_configs.configs import POSTGRES_DEFAULT_SCHEMA
 
-DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_VECTOR_DB = 200
-DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_NO_VECTOR_DB = 10000
-
 
 class PageType(str, Enum):
     CHAT = "chat"
@@ -85,12 +78,7 @@ class Settings(BaseModel):
 
     # User Knowledge settings
     user_knowledge_enabled: bool | None = True
-    user_file_max_upload_size_mb: int | None = Field(
-        default=DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB, ge=0
-    )
-    file_token_count_threshold_k: int | None = Field(
-        default=None, ge=0  # thousands of tokens; None = context-aware default
-    )
+    user_file_max_upload_size_mb: int | None = None
 
     # Connector settings
     show_extra_connectors: bool | None = True
@@ -120,14 +108,3 @@ class UserSettings(Settings):
     hooks_enabled: bool = False
     # Application version, read from the ONYX_VERSION env var at startup.
     version: str | None = None
-    # Hard ceiling for user_file_max_upload_size_mb, derived from env var.
-    max_allowed_upload_size_mb: int = MAX_ALLOWED_UPLOAD_SIZE_MB
-    # Factory defaults so the frontend can show a "restore default" button.
-    default_user_file_max_upload_size_mb: int = DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB
-    default_file_token_count_threshold_k: int = Field(
-        default_factory=lambda: (
-            DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_NO_VECTOR_DB
-            if DISABLE_VECTOR_DB
-            else DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_VECTOR_DB
-        )
-    )

backend/onyx/server/settings/store.py

@@ -1,19 +1,13 @@
 from onyx.cache.factory import get_cache_backend
-from onyx.configs.app_configs import DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB
 from onyx.configs.app_configs import DISABLE_USER_KNOWLEDGE
-from onyx.configs.app_configs import DISABLE_VECTOR_DB
 from onyx.configs.app_configs import ENABLE_OPENSEARCH_INDEXING_FOR_ONYX
-from onyx.configs.app_configs import MAX_ALLOWED_UPLOAD_SIZE_MB
 from onyx.configs.app_configs import ONYX_QUERY_HISTORY_TYPE
 from onyx.configs.app_configs import SHOW_EXTRA_CONNECTORS
+from onyx.configs.app_configs import USER_FILE_MAX_UPLOAD_SIZE_MB
 from onyx.configs.constants import KV_SETTINGS_KEY
 from onyx.configs.constants import OnyxRedisLocks
 from onyx.key_value_store.factory import get_kv_store
 from onyx.key_value_store.interface import KvKeyNotFoundError
-from onyx.server.settings.models import (
-    DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_NO_VECTOR_DB,
-)
-from onyx.server.settings.models import DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_VECTOR_DB
 from onyx.server.settings.models import Settings
 from onyx.utils.logger import setup_logger
 
@@ -57,36 +51,9 @@ def load_settings() -> Settings:
     if DISABLE_USER_KNOWLEDGE:
         settings.user_knowledge_enabled = False
 
+    settings.user_file_max_upload_size_mb = USER_FILE_MAX_UPLOAD_SIZE_MB
     settings.show_extra_connectors = SHOW_EXTRA_CONNECTORS
     settings.opensearch_indexing_enabled = ENABLE_OPENSEARCH_INDEXING_FOR_ONYX
-
-    # Resolve context-aware defaults for token threshold.
-    # None = admin hasn't set a value yet → use context-aware default.
-    # 0 = admin explicitly chose "no limit" → preserve as-is.
-    if settings.file_token_count_threshold_k is None:
-        settings.file_token_count_threshold_k = (
-            DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_NO_VECTOR_DB
-            if DISABLE_VECTOR_DB
-            else DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_VECTOR_DB
-        )
-
-    # Upload size: 0 and None are treated as "unset" (not "no limit") →
-    # fall back to min(configured default, hard ceiling).
-    if not settings.user_file_max_upload_size_mb:
-        settings.user_file_max_upload_size_mb = min(
-            DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB,
-            MAX_ALLOWED_UPLOAD_SIZE_MB,
-        )
-
-    # Clamp to env ceiling so stale KV values are capped even if the
-    # operator lowered MAX_ALLOWED_UPLOAD_SIZE_MB after a higher value
-    # was already saved (api.py only guards new writes).
-    if (
-        settings.user_file_max_upload_size_mb > 0
-        and settings.user_file_max_upload_size_mb > MAX_ALLOWED_UPLOAD_SIZE_MB
-    ):
-        settings.user_file_max_upload_size_mb = MAX_ALLOWED_UPLOAD_SIZE_MB
-
     return settings
 
 

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 TABULAR are guaranteed to contain actual text bytes.
+        # Only PLAIN_TEXT and CSV 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.TABULAR):
+        if chat_file.file_type not in (ChatFileType.PLAIN_TEXT, ChatFileType.CSV):
             raise ToolCallException(
                 message=f"File {file_id} is not a text file (type={chat_file.file_type})",
                 llm_facing_message=(

backend/shared_configs/configs.py

@@ -191,6 +191,25 @@ IGNORED_SYNCING_TENANT_LIST = (
     else None
 )
 
+# Global flag to skip userfile threshold for all users/tenants
+SKIP_USERFILE_THRESHOLD = (
+    os.environ.get("SKIP_USERFILE_THRESHOLD", "").lower() == "true"
+)
+
+# Comma-separated list of specific tenant IDs to skip threshold (multi-tenant only)
+SKIP_USERFILE_THRESHOLD_TENANT_IDS = os.environ.get(
+    "SKIP_USERFILE_THRESHOLD_TENANT_IDS"
+)
+SKIP_USERFILE_THRESHOLD_TENANT_LIST = (
+    [
+        tenant.strip()
+        for tenant in SKIP_USERFILE_THRESHOLD_TENANT_IDS.split(",")
+        if tenant.strip()
+    ]
+    if SKIP_USERFILE_THRESHOLD_TENANT_IDS
+    else None
+)
+
 ENVIRONMENT = os.environ.get("ENVIRONMENT") or "not_explicitly_set"
 
 

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

@@ -1,6 +1,4 @@
 import time
-from datetime import datetime
-from datetime import timezone
 
 import pytest
 
@@ -19,10 +17,6 @@ PRIVATE_CHANNEL_USERS = [
     "test_user_2@onyx-test.com",
 ]
 
-# Predates any test workspace messages, so the result set should match
-# the "no start time" case while exercising the oldest= parameter.
-OLDEST_TS_2016 = datetime(2016, 1, 1, tzinfo=timezone.utc).timestamp()
-
 pytestmark = pytest.mark.usefixtures("enable_ee")
 
 
@@ -111,17 +105,15 @@ def test_load_from_checkpoint_access__private_channel(
     ],
     indirect=True,
 )
-@pytest.mark.parametrize("start_ts", [None, OLDEST_TS_2016])
 def test_slim_documents_access__public_channel(
     slack_connector: SlackConnector,
-    start_ts: float | None,
 ) -> None:
     """Test that retrieve_all_slim_docs_perm_sync returns correct access information for slim documents."""
     if not slack_connector.client:
         raise RuntimeError("Web client must be defined")
 
     slim_docs_generator = slack_connector.retrieve_all_slim_docs_perm_sync(
-        start=start_ts,
+        start=0.0,
         end=time.time(),
     )
 
@@ -157,7 +149,7 @@ def test_slim_documents_access__private_channel(
         raise RuntimeError("Web client must be defined")
 
     slim_docs_generator = slack_connector.retrieve_all_slim_docs_perm_sync(
-        start=None,
+        start=0.0,
         end=time.time(),
     )
 

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.TABULAR,
+        "type": ChatFileType.CSV,
         "name": "data.csv",
         "user_file_id": str(user_file.id),
     }

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.TABULAR
+    assert result[0]["type"] == ChatFileType.CSV
 
 
 def test_unknown_extension_defaults_to_plain_text() -> None:

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

@@ -1,5 +1,3 @@
-from datetime import datetime
-from datetime import timezone
 from unittest.mock import MagicMock
 from unittest.mock import patch
 
@@ -33,7 +31,6 @@ def mock_jira_cc_pair(
         "jira_base_url": jira_base_url,
         "project_key": project_key,
     }
-    mock_cc_pair.connector.indexing_start = None
 
     return mock_cc_pair
 
@@ -68,75 +65,3 @@ def test_jira_permission_sync(
             fetch_all_existing_docs_ids_fn=mock_fetch_all_existing_docs_ids_fn,
         ):
             print(doc)
-
-
-def test_jira_doc_sync_passes_indexing_start(
-    jira_connector: JiraConnector,
-    mock_jira_cc_pair: MagicMock,
-    mock_fetch_all_existing_docs_fn: MagicMock,
-    mock_fetch_all_existing_docs_ids_fn: MagicMock,
-) -> None:
-    """Verify that generic_doc_sync derives indexing_start from cc_pair
-    and forwards it to retrieve_all_slim_docs_perm_sync."""
-    indexing_start_dt = datetime(2025, 6, 1, tzinfo=timezone.utc)
-    mock_jira_cc_pair.connector.indexing_start = indexing_start_dt
-
-    with patch("onyx.connectors.jira.connector.build_jira_client") as mock_build_client:
-        mock_build_client.return_value = jira_connector._jira_client
-        assert jira_connector._jira_client is not None
-        jira_connector._jira_client._options = MagicMock()
-        jira_connector._jira_client._options.return_value = {
-            "rest_api_version": JIRA_SERVER_API_VERSION
-        }
-
-        with patch.object(
-            type(jira_connector),
-            "retrieve_all_slim_docs_perm_sync",
-            return_value=iter([]),
-        ) as mock_retrieve:
-            list(
-                jira_doc_sync(
-                    cc_pair=mock_jira_cc_pair,
-                    fetch_all_existing_docs_fn=mock_fetch_all_existing_docs_fn,
-                    fetch_all_existing_docs_ids_fn=mock_fetch_all_existing_docs_ids_fn,
-                )
-            )
-
-            mock_retrieve.assert_called_once()
-            call_kwargs = mock_retrieve.call_args
-            assert call_kwargs.kwargs["start"] == indexing_start_dt.timestamp()
-
-
-def test_jira_doc_sync_passes_none_when_no_indexing_start(
-    jira_connector: JiraConnector,
-    mock_jira_cc_pair: MagicMock,
-    mock_fetch_all_existing_docs_fn: MagicMock,
-    mock_fetch_all_existing_docs_ids_fn: MagicMock,
-) -> None:
-    """Verify that indexing_start is None when the connector has no indexing_start set."""
-    mock_jira_cc_pair.connector.indexing_start = None
-
-    with patch("onyx.connectors.jira.connector.build_jira_client") as mock_build_client:
-        mock_build_client.return_value = jira_connector._jira_client
-        assert jira_connector._jira_client is not None
-        jira_connector._jira_client._options = MagicMock()
-        jira_connector._jira_client._options.return_value = {
-            "rest_api_version": JIRA_SERVER_API_VERSION
-        }
-
-        with patch.object(
-            type(jira_connector),
-            "retrieve_all_slim_docs_perm_sync",
-            return_value=iter([]),
-        ) as mock_retrieve:
-            list(
-                jira_doc_sync(
-                    cc_pair=mock_jira_cc_pair,
-                    fetch_all_existing_docs_fn=mock_fetch_all_existing_docs_fn,
-                    fetch_all_existing_docs_ids_fn=mock_fetch_all_existing_docs_ids_fn,
-                )
-            )
-
-            mock_retrieve.assert_called_once()
-            call_kwargs = mock_retrieve.call_args
-            assert call_kwargs.kwargs["start"] is None

backend/tests/unit/onyx/llm/test_multi_llm.py

@@ -11,6 +11,7 @@ from litellm.types.utils import ChatCompletionDeltaToolCall
 from litellm.types.utils import Delta
 from litellm.types.utils import Function as LiteLLMFunction
 
+import onyx.llm.models
 from onyx.configs.app_configs import MOCK_LLM_RESPONSE
 from onyx.llm.constants import LlmProviderNames
 from onyx.llm.interfaces import LLMUserIdentity
@@ -1479,6 +1480,147 @@ def test_bifrost_normalizes_api_base_in_model_kwargs() -> None:
     assert llm._model_kwargs["api_base"] == "https://bifrost.example.com/v1"
 
 
+def test_prompt_contains_tool_call_history_true() -> None:
+    from onyx.llm.multi_llm import _prompt_contains_tool_call_history
+
+    messages: LanguageModelInput = [
+        UserMessage(content="What's the weather?"),
+        AssistantMessage(
+            content=None,
+            tool_calls=[
+                ToolCall(
+                    id="tc_1",
+                    function=FunctionCall(name="get_weather", arguments="{}"),
+                )
+            ],
+        ),
+    ]
+    assert _prompt_contains_tool_call_history(messages) is True
+
+
+def test_prompt_contains_tool_call_history_false_no_tools() -> None:
+    from onyx.llm.multi_llm import _prompt_contains_tool_call_history
+
+    messages: LanguageModelInput = [
+        UserMessage(content="Hello"),
+        AssistantMessage(content="Hi there!"),
+    ]
+    assert _prompt_contains_tool_call_history(messages) is False
+
+
+def test_prompt_contains_tool_call_history_false_user_only() -> None:
+    from onyx.llm.multi_llm import _prompt_contains_tool_call_history
+
+    messages: LanguageModelInput = [UserMessage(content="Hello")]
+    assert _prompt_contains_tool_call_history(messages) is False
+
+
+def test_bedrock_claude_drops_thinking_when_thinking_blocks_missing() -> None:
+    """When thinking is enabled but assistant messages with tool_calls lack
+    thinking_blocks, the thinking param must be dropped to avoid the Bedrock
+    BadRequestError about missing thinking blocks."""
+    llm = LitellmLLM(
+        api_key=None,
+        timeout=30,
+        model_provider=LlmProviderNames.BEDROCK,
+        model_name="anthropic.claude-sonnet-4-20250514-v1:0",
+        max_input_tokens=200000,
+    )
+
+    messages: LanguageModelInput = [
+        UserMessage(content="What's the weather?"),
+        AssistantMessage(
+            content=None,
+            tool_calls=[
+                ToolCall(
+                    id="tc_1",
+                    function=FunctionCall(
+                        name="get_weather",
+                        arguments='{"city": "Paris"}',
+                    ),
+                )
+            ],
+        ),
+        onyx.llm.models.ToolMessage(
+            content="22°C sunny",
+            tool_call_id="tc_1",
+        ),
+    ]
+
+    tools = [
+        {
+            "type": "function",
+            "function": {
+                "name": "get_weather",
+                "description": "Get the weather",
+                "parameters": {
+                    "type": "object",
+                    "properties": {"city": {"type": "string"}},
+                },
+            },
+        }
+    ]
+
+    with (
+        patch("litellm.completion") as mock_completion,
+        patch("onyx.llm.multi_llm.model_is_reasoning_model", return_value=True),
+    ):
+        mock_completion.return_value = []
+
+        list(llm.stream(messages, tools=tools, reasoning_effort=ReasoningEffort.HIGH))
+
+        kwargs = mock_completion.call_args.kwargs
+        assert "thinking" not in kwargs, (
+            "thinking param should be dropped when thinking_blocks are missing "
+            "from assistant messages with tool_calls"
+        )
+
+
+def test_bedrock_claude_keeps_thinking_when_no_tool_history() -> None:
+    """When thinking is enabled and there are no historical assistant messages
+    with tool_calls, the thinking param should be preserved."""
+    llm = LitellmLLM(
+        api_key=None,
+        timeout=30,
+        model_provider=LlmProviderNames.BEDROCK,
+        model_name="anthropic.claude-sonnet-4-20250514-v1:0",
+        max_input_tokens=200000,
+    )
+
+    messages: LanguageModelInput = [
+        UserMessage(content="What's the weather?"),
+    ]
+
+    tools = [
+        {
+            "type": "function",
+            "function": {
+                "name": "get_weather",
+                "description": "Get the weather",
+                "parameters": {
+                    "type": "object",
+                    "properties": {"city": {"type": "string"}},
+                },
+            },
+        }
+    ]
+
+    with (
+        patch("litellm.completion") as mock_completion,
+        patch("onyx.llm.multi_llm.model_is_reasoning_model", return_value=True),
+    ):
+        mock_completion.return_value = []
+
+        list(llm.stream(messages, tools=tools, reasoning_effort=ReasoningEffort.HIGH))
+
+        kwargs = mock_completion.call_args.kwargs
+        assert "thinking" in kwargs, (
+            "thinking param should be preserved when no assistant messages "
+            "with tool_calls exist in history"
+        )
+        assert kwargs["thinking"]["type"] == "enabled"
+
+
 def test_bifrost_claude_includes_allowed_openai_params() -> None:
     llm = LitellmLLM(
         api_key="test_key",

backend/tests/unit/onyx/server/test_projects_file_utils.py

@@ -4,23 +4,13 @@ from unittest.mock import MagicMock
 import pytest
 from fastapi import UploadFile
 
-from onyx.natural_language_processing import utils as nlp_utils
-from onyx.natural_language_processing.utils import BaseTokenizer
-from onyx.natural_language_processing.utils import count_tokens
 from onyx.server.features.projects import projects_file_utils as utils
-from onyx.server.settings.models import Settings
 
 
-class _Tokenizer(BaseTokenizer):
+class _Tokenizer:
     def encode(self, text: str) -> list[int]:
         return [1] * len(text)
 
-    def tokenize(self, text: str) -> list[str]:
-        return list(text)
-
-    def decode(self, _tokens: list[int]) -> str:
-        return ""
-
 
 class _NonSeekableFile(BytesIO):
     def tell(self) -> int:
@@ -39,26 +29,10 @@ def _make_upload_no_size(filename: str, content: bytes) -> UploadFile:
     return UploadFile(filename=filename, file=BytesIO(content), size=None)
 
 
-def _make_settings(upload_size_mb: int = 1, token_threshold_k: int = 100) -> Settings:
-    return Settings(
-        user_file_max_upload_size_mb=upload_size_mb,
-        file_token_count_threshold_k=token_threshold_k,
-    )
-
-
-def _patch_common_dependencies(
-    monkeypatch: pytest.MonkeyPatch,
-    upload_size_mb: int = 1,
-    token_threshold_k: int = 100,
-) -> None:
+def _patch_common_dependencies(monkeypatch: pytest.MonkeyPatch) -> None:
     monkeypatch.setattr(utils, "fetch_default_llm_model", lambda _db: None)
     monkeypatch.setattr(utils, "get_tokenizer", lambda **_kwargs: _Tokenizer())
     monkeypatch.setattr(utils, "is_file_password_protected", lambda **_kwargs: False)
-    monkeypatch.setattr(
-        utils,
-        "load_settings",
-        lambda: _make_settings(upload_size_mb, token_threshold_k),
-    )
 
 
 def test_get_upload_size_bytes_falls_back_to_stream_size() -> None:
@@ -102,8 +76,9 @@ def test_is_upload_too_large_logs_warning_when_size_unknown(
 def test_categorize_uploaded_files_accepts_size_under_limit(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
-    # upload_size_mb=1 → max_bytes = 1*1024*1024; file size 99 is well under
-    _patch_common_dependencies(monkeypatch, upload_size_mb=1)
+    _patch_common_dependencies(monkeypatch)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_BYTES", 100)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_MB", 1)
     monkeypatch.setattr(utils, "estimate_image_tokens_for_upload", lambda _upload: 10)
 
     upload = _make_upload("small.png", size=99)
@@ -116,7 +91,9 @@ def test_categorize_uploaded_files_accepts_size_under_limit(
 def test_categorize_uploaded_files_uses_seek_fallback_when_upload_size_missing(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
-    _patch_common_dependencies(monkeypatch, upload_size_mb=1)
+    _patch_common_dependencies(monkeypatch)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_BYTES", 100)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_MB", 1)
     monkeypatch.setattr(utils, "estimate_image_tokens_for_upload", lambda _upload: 10)
 
     upload = _make_upload_no_size("small.png", content=b"x" * 99)
@@ -129,11 +106,12 @@ def test_categorize_uploaded_files_uses_seek_fallback_when_upload_size_missing(
 def test_categorize_uploaded_files_accepts_size_at_limit(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
-    _patch_common_dependencies(monkeypatch, upload_size_mb=1)
+    _patch_common_dependencies(monkeypatch)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_BYTES", 100)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_MB", 1)
     monkeypatch.setattr(utils, "estimate_image_tokens_for_upload", lambda _upload: 10)
 
-    # 1 MB = 1048576 bytes; file at exactly that boundary should be accepted
-    upload = _make_upload("edge.png", size=1048576)
+    upload = _make_upload("edge.png", size=100)
     result = utils.categorize_uploaded_files([upload], MagicMock())
 
     assert len(result.acceptable) == 1
@@ -143,10 +121,12 @@ def test_categorize_uploaded_files_accepts_size_at_limit(
 def test_categorize_uploaded_files_rejects_size_over_limit_with_reason(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
-    _patch_common_dependencies(monkeypatch, upload_size_mb=1)
+    _patch_common_dependencies(monkeypatch)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_BYTES", 100)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_MB", 1)
     monkeypatch.setattr(utils, "estimate_image_tokens_for_upload", lambda _upload: 10)
 
-    upload = _make_upload("large.png", size=1048577)  # 1 byte over 1 MB
+    upload = _make_upload("large.png", size=101)
     result = utils.categorize_uploaded_files([upload], MagicMock())
 
     assert len(result.acceptable) == 0
@@ -157,11 +137,13 @@ def test_categorize_uploaded_files_rejects_size_over_limit_with_reason(
 def test_categorize_uploaded_files_mixed_batch_keeps_valid_and_rejects_oversized(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
-    _patch_common_dependencies(monkeypatch, upload_size_mb=1)
+    _patch_common_dependencies(monkeypatch)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_BYTES", 100)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_MB", 1)
     monkeypatch.setattr(utils, "estimate_image_tokens_for_upload", lambda _upload: 10)
 
     small = _make_upload("small.png", size=50)
-    large = _make_upload("large.png", size=1048577)
+    large = _make_upload("large.png", size=101)
 
     result = utils.categorize_uploaded_files([small, large], MagicMock())
 
@@ -171,12 +153,15 @@ def test_categorize_uploaded_files_mixed_batch_keeps_valid_and_rejects_oversized
     assert result.rejected[0].reason == "Exceeds 1 MB file size limit"
 
 
-def test_categorize_uploaded_files_enforces_size_limit_always(
+def test_categorize_uploaded_files_enforces_size_limit_even_when_threshold_is_skipped(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
-    _patch_common_dependencies(monkeypatch, upload_size_mb=1)
+    _patch_common_dependencies(monkeypatch)
+    monkeypatch.setattr(utils, "SKIP_USERFILE_THRESHOLD", True)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_BYTES", 100)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_MB", 1)
 
-    upload = _make_upload("oversized.pdf", size=1048577)
+    upload = _make_upload("oversized.pdf", size=101)
     result = utils.categorize_uploaded_files([upload], MagicMock())
 
     assert len(result.acceptable) == 0
@@ -187,12 +172,14 @@ def test_categorize_uploaded_files_enforces_size_limit_always(
 def test_categorize_uploaded_files_checks_size_before_text_extraction(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
-    _patch_common_dependencies(monkeypatch, upload_size_mb=1)
+    _patch_common_dependencies(monkeypatch)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_BYTES", 100)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_MB", 1)
 
     extract_mock = MagicMock(return_value="this should not run")
     monkeypatch.setattr(utils, "extract_file_text", extract_mock)
 
-    oversized_doc = _make_upload("oversized.pdf", size=1048577)
+    oversized_doc = _make_upload("oversized.pdf", size=101)
     result = utils.categorize_uploaded_files([oversized_doc], MagicMock())
 
     extract_mock.assert_not_called()
@@ -201,219 +188,40 @@ def test_categorize_uploaded_files_checks_size_before_text_extraction(
     assert result.rejected[0].reason == "Exceeds 1 MB file size limit"
 
 
-def test_categorize_enforces_size_limit_when_upload_size_mb_is_positive(
+def test_categorize_uploaded_files_accepts_python_file(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
-    """A positive upload_size_mb is always enforced."""
-    _patch_common_dependencies(monkeypatch, upload_size_mb=1)
-    monkeypatch.setattr(utils, "estimate_image_tokens_for_upload", lambda _upload: 10)
+    _patch_common_dependencies(monkeypatch)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_BYTES", 10_000)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_MB", 1)
 
-    upload = _make_upload("huge.png", size=1048577, content=b"x")
-    result = utils.categorize_uploaded_files([upload], MagicMock())
-
-    assert len(result.acceptable) == 0
-    assert len(result.rejected) == 1
-
-
-def test_categorize_enforces_token_limit_when_threshold_k_is_positive(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """A positive token_threshold_k is always enforced."""
-    _patch_common_dependencies(monkeypatch, upload_size_mb=1000, token_threshold_k=5)
-    monkeypatch.setattr(utils, "estimate_image_tokens_for_upload", lambda _upload: 6000)
-
-    upload = _make_upload("big_image.png", size=100)
-    result = utils.categorize_uploaded_files([upload], MagicMock())
-
-    assert len(result.acceptable) == 0
-    assert len(result.rejected) == 1
-
-
-def test_categorize_no_token_limit_when_threshold_k_is_zero(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """token_threshold_k=0 means no token limit; high-token files are accepted."""
-    _patch_common_dependencies(monkeypatch, upload_size_mb=1000, token_threshold_k=0)
+    py_source = b'def hello():\n    print("world")\n'
     monkeypatch.setattr(
-        utils, "estimate_image_tokens_for_upload", lambda _upload: 999_999
+        utils, "extract_file_text", lambda **_kwargs: py_source.decode()
     )
 
-    upload = _make_upload("huge_image.png", size=100)
+    upload = _make_upload("script.py", size=len(py_source), content=py_source)
     result = utils.categorize_uploaded_files([upload], MagicMock())
 
-    assert len(result.rejected) == 0
     assert len(result.acceptable) == 1
+    assert result.acceptable[0].filename == "script.py"
+    assert len(result.rejected) == 0
 
 
-def test_categorize_both_limits_enforced(
+def test_categorize_uploaded_files_rejects_binary_file(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
-    """Both positive limits are enforced; file exceeding token limit is rejected."""
-    _patch_common_dependencies(monkeypatch, upload_size_mb=10, token_threshold_k=5)
-    monkeypatch.setattr(utils, "estimate_image_tokens_for_upload", lambda _upload: 6000)
+    _patch_common_dependencies(monkeypatch)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_BYTES", 10_000)
+    monkeypatch.setattr(utils, "USER_FILE_MAX_UPLOAD_SIZE_MB", 1)
 
-    upload = _make_upload("over_tokens.png", size=100)
+    monkeypatch.setattr(utils, "extract_file_text", lambda **_kwargs: "")
+
+    binary_content = bytes(range(256)) * 4
+    upload = _make_upload("data.bin", size=len(binary_content), content=binary_content)
     result = utils.categorize_uploaded_files([upload], MagicMock())
 
     assert len(result.acceptable) == 0
     assert len(result.rejected) == 1
-    assert result.rejected[0].reason == "Exceeds 5K token limit"
-
-
-def test_categorize_rejection_reason_contains_dynamic_values(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """Rejection reasons reflect the admin-configured limits, not hardcoded values."""
-    _patch_common_dependencies(monkeypatch, upload_size_mb=42, token_threshold_k=7)
-    monkeypatch.setattr(utils, "estimate_image_tokens_for_upload", lambda _upload: 8000)
-
-    # File within size limit but over token limit
-    upload = _make_upload("tokens.png", size=100)
-    result = utils.categorize_uploaded_files([upload], MagicMock())
-
-    assert result.rejected[0].reason == "Exceeds 7K token limit"
-
-    # File over size limit
-    _patch_common_dependencies(monkeypatch, upload_size_mb=42, token_threshold_k=7)
-    oversized = _make_upload("big.png", size=42 * 1024 * 1024 + 1)
-    result2 = utils.categorize_uploaded_files([oversized], MagicMock())
-
-    assert result2.rejected[0].reason == "Exceeds 42 MB file size limit"
-
-
-# --- count_tokens tests ---
-
-
-def test_count_tokens_small_text() -> None:
-    """Small text should be encoded in a single call and return correct count."""
-    tokenizer = _Tokenizer()
-    text = "hello world"
-    assert count_tokens(text, tokenizer) == len(tokenizer.encode(text))
-
-
-def test_count_tokens_chunked_matches_single_call() -> None:
-    """Chunked encoding should produce the same result as single-call for small text."""
-    tokenizer = _Tokenizer()
-    text = "a" * 1000
-    assert count_tokens(text, tokenizer) == len(tokenizer.encode(text))
-
-
-def test_count_tokens_large_text_is_chunked(monkeypatch: pytest.MonkeyPatch) -> None:
-    """Text exceeding _ENCODE_CHUNK_SIZE should be split into multiple encode calls."""
-    monkeypatch.setattr(nlp_utils, "_ENCODE_CHUNK_SIZE", 100)
-    tokenizer = _Tokenizer()
-    text = "a" * 250
-    # _Tokenizer returns 1 token per char, so total should be 250
-    assert count_tokens(text, tokenizer) == 250
-
-
-def test_count_tokens_with_token_limit_exits_early(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """When token_limit is set and exceeded, count_tokens should stop early."""
-    monkeypatch.setattr(nlp_utils, "_ENCODE_CHUNK_SIZE", 100)
-
-    encode_call_count = 0
-    original_tokenizer = _Tokenizer()
-
-    class _CountingTokenizer(BaseTokenizer):
-        def encode(self, text: str) -> list[int]:
-            nonlocal encode_call_count
-            encode_call_count += 1
-            return original_tokenizer.encode(text)
-
-        def tokenize(self, text: str) -> list[str]:
-            return list(text)
-
-        def decode(self, _tokens: list[int]) -> str:
-            return ""
-
-    tokenizer = _CountingTokenizer()
-    # 500 chars → 5 chunks of 100; limit=150 → should stop after 2 chunks
-    text = "a" * 500
-    result = count_tokens(text, tokenizer, token_limit=150)
-
-    assert result == 200  # 2 chunks × 100 tokens each
-    assert encode_call_count == 2, "Should have stopped after 2 chunks"
-
-
-def test_count_tokens_with_token_limit_not_exceeded(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """When token_limit is set but not exceeded, all chunks are encoded."""
-    monkeypatch.setattr(nlp_utils, "_ENCODE_CHUNK_SIZE", 100)
-    tokenizer = _Tokenizer()
-    text = "a" * 250
-    result = count_tokens(text, tokenizer, token_limit=1000)
-    assert result == 250
-
-
-def test_count_tokens_no_limit_encodes_all_chunks(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """Without token_limit, all chunks are encoded regardless of count."""
-    monkeypatch.setattr(nlp_utils, "_ENCODE_CHUNK_SIZE", 100)
-    tokenizer = _Tokenizer()
-    text = "a" * 500
-    result = count_tokens(text, tokenizer)
-    assert result == 500
-
-
-# --- early exit via token_limit in categorize tests ---
-
-
-def test_categorize_early_exits_tokenization_for_large_text(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """Large text files should be rejected via early-exit tokenization
-    without encoding all chunks."""
-    _patch_common_dependencies(monkeypatch, upload_size_mb=1000, token_threshold_k=1)
-    # token_threshold = 1000; _ENCODE_CHUNK_SIZE = 100 → text of 500 chars = 5 chunks
-    # Should stop after 2nd chunk (200 tokens > 1000? No... need 1 token per char)
-    # With _Tokenizer: 1 token per char. threshold=1000, chunk=100 → need 11 chunks
-    # Let's use a bigger text
-    monkeypatch.setattr(nlp_utils, "_ENCODE_CHUNK_SIZE", 100)
-    large_text = "x" * 5000  # 5000 tokens, threshold 1000
-    monkeypatch.setattr(utils, "extract_file_text", lambda **_kwargs: large_text)
-
-    encode_call_count = 0
-    original_tokenizer = _Tokenizer()
-
-    class _CountingTokenizer(BaseTokenizer):
-        def encode(self, text: str) -> list[int]:
-            nonlocal encode_call_count
-            encode_call_count += 1
-            return original_tokenizer.encode(text)
-
-        def tokenize(self, text: str) -> list[str]:
-            return list(text)
-
-        def decode(self, _tokens: list[int]) -> str:
-            return ""
-
-    monkeypatch.setattr(utils, "get_tokenizer", lambda **_kwargs: _CountingTokenizer())
-
-    upload = _make_upload("big.txt", size=5000, content=large_text.encode())
-    result = utils.categorize_uploaded_files([upload], MagicMock())
-
-    assert len(result.rejected) == 1
-    assert "token limit" in result.rejected[0].reason
-    # 5000 chars / 100 chunk_size = 50 chunks total; should stop well before all 50
-    assert (
-        encode_call_count < 50
-    ), f"Expected early exit but encoded {encode_call_count} chunks out of 50"
-
-
-def test_categorize_text_under_token_limit_accepted(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """Text files under the token threshold should be accepted with exact count."""
-    _patch_common_dependencies(monkeypatch, upload_size_mb=1000, token_threshold_k=1)
-    small_text = "x" * 500  # 500 tokens < 1000 threshold
-    monkeypatch.setattr(utils, "extract_file_text", lambda **_kwargs: small_text)
-
-    upload = _make_upload("ok.txt", size=500, content=small_text.encode())
-    result = utils.categorize_uploaded_files([upload], MagicMock())
-
-    assert len(result.acceptable) == 1
-    assert result.acceptable_file_to_token_count["ok.txt"] == 500
+    assert result.rejected[0].filename == "data.bin"
+    assert "Unsupported file type" in result.rejected[0].reason

backend/tests/unit/onyx/server/test_settings_store.py

@@ -1,23 +1,12 @@
 import pytest
 
-from onyx.configs.app_configs import DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB
 from onyx.key_value_store.interface import KvKeyNotFoundError
 from onyx.server.settings import store as settings_store
-from onyx.server.settings.models import (
-    DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_NO_VECTOR_DB,
-)
-from onyx.server.settings.models import DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_VECTOR_DB
-from onyx.server.settings.models import Settings
 
 
 class _FakeKvStore:
-    def __init__(self, data: dict | None = None) -> None:
-        self._data = data
-
     def load(self, _key: str) -> dict:
-        if self._data is None:
-            raise KvKeyNotFoundError()
-        return self._data
+        raise KvKeyNotFoundError()
 
 
 class _FakeCache:
@@ -31,140 +20,13 @@ class _FakeCache:
         self._vals[key] = value.encode("utf-8")
 
 
-def test_load_settings_uses_model_defaults_when_no_stored_value(
+def test_load_settings_includes_user_file_max_upload_size_mb(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
-    """When no settings are stored (vector DB enabled), load_settings() should
-    resolve the default token threshold to 200."""
     monkeypatch.setattr(settings_store, "get_kv_store", lambda: _FakeKvStore())
     monkeypatch.setattr(settings_store, "get_cache_backend", lambda: _FakeCache())
-    monkeypatch.setattr(settings_store, "DISABLE_VECTOR_DB", False)
+    monkeypatch.setattr(settings_store, "USER_FILE_MAX_UPLOAD_SIZE_MB", 77)
 
     settings = settings_store.load_settings()
 
-    assert settings.user_file_max_upload_size_mb == DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB
-    assert (
-        settings.file_token_count_threshold_k
-        == DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_VECTOR_DB
-    )
-
-
-def test_load_settings_uses_high_token_default_when_vector_db_disabled(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """When vector DB is disabled and no settings are stored, the token
-    threshold should default to 10000 (10M tokens)."""
-    monkeypatch.setattr(settings_store, "get_kv_store", lambda: _FakeKvStore())
-    monkeypatch.setattr(settings_store, "get_cache_backend", lambda: _FakeCache())
-    monkeypatch.setattr(settings_store, "DISABLE_VECTOR_DB", True)
-
-    settings = settings_store.load_settings()
-
-    assert settings.user_file_max_upload_size_mb == DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB
-    assert (
-        settings.file_token_count_threshold_k
-        == DEFAULT_FILE_TOKEN_COUNT_THRESHOLD_K_NO_VECTOR_DB
-    )
-
-
-def test_load_settings_preserves_explicit_value_when_vector_db_disabled(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """When vector DB is disabled but admin explicitly set a token threshold,
-    that value should be preserved (not overridden by the 10000 default)."""
-    stored = Settings(file_token_count_threshold_k=500).model_dump()
-    monkeypatch.setattr(settings_store, "get_kv_store", lambda: _FakeKvStore(stored))
-    monkeypatch.setattr(settings_store, "get_cache_backend", lambda: _FakeCache())
-    monkeypatch.setattr(settings_store, "DISABLE_VECTOR_DB", True)
-
-    settings = settings_store.load_settings()
-
-    assert settings.file_token_count_threshold_k == 500
-
-
-def test_load_settings_preserves_zero_token_threshold(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """A value of 0 means 'no limit' and should be preserved."""
-    stored = Settings(file_token_count_threshold_k=0).model_dump()
-    monkeypatch.setattr(settings_store, "get_kv_store", lambda: _FakeKvStore(stored))
-    monkeypatch.setattr(settings_store, "get_cache_backend", lambda: _FakeCache())
-    monkeypatch.setattr(settings_store, "DISABLE_VECTOR_DB", True)
-
-    settings = settings_store.load_settings()
-
-    assert settings.file_token_count_threshold_k == 0
-
-
-def test_load_settings_resolves_zero_upload_size_to_default(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """A value of 0 should be treated as unset and resolved to the default."""
-    stored = Settings(user_file_max_upload_size_mb=0).model_dump()
-    monkeypatch.setattr(settings_store, "get_kv_store", lambda: _FakeKvStore(stored))
-    monkeypatch.setattr(settings_store, "get_cache_backend", lambda: _FakeCache())
-
-    settings = settings_store.load_settings()
-
-    assert settings.user_file_max_upload_size_mb == DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB
-
-
-def test_load_settings_clamps_upload_size_to_env_max(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """When the stored upload size exceeds MAX_ALLOWED_UPLOAD_SIZE_MB, it should
-    be clamped to the env-configured maximum."""
-    stored = Settings(user_file_max_upload_size_mb=500).model_dump()
-    monkeypatch.setattr(settings_store, "get_kv_store", lambda: _FakeKvStore(stored))
-    monkeypatch.setattr(settings_store, "get_cache_backend", lambda: _FakeCache())
-    monkeypatch.setattr(settings_store, "MAX_ALLOWED_UPLOAD_SIZE_MB", 250)
-
-    settings = settings_store.load_settings()
-
-    assert settings.user_file_max_upload_size_mb == 250
-
-
-def test_load_settings_preserves_upload_size_within_max(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """When the stored upload size is within MAX_ALLOWED_UPLOAD_SIZE_MB, it should
-    be preserved unchanged."""
-    stored = Settings(user_file_max_upload_size_mb=150).model_dump()
-    monkeypatch.setattr(settings_store, "get_kv_store", lambda: _FakeKvStore(stored))
-    monkeypatch.setattr(settings_store, "get_cache_backend", lambda: _FakeCache())
-    monkeypatch.setattr(settings_store, "MAX_ALLOWED_UPLOAD_SIZE_MB", 250)
-
-    settings = settings_store.load_settings()
-
-    assert settings.user_file_max_upload_size_mb == 150
-
-
-def test_load_settings_zero_upload_size_resolves_to_default(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """A value of 0 should be treated as unset and resolved to the default,
-    clamped to MAX_ALLOWED_UPLOAD_SIZE_MB."""
-    stored = Settings(user_file_max_upload_size_mb=0).model_dump()
-    monkeypatch.setattr(settings_store, "get_kv_store", lambda: _FakeKvStore(stored))
-    monkeypatch.setattr(settings_store, "get_cache_backend", lambda: _FakeCache())
-    monkeypatch.setattr(settings_store, "MAX_ALLOWED_UPLOAD_SIZE_MB", 100)
-    monkeypatch.setattr(settings_store, "DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB", 100)
-
-    settings = settings_store.load_settings()
-
-    assert settings.user_file_max_upload_size_mb == 100
-
-
-def test_load_settings_default_clamped_to_max(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """When DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB exceeds MAX_ALLOWED_UPLOAD_SIZE_MB,
-    the effective default should be min(DEFAULT, MAX)."""
-    monkeypatch.setattr(settings_store, "get_kv_store", lambda: _FakeKvStore())
-    monkeypatch.setattr(settings_store, "get_cache_backend", lambda: _FakeCache())
-    monkeypatch.setattr(settings_store, "DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB", 100)
-    monkeypatch.setattr(settings_store, "MAX_ALLOWED_UPLOAD_SIZE_MB", 50)
-
-    settings = settings_store.load_settings()
-
-    assert settings.user_file_max_upload_size_mb == 50
+    assert settings.user_file_max_upload_size_mb == 77

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.TABULAR,
+                file_type=ChatFileType.CSV,
                 filename="data.csv",
                 content_text="csv,data\n1,2",
                 token_count=5,

deployment/data/nginx/app.conf.template

@@ -39,22 +39,6 @@ server {
     # Conditionally include MCP location configuration
     include /etc/nginx/conf.d/mcp.conf.inc;
 
-    location ~ ^/scim(/.*)?$ {
-        proxy_set_header X-Real-IP $remote_addr;
-        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
-        proxy_set_header X-Forwarded-Proto $scheme;
-        proxy_set_header X-Forwarded-Host $host;
-        proxy_set_header X-Forwarded-Port $server_port;
-        proxy_set_header Host $host;
-        proxy_http_version 1.1;
-        proxy_buffering off;
-        proxy_redirect off;
-        proxy_connect_timeout ${NGINX_PROXY_CONNECT_TIMEOUT}s;
-        proxy_send_timeout ${NGINX_PROXY_SEND_TIMEOUT}s;
-        proxy_read_timeout ${NGINX_PROXY_READ_TIMEOUT}s;
-        proxy_pass http://api_server;
-    }
-
     # Match both /api/* and /openapi.json in a single rule
     location ~ ^/(api|openapi.json)(/.*)?$ {
         # Rewrite /api prefixed matched paths

deployment/data/nginx/app.conf.template.no-letsencrypt

@@ -39,20 +39,6 @@ server {
     # Conditionally include MCP location configuration
     include /etc/nginx/conf.d/mcp.conf.inc;
 
-    location ~ ^/scim(/.*)?$ {
-        proxy_set_header X-Real-IP $remote_addr;
-        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
-        # don't trust client-supplied X-Forwarded-* headers — use nginx's own values
-        proxy_set_header X-Forwarded-Proto $scheme;
-        proxy_set_header X-Forwarded-Host $host;
-        proxy_set_header X-Forwarded-Port $server_port;
-        proxy_set_header Host $host;
-        proxy_http_version 1.1;
-        proxy_buffering off;
-        proxy_redirect off;
-        proxy_pass http://api_server;
-    }
-
     # Match both /api/* and /openapi.json in a single rule
     location ~ ^/(api|openapi.json)(/.*)?$ {
         # Rewrite /api prefixed matched paths

deployment/data/nginx/app.conf.template.prod

@@ -39,23 +39,6 @@ server {
     # Conditionally include MCP location configuration 
     include /etc/nginx/conf.d/mcp.conf.inc;
 
-    location ~ ^/scim(/.*)?$ {
-        proxy_set_header X-Real-IP $remote_addr;
-        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
-        # don't trust client-supplied X-Forwarded-* headers — use nginx's own values
-        proxy_set_header X-Forwarded-Proto $scheme;
-        proxy_set_header X-Forwarded-Host $host;
-        proxy_set_header X-Forwarded-Port $server_port;
-        proxy_set_header Host $host;
-        proxy_http_version 1.1;
-        proxy_buffering off;
-        proxy_redirect off;
-        proxy_connect_timeout ${NGINX_PROXY_CONNECT_TIMEOUT}s;
-        proxy_send_timeout ${NGINX_PROXY_SEND_TIMEOUT}s;
-        proxy_read_timeout ${NGINX_PROXY_READ_TIMEOUT}s;
-        proxy_pass http://api_server;
-    }
-
     # Match both /api/* and /openapi.json in a single rule
     location ~ ^/(api|openapi.json)(/.*)?$ {
         # Rewrite /api prefixed matched paths

deployment/docker_compose/env.prod.template

@@ -66,3 +66,10 @@ DB_READONLY_PASSWORD=password
 # Show extra/uncommon connectors
 # See https://docs.onyx.app/admins/connectors/overview for a full list of connectors
 SHOW_EXTRA_CONNECTORS=False
+
+# User File Upload Configuration
+# Skip the token count threshold check (100,000 tokens) for uploaded files
+# For self-hosted: set to true to skip for all users
+#SKIP_USERFILE_THRESHOLD=false
+# For multi-tenant: comma-separated list of tenant IDs to skip threshold
+#SKIP_USERFILE_THRESHOLD_TENANT_IDS=

deployment/docker_compose/env.template

@@ -35,10 +35,6 @@ USER_AUTH_SECRET=""
 
 ## Chat Configuration
 # HARD_DELETE_CHATS=
-# MAX_ALLOWED_UPLOAD_SIZE_MB=250
-# Default per-user upload size limit (MB) when no admin value is set.
-# Automatically clamped to MAX_ALLOWED_UPLOAD_SIZE_MB at runtime.
-# DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB=100
 
 ## Base URL for redirects
 # WEB_DOMAIN=
@@ -46,6 +42,13 @@ USER_AUTH_SECRET=""
 ## Enterprise Features, requires a paid plan and licenses
 ENABLE_PAID_ENTERPRISE_EDITION_FEATURES=false
 
+## User File Upload Configuration
+# Skip the token count threshold check (100,000 tokens) for uploaded files
+# For self-hosted: set to true to skip for all users
+# SKIP_USERFILE_THRESHOLD=false
+# For multi-tenant: comma-separated list of tenant IDs to skip threshold
+# SKIP_USERFILE_THRESHOLD_TENANT_IDS=
+
 
 ################################################################################
 ## SERVICES CONFIGURATIONS

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: 0.4.37
+version: 0.4.36
 appVersion: latest
 annotations:
   category: Productivity

deployment/helm/charts/onyx/templates/nginx-conf.yaml

@@ -63,22 +63,6 @@ data:
         }
         {{- end }}
 
-        location ~ ^/scim(/.*)?$ {
-            proxy_set_header X-Real-IP $remote_addr;
-            proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
-            proxy_set_header X-Forwarded-Proto $scheme;
-            proxy_set_header X-Forwarded-Host $host;
-            proxy_set_header Host $host;
-            proxy_http_version 1.1;
-            proxy_buffering off;
-            proxy_redirect off;
-            # timeout settings
-            proxy_connect_timeout {{ .Values.nginx.timeouts.connect }}s;
-            proxy_send_timeout {{ .Values.nginx.timeouts.send }}s;
-            proxy_read_timeout {{ .Values.nginx.timeouts.read }}s;
-            proxy_pass http://api_server;
-        }
-
         location ~ ^/(api|openapi\.json)(/.*)?$ {
             rewrite ^/api(/.*)$ $1 break;
             proxy_set_header X-Real-IP $remote_addr;

deployment/helm/charts/onyx/values.yaml

@@ -282,7 +282,7 @@ nginx:
     # The ingress-nginx subchart doesn't auto-detect our custom ConfigMap changes.
     # Workaround: Helm upgrade will restart if the following annotation value changes.
     podAnnotations:
-      onyx.app/nginx-config-version: "3"
+      onyx.app/nginx-config-version: "2"
 
     # Propagate DOMAIN into nginx so server_name continues to use the same env var
     extraEnvs:
@@ -1285,5 +1285,11 @@ configMap:
   DOMAIN: "localhost"
   # Chat Configs
   HARD_DELETE_CHATS: ""
-  MAX_ALLOWED_UPLOAD_SIZE_MB: ""
-  DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB: ""
+  # User File Upload Configuration
+  # Skip the token count threshold check (100,000 tokens) for uploaded files
+  # For self-hosted: set to true to skip for all users
+  SKIP_USERFILE_THRESHOLD: ""
+  # For multi-tenant: comma-separated list of tenant IDs to skip threshold
+  SKIP_USERFILE_THRESHOLD_TENANT_IDS: ""
+  # Maximum user upload file size in MB for chat/projects uploads
+  USER_FILE_MAX_UPLOAD_SIZE_MB: ""

tools/ods/README.md

@@ -28,7 +28,7 @@ Some commands require external tools to be installed and configured:
 - **uv** - Required for `backend` commands
   - Install from [docs.astral.sh/uv](https://docs.astral.sh/uv/)
 
-- **GitHub CLI** (`gh`) - Required for `run-ci`, `cherry-pick`, and `trace` commands
+- **GitHub CLI** (`gh`) - Required for `run-ci` and `cherry-pick` commands
   - Install from [cli.github.com](https://cli.github.com/)
   - Authenticate with `gh auth login`
 
@@ -412,62 +412,6 @@ The `compare` subcommand writes a `summary.json` alongside the report with aggre
 counts (changed, added, removed, unchanged). The HTML report is only generated when
 visual differences are detected.
 
-### `trace` - View Playwright Traces from CI
-
-Download Playwright trace artifacts from a GitHub Actions run and open them
-with `playwright show-trace`. Traces are only generated for failing tests
-(`retain-on-failure`).
-
-```shell
-ods trace [run-id-or-url]
-```
-
-The run can be specified as a numeric run ID, a full GitHub Actions URL, or
-omitted to find the latest Playwright run for the current branch.
-
-**Flags:**
-
-| Flag | Default | Description |
-|------|---------|-------------|
-| `--branch`, `-b` | | Find latest run for this branch |
-| `--pr` | | Find latest run for this PR number |
-| `--project`, `-p` | | Filter to a specific project (`admin`, `exclusive`, `lite`) |
-| `--list`, `-l` | `false` | List available traces without opening |
-| `--no-open` | `false` | Download traces but don't open them |
-
-When multiple traces are found, an interactive picker lets you select which
-traces to open. Use arrow keys or `j`/`k` to navigate, `space` to toggle,
-`a` to select all, `n` to deselect all, and `enter` to open. Falls back to a
-plain-text prompt when no TTY is available.
-
-Downloaded artifacts are cached in `/tmp/ods-traces/<run-id>/` so repeated
-invocations for the same run are instant.
-
-**Examples:**
-
-```shell
-# Latest run for the current branch
-ods trace
-
-# Specific run ID
-ods trace 12345678
-
-# Full GitHub Actions URL
-ods trace https://github.com/onyx-dot-app/onyx/actions/runs/12345678
-
-# Latest run for a PR
-ods trace --pr 9500
-
-# Latest run for a specific branch
-ods trace --branch main
-
-# Only download admin project traces
-ods trace --project admin
-
-# List traces without opening
-ods trace --list
-```
-
 ### Testing Changes Locally (Dry Run)
 
 Both `run-ci` and `cherry-pick` support `--dry-run` to test without making remote changes:

tools/ods/cmd/root.go

@@ -55,7 +55,6 @@ func NewRootCommand() *cobra.Command {
 	cmd.AddCommand(NewWebCommand())
 	cmd.AddCommand(NewLatestStableTagCommand())
 	cmd.AddCommand(NewWhoisCommand())
-	cmd.AddCommand(NewTraceCommand())
 
 	return cmd
 }

tools/ods/cmd/trace.go

@@ -1,556 +0,0 @@
-package cmd
-
-import (
-	"bufio"
-	"encoding/json"
-	"errors"
-	"fmt"
-	"os"
-	"os/exec"
-	"path/filepath"
-	"regexp"
-	"sort"
-	"strconv"
-	"strings"
-
-	log "github.com/sirupsen/logrus"
-	"github.com/spf13/cobra"
-
-	"github.com/onyx-dot-app/onyx/tools/ods/internal/git"
-	"github.com/onyx-dot-app/onyx/tools/ods/internal/paths"
-	"github.com/onyx-dot-app/onyx/tools/ods/internal/tui"
-)
-
-const playwrightWorkflow = "Run Playwright Tests"
-
-// TraceOptions holds options for the trace command
-type TraceOptions struct {
-	Branch  string
-	PR      string
-	Project string
-	List    bool
-	NoOpen  bool
-}
-
-// traceInfo describes a single trace.zip found in the downloaded artifacts.
-type traceInfo struct {
-	Path    string // absolute path to trace.zip
-	Project string // project group extracted from artifact dir (e.g. "admin", "admin-shard-1")
-	TestDir string // test directory name (human-readable-ish)
-}
-
-// NewTraceCommand creates a new trace command
-func NewTraceCommand() *cobra.Command {
-	opts := &TraceOptions{}
-
-	cmd := &cobra.Command{
-		Use:   "trace [run-id-or-url]",
-		Short: "Download and view Playwright traces from GitHub Actions",
-		Long: `Download Playwright trace artifacts from a GitHub Actions run and open them
-with 'playwright show-trace'.
-
-The run can be specified as:
-  - A GitHub Actions run ID (numeric)
-  - A full GitHub Actions run URL
-  - Omitted, to find the latest Playwright run for the current branch
-
-You can also look up the latest run by branch name or PR number.
-
-Examples:
-  ods trace                          # latest run for current branch
-  ods trace 12345678                 # specific run ID
-  ods trace https://github.com/onyx-dot-app/onyx/actions/runs/12345678
-  ods trace --pr 9500                # latest run for PR #9500
-  ods trace --branch main            # latest run for main branch
-  ods trace --project admin          # only download admin project traces
-  ods trace --list                   # list available traces without opening`,
-		Args: cobra.MaximumNArgs(1),
-		Run: func(cmd *cobra.Command, args []string) {
-			runTrace(args, opts)
-		},
-	}
-
-	cmd.Flags().StringVarP(&opts.Branch, "branch", "b", "", "Find latest run for this branch")
-	cmd.Flags().StringVar(&opts.PR, "pr", "", "Find latest run for this PR number")
-	cmd.Flags().StringVarP(&opts.Project, "project", "p", "", "Filter to a specific project (admin, exclusive, lite)")
-	cmd.Flags().BoolVarP(&opts.List, "list", "l", false, "List available traces without opening")
-	cmd.Flags().BoolVar(&opts.NoOpen, "no-open", false, "Download traces but don't open them")
-
-	return cmd
-}
-
-// ghRun represents a GitHub Actions workflow run from `gh run list`
-type ghRun struct {
-	DatabaseID int64  `json:"databaseId"`
-	Status     string `json:"status"`
-	Conclusion string `json:"conclusion"`
-	HeadBranch string `json:"headBranch"`
-	URL        string `json:"url"`
-}
-
-func runTrace(args []string, opts *TraceOptions) {
-	git.CheckGitHubCLI()
-
-	runID, err := resolveRunID(args, opts)
-	if err != nil {
-		log.Fatalf("Failed to resolve run: %v", err)
-	}
-
-	log.Infof("Using run ID: %s", runID)
-
-	destDir, err := downloadTraceArtifacts(runID, opts.Project)
-	if err != nil {
-		log.Fatalf("Failed to download artifacts: %v", err)
-	}
-
-	traces, err := findTraceInfos(destDir, runID)
-	if err != nil {
-		log.Fatalf("Failed to find traces: %v", err)
-	}
-
-	if len(traces) == 0 {
-		log.Info("No trace files found in the downloaded artifacts.")
-		log.Info("Traces are only generated for failing tests (retain-on-failure).")
-		return
-	}
-
-	projects := groupByProject(traces)
-
-	if opts.List || opts.NoOpen {
-		printTraceList(traces, projects)
-		fmt.Printf("\nTraces downloaded to: %s\n", destDir)
-		return
-	}
-
-	if len(traces) == 1 {
-		openTraces(traces)
-		return
-	}
-
-	for {
-		selected := selectTraces(traces, projects)
-		if len(selected) == 0 {
-			return
-		}
-		openTraces(selected)
-	}
-}
-
-// resolveRunID determines the run ID from the provided arguments and options.
-func resolveRunID(args []string, opts *TraceOptions) (string, error) {
-	if len(args) == 1 {
-		return parseRunIDFromArg(args[0])
-	}
-
-	if opts.PR != "" {
-		return findLatestRunForPR(opts.PR)
-	}
-
-	branch := opts.Branch
-	if branch == "" {
-		var err error
-		branch, err = git.GetCurrentBranch()
-		if err != nil {
-			return "", fmt.Errorf("failed to get current branch: %w", err)
-		}
-		if branch == "" {
-			return "", fmt.Errorf("detached HEAD; specify a --branch, --pr, or run ID")
-		}
-		log.Infof("Using current branch: %s", branch)
-	}
-
-	return findLatestRunForBranch(branch)
-}
-
-var runURLPattern = regexp.MustCompile(`/actions/runs/(\d+)`)
-
-// parseRunIDFromArg extracts a run ID from either a numeric string or a full URL.
-func parseRunIDFromArg(arg string) (string, error) {
-	if matched, _ := regexp.MatchString(`^\d+$`, arg); matched {
-		return arg, nil
-	}
-
-	matches := runURLPattern.FindStringSubmatch(arg)
-	if matches != nil {
-		return matches[1], nil
-	}
-
-	return "", fmt.Errorf("could not parse run ID from %q; expected a numeric ID or GitHub Actions URL", arg)
-}
-
-// findLatestRunForBranch finds the most recent Playwright workflow run for a branch.
-func findLatestRunForBranch(branch string) (string, error) {
-	log.Infof("Looking up latest Playwright run for branch: %s", branch)
-
-	cmd := exec.Command("gh", "run", "list",
-		"--workflow", playwrightWorkflow,
-		"--branch", branch,
-		"--limit", "1",
-		"--json", "databaseId,status,conclusion,headBranch,url",
-	)
-	output, err := cmd.Output()
-	if err != nil {
-		return "", ghError(err, "gh run list failed")
-	}
-
-	var runs []ghRun
-	if err := json.Unmarshal(output, &runs); err != nil {
-		return "", fmt.Errorf("failed to parse run list: %w", err)
-	}
-
-	if len(runs) == 0 {
-		return "", fmt.Errorf("no Playwright runs found for branch %q", branch)
-	}
-
-	run := runs[0]
-	log.Infof("Found run: %s (status: %s, conclusion: %s)", run.URL, run.Status, run.Conclusion)
-	return fmt.Sprintf("%d", run.DatabaseID), nil
-}
-
-// findLatestRunForPR finds the most recent Playwright workflow run for a PR.
-func findLatestRunForPR(prNumber string) (string, error) {
-	log.Infof("Looking up branch for PR #%s", prNumber)
-
-	cmd := exec.Command("gh", "pr", "view", prNumber,
-		"--json", "headRefName",
-		"--jq", ".headRefName",
-	)
-	output, err := cmd.Output()
-	if err != nil {
-		return "", ghError(err, "gh pr view failed")
-	}
-
-	branch := strings.TrimSpace(string(output))
-	if branch == "" {
-		return "", fmt.Errorf("could not determine branch for PR #%s", prNumber)
-	}
-
-	log.Infof("PR #%s is on branch: %s", prNumber, branch)
-	return findLatestRunForBranch(branch)
-}
-
-// downloadTraceArtifacts downloads playwright trace artifacts for a run.
-// Returns the path to the download directory.
-func downloadTraceArtifacts(runID string, project string) (string, error) {
-	cacheKey := runID
-	if project != "" {
-		cacheKey = runID + "-" + project
-	}
-	destDir := filepath.Join(os.TempDir(), "ods-traces", cacheKey)
-
-	// Reuse a previous download if traces exist
-	if info, err := os.Stat(destDir); err == nil && info.IsDir() {
-		traces, _ := findTraces(destDir)
-		if len(traces) > 0 {
-			log.Infof("Using cached download at %s", destDir)
-			return destDir, nil
-		}
-		_ = os.RemoveAll(destDir)
-	}
-
-	if err := os.MkdirAll(destDir, 0755); err != nil {
-		return "", fmt.Errorf("failed to create directory %s: %w", destDir, err)
-	}
-
-	ghArgs := []string{"run", "download", runID, "--dir", destDir}
-
-	if project != "" {
-		ghArgs = append(ghArgs, "--pattern", fmt.Sprintf("playwright-test-results-%s-*", project))
-	} else {
-		ghArgs = append(ghArgs, "--pattern", "playwright-test-results-*")
-	}
-
-	log.Infof("Downloading trace artifacts...")
-	log.Debugf("Running: gh %s", strings.Join(ghArgs, " "))
-
-	cmd := exec.Command("gh", ghArgs...)
-	cmd.Stdout = os.Stdout
-	cmd.Stderr = os.Stderr
-	if err := cmd.Run(); err != nil {
-		_ = os.RemoveAll(destDir)
-		return "", fmt.Errorf("gh run download failed: %w\nMake sure the run ID is correct and the artifacts haven't expired (30 day retention)", err)
-	}
-
-	return destDir, nil
-}
-
-// findTraces recursively finds all trace.zip files under a directory.
-func findTraces(root string) ([]string, error) {
-	var traces []string
-	err := filepath.Walk(root, func(path string, info os.FileInfo, err error) error {
-		if err != nil {
-			return err
-		}
-		if !info.IsDir() && info.Name() == "trace.zip" {
-			traces = append(traces, path)
-		}
-		return nil
-	})
-	return traces, err
-}
-
-// findTraceInfos walks the download directory and returns structured trace info.
-// Expects: destDir/{artifact-dir}/{test-dir}/trace.zip
-func findTraceInfos(destDir, runID string) ([]traceInfo, error) {
-	var traces []traceInfo
-	err := filepath.Walk(destDir, func(path string, info os.FileInfo, err error) error {
-		if err != nil {
-			return err
-		}
-		if info.IsDir() || info.Name() != "trace.zip" {
-			return nil
-		}
-
-		rel, _ := filepath.Rel(destDir, path)
-		parts := strings.SplitN(rel, string(filepath.Separator), 3)
-
-		artifactDir := ""
-		testDir := filepath.Base(filepath.Dir(path))
-		if len(parts) >= 2 {
-			artifactDir = parts[0]
-			testDir = parts[1]
-		}
-
-		traces = append(traces, traceInfo{
-			Path:    path,
-			Project: extractProject(artifactDir, runID),
-			TestDir: testDir,
-		})
-		return nil
-	})
-
-	sort.Slice(traces, func(i, j int) bool {
-		pi, pj := projectSortKey(traces[i].Project), projectSortKey(traces[j].Project)
-		if pi != pj {
-			return pi < pj
-		}
-		return traces[i].TestDir < traces[j].TestDir
-	})
-
-	return traces, err
-}
-
-// extractProject derives a project group from an artifact directory name.
-// e.g. "playwright-test-results-admin-12345" -> "admin"
-//
-//	"playwright-test-results-admin-shard-1-12345" -> "admin-shard-1"
-func extractProject(artifactDir, runID string) string {
-	name := strings.TrimPrefix(artifactDir, "playwright-test-results-")
-	name = strings.TrimSuffix(name, "-"+runID)
-	if name == "" {
-		return artifactDir
-	}
-	return name
-}
-
-// projectSortKey returns a sort-friendly key that orders admin < exclusive < lite.
-func projectSortKey(project string) string {
-	switch {
-	case strings.HasPrefix(project, "admin"):
-		return "0-" + project
-	case strings.HasPrefix(project, "exclusive"):
-		return "1-" + project
-	case strings.HasPrefix(project, "lite"):
-		return "2-" + project
-	default:
-		return "3-" + project
-	}
-}
-
-// groupByProject returns an ordered list of unique project names found in traces.
-func groupByProject(traces []traceInfo) []string {
-	seen := map[string]bool{}
-	var projects []string
-	for _, t := range traces {
-		if !seen[t.Project] {
-			seen[t.Project] = true
-			projects = append(projects, t.Project)
-		}
-	}
-	sort.Slice(projects, func(i, j int) bool {
-		return projectSortKey(projects[i]) < projectSortKey(projects[j])
-	})
-	return projects
-}
-
-// printTraceList displays traces grouped by project.
-func printTraceList(traces []traceInfo, projects []string) {
-	fmt.Printf("\nFound %d trace(s) across %d project(s):\n", len(traces), len(projects))
-
-	idx := 1
-	for _, proj := range projects {
-		count := 0
-		for _, t := range traces {
-			if t.Project == proj {
-				count++
-			}
-		}
-		fmt.Printf("\n  %s (%d):\n", proj, count)
-		for _, t := range traces {
-			if t.Project == proj {
-				fmt.Printf("    [%2d] %s\n", idx, t.TestDir)
-				idx++
-			}
-		}
-	}
-}
-
-// selectTraces tries the TUI picker first, falling back to a plain-text
-// prompt when the terminal cannot be initialised (e.g. piped output).
-func selectTraces(traces []traceInfo, projects []string) []traceInfo {
-	// Build picker groups in the same order as the sorted traces slice.
-	var groups []tui.PickerGroup
-	for _, proj := range projects {
-		var items []string
-		for _, t := range traces {
-			if t.Project == proj {
-				items = append(items, t.TestDir)
-			}
-		}
-		groups = append(groups, tui.PickerGroup{Label: proj, Items: items})
-	}
-
-	indices, err := tui.Pick(groups)
-	if err != nil {
-		// Terminal not available — fall back to text prompt
-		log.Debugf("TUI picker unavailable: %v", err)
-		printTraceList(traces, projects)
-		return promptTraceSelection(traces, projects)
-	}
-	if indices == nil {
-		return nil // user cancelled
-	}
-
-	selected := make([]traceInfo, len(indices))
-	for i, idx := range indices {
-		selected[i] = traces[idx]
-	}
-	return selected
-}
-
-// promptTraceSelection asks the user which traces to open via plain text.
-// Accepts numbers (1,3,5), ranges (1-5), "all", or a project name.
-func promptTraceSelection(traces []traceInfo, projects []string) []traceInfo {
-	fmt.Printf("\nOpen which traces? (e.g. 1,3,5 | 1-5 | all | %s): ", strings.Join(projects, " | "))
-
-	reader := bufio.NewReader(os.Stdin)
-	input, err := reader.ReadString('\n')
-	if err != nil {
-		log.Fatalf("Failed to read input: %v", err)
-	}
-	input = strings.TrimSpace(input)
-
-	if input == "" || strings.EqualFold(input, "all") {
-		return traces
-	}
-
-	// Check if input matches a project name
-	for _, proj := range projects {
-		if strings.EqualFold(input, proj) {
-			var selected []traceInfo
-			for _, t := range traces {
-				if t.Project == proj {
-					selected = append(selected, t)
-				}
-			}
-			return selected
-		}
-	}
-
-	// Parse as number/range selection
-	indices := parseTraceSelection(input, len(traces))
-	if len(indices) == 0 {
-		log.Warn("No valid selection; opening all traces")
-		return traces
-	}
-
-	selected := make([]traceInfo, len(indices))
-	for i, idx := range indices {
-		selected[i] = traces[idx]
-	}
-	return selected
-}
-
-// parseTraceSelection parses a comma-separated list of numbers and ranges into
-// 0-based indices. Input is 1-indexed (matches display). Out-of-range values
-// are silently ignored.
-func parseTraceSelection(input string, max int) []int {
-	var result []int
-	seen := map[int]bool{}
-
-	for _, part := range strings.Split(input, ",") {
-		part = strings.TrimSpace(part)
-		if part == "" {
-			continue
-		}
-
-		if idx := strings.Index(part, "-"); idx > 0 {
-			lo, err1 := strconv.Atoi(strings.TrimSpace(part[:idx]))
-			hi, err2 := strconv.Atoi(strings.TrimSpace(part[idx+1:]))
-			if err1 != nil || err2 != nil {
-				continue
-			}
-			for i := lo; i <= hi; i++ {
-				zi := i - 1
-				if zi >= 0 && zi < max && !seen[zi] {
-					result = append(result, zi)
-					seen[zi] = true
-				}
-			}
-		} else {
-			n, err := strconv.Atoi(part)
-			if err != nil {
-				continue
-			}
-			zi := n - 1
-			if zi >= 0 && zi < max && !seen[zi] {
-				result = append(result, zi)
-				seen[zi] = true
-			}
-		}
-	}
-
-	return result
-}
-
-// openTraces opens the selected traces with playwright show-trace,
-// running npx from the web/ directory to use the project's Playwright version.
-func openTraces(traces []traceInfo) {
-	tracePaths := make([]string, len(traces))
-	for i, t := range traces {
-		tracePaths[i] = t.Path
-	}
-
-	args := append([]string{"playwright", "show-trace"}, tracePaths...)
-
-	log.Infof("Opening %d trace(s) with playwright show-trace...", len(traces))
-	cmd := exec.Command("npx", args...)
-
-	// Run from web/ to pick up the locally-installed Playwright version
-	if root, err := paths.GitRoot(); err == nil {
-		cmd.Dir = filepath.Join(root, "web")
-	}
-
-	cmd.Stdout = os.Stdout
-	cmd.Stderr = os.Stderr
-	cmd.Stdin = os.Stdin
-
-	if err := cmd.Run(); err != nil {
-		var exitErr *exec.ExitError
-		if errors.As(err, &exitErr) {
-			// Normal exit (e.g. user closed the window) — just log and return
-			// so the picker loop can continue.
-			log.Debugf("playwright exited with code %d", exitErr.ExitCode())
-			return
-		}
-		log.Errorf("playwright show-trace failed: %v\nMake sure Playwright is installed (npx playwright install)", err)
-	}
-}
-
-// ghError wraps a gh CLI error with stderr output.
-func ghError(err error, msg string) error {
-	if exitErr, ok := err.(*exec.ExitError); ok {
-		return fmt.Errorf("%s: %w: %s", msg, err, string(exitErr.Stderr))
-	}
-	return fmt.Errorf("%s: %w", msg, err)
-}

tools/ods/go.mod

@@ -3,19 +3,13 @@ module github.com/onyx-dot-app/onyx/tools/ods
 go 1.26.0
 
 require (
-	github.com/gdamore/tcell/v2 v2.13.8
 	github.com/jmelahman/tag v0.5.2
-	github.com/sirupsen/logrus v1.9.4
+	github.com/sirupsen/logrus v1.9.3
 	github.com/spf13/cobra v1.10.2
 	github.com/spf13/pflag v1.0.10
 )
 
 require (
-	github.com/gdamore/encoding v1.0.1 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
-	github.com/lucasb-eyer/go-colorful v1.3.0 // indirect
-	github.com/rivo/uniseg v0.4.7 // indirect
-	golang.org/x/sys v0.42.0 // indirect
-	golang.org/x/term v0.41.0 // indirect
-	golang.org/x/text v0.35.0 // indirect
+	golang.org/x/sys v0.39.0 // indirect
 )

tools/ods/go.sum

@@ -1,68 +1,30 @@
 github.com/cpuguy83/go-md2man/v2 v2.0.6/go.mod h1:oOW0eioCTA6cOiMLiUPZOpcVxMig6NIQQ7OS05n1F4g=
+github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
-github.com/gdamore/encoding v1.0.1 h1:YzKZckdBL6jVt2Gc+5p82qhrGiqMdG/eNs6Wy0u3Uhw=
-github.com/gdamore/encoding v1.0.1/go.mod h1:0Z0cMFinngz9kS1QfMjCP8TY7em3bZYeeklsSDPivEo=
-github.com/gdamore/tcell/v2 v2.13.8 h1:Mys/Kl5wfC/GcC5Cx4C2BIQH9dbnhnkPgS9/wF3RlfU=
-github.com/gdamore/tcell/v2 v2.13.8/go.mod h1:+Wfe208WDdB7INEtCsNrAN6O2m+wsTPk1RAovjaILlo=
 github.com/inconshreveable/mousetrap v1.1.0 h1:wN+x4NVGpMsO7ErUn/mUI3vEoE6Jt13X2s0bqwp9tc8=
 github.com/inconshreveable/mousetrap v1.1.0/go.mod h1:vpF70FUmC8bwa3OWnCshd2FqLfsEA9PFc4w1p2J65bw=
 github.com/jmelahman/tag v0.5.2 h1:g6A/aHehu5tkA31mPoDsXBNr1FigZ9A82Y8WVgb/WsM=
 github.com/jmelahman/tag v0.5.2/go.mod h1:qmuqk19B1BKkpcg3kn7l/Eey+UqucLxgOWkteUGiG4Q=
-github.com/lucasb-eyer/go-colorful v1.3.0 h1:2/yBRLdWBZKrf7gB40FoiKfAWYQ0lqNcbuQwVHXptag=
-github.com/lucasb-eyer/go-colorful v1.3.0/go.mod h1:R4dSotOR9KMtayYi1e77YzuveK+i7ruzyGqttikkLy0=
 github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
-github.com/rivo/uniseg v0.4.7 h1:WUdvkW8uEhrYfLC4ZzdpI2ztxP1I582+49Oc5Mq64VQ=
-github.com/rivo/uniseg v0.4.7/go.mod h1:FN3SvrM+Zdj16jyLfmOkMNblXMcoc8DfTHruCPUcx88=
 github.com/russross/blackfriday/v2 v2.1.0/go.mod h1:+Rmxgy9KzJVeS9/2gXHxylqXiyQDYRxCVz55jmeOWTM=
-github.com/sirupsen/logrus v1.9.4 h1:TsZE7l11zFCLZnZ+teH4Umoq5BhEIfIzfRDZ1Uzql2w=
-github.com/sirupsen/logrus v1.9.4/go.mod h1:ftWc9WdOfJ0a92nsE2jF5u5ZwH8Bv2zdeOC42RjbV2g=
+github.com/sirupsen/logrus v1.9.3 h1:dueUQJ1C2q9oE3F7wvmSGAaVtTmUizReu6fjN8uqzbQ=
+github.com/sirupsen/logrus v1.9.3/go.mod h1:naHLuLoDiP4jHNo9R0sCBMtWGeIprob74mVsIT4qYEQ=
 github.com/spf13/cobra v1.10.2 h1:DMTTonx5m65Ic0GOoRY2c16WCbHxOOw6xxezuLaBpcU=
 github.com/spf13/cobra v1.10.2/go.mod h1:7C1pvHqHw5A4vrJfjNwvOdzYu0Gml16OCs2GRiTUUS4=
 github.com/spf13/pflag v1.0.9/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=
 github.com/spf13/pflag v1.0.10 h1:4EBh2KAYBwaONj6b2Ye1GiHfwjqyROoF4RwYO+vPwFk=
 github.com/spf13/pflag v1.0.10/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=
+github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
+github.com/stretchr/testify v1.7.0/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
 github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U=
 github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
-github.com/yuin/goldmark v1.4.13/go.mod h1:6yULJ656Px+3vBD8DxQVa3kxgyrAnzto9xy5taEt/CY=
 go.yaml.in/yaml/v3 v3.0.4/go.mod h1:DhzuOOF2ATzADvBadXxruRBLzYTpT36CKvDb3+aBEFg=
-golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
-golang.org/x/crypto v0.0.0-20210921155107-089bfa567519/go.mod h1:GvvjBRRGRdwPK5ydBHafDWAxML/pGHZbMvKqRZ5+Abc=
-golang.org/x/mod v0.6.0-dev.0.20220419223038-86c51ed26bb4/go.mod h1:jJ57K6gSWd91VN4djpZkiMVwK6gcyfeH4XE8wZrZaV4=
-golang.org/x/mod v0.8.0/go.mod h1:iBbtSCu2XBx23ZKBPSOrRkjjQPZFPuis4dIYUhu/chs=
-golang.org/x/net v0.0.0-20190620200207-3b0461eec859/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
-golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
-golang.org/x/net v0.0.0-20220722155237-a158d28d115b/go.mod h1:XRhObCWvk6IyKnWLug+ECip1KBveYUHfp+8e9klMJ9c=
-golang.org/x/net v0.6.0/go.mod h1:2Tu9+aMcznHK/AK1HMvgo6xiTLG5rD5rZLDS+rp2Bjs=
-golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
-golang.org/x/sync v0.0.0-20220722155255-886fb9371eb4/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
-golang.org/x/sync v0.1.0/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
-golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
-golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20210615035016-665e8c7367d1/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20220520151302-bc2c85ada10a/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20220722155257-8c9f86f7a55f/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.5.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.42.0 h1:omrd2nAlyT5ESRdCLYdm3+fMfNFE/+Rf4bDIQImRJeo=
-golang.org/x/sys v0.42.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
-golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
-golang.org/x/term v0.0.0-20210927222741-03fcf44c2211/go.mod h1:jbD1KX2456YbFQfuXm/mYQcufACuNUgVhRMnK/tPxf8=
-golang.org/x/term v0.5.0/go.mod h1:jMB1sMXY+tzblOD4FWmEbocvup2/aLOaQEp7JmGp78k=
-golang.org/x/term v0.41.0 h1:QCgPso/Q3RTJx2Th4bDLqML4W6iJiaXFq2/ftQF13YU=
-golang.org/x/term v0.41.0/go.mod h1:3pfBgksrReYfZ5lvYM0kSO0LIkAl4Yl2bXOkKP7Ec2A=
-golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
-golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
-golang.org/x/text v0.3.7/go.mod h1:u+2+/6zg+i71rQMx5EYifcz6MCKuco9NR6JIITiCfzQ=
-golang.org/x/text v0.7.0/go.mod h1:mrYo+phRRbMaCq/xk9113O4dZlRixOauAjOtrjsXDZ8=
-golang.org/x/text v0.14.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
-golang.org/x/text v0.35.0 h1:JOVx6vVDFokkpaq1AEptVzLTpDe9KGpj5tR4/X+ybL8=
-golang.org/x/text v0.35.0/go.mod h1:khi/HExzZJ2pGnjenulevKNX1W67CUy0AsXcNubPGCA=
-golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
-golang.org/x/tools v0.0.0-20191119224855-298f0cb1881e/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
-golang.org/x/tools v0.1.12/go.mod h1:hNGJHUnrk76NpqgfD5Aqm5Crs+Hm0VOH/i9J2+nxYbc=
-golang.org/x/tools v0.6.0/go.mod h1:Xwgl3UAJ/d3gWutnCtw505GrjyAbvKui8lOU390QaIU=
-golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
+golang.org/x/sys v0.0.0-20220715151400-c0bba94af5f8/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.39.0 h1:CvCKL8MeisomCi6qNZ+wbb0DN9E5AATixKsvNtMoMFk=
+golang.org/x/sys v0.39.0/go.mod h1:OgkHotnGiDImocRcuBABYBEXf8A9a87e/uXjp9XT3ks=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
+gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
 gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
 gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=

tools/ods/internal/tui/picker.go

@@ -1,419 +0,0 @@
-package tui
-
-import (
-	"fmt"
-
-	"github.com/gdamore/tcell/v2"
-)
-
-// PickerGroup represents a labelled group of selectable items.
-type PickerGroup struct {
-	Label string
-	Items []string
-}
-
-// entry is a single row in the picker (either a group header or an item).
-type entry struct {
-	label    string
-	isHeader bool
-	selected bool
-	groupIdx int
-	flatIdx  int // index across all items (ignoring headers), -1 for headers
-}
-
-// Pick shows a full-screen grouped multi-select picker.
-// All items start deselected. Returns the flat indices of selected items
-// (0-based, spanning all groups in order). Returns nil if cancelled.
-// Returns a non-nil error if the terminal cannot be initialised, in which
-// case the caller should fall back to a simpler prompt.
-func Pick(groups []PickerGroup) ([]int, error) {
-	screen, err := tcell.NewScreen()
-	if err != nil {
-		return nil, err
-	}
-	if err := screen.Init(); err != nil {
-		return nil, err
-	}
-	defer screen.Fini()
-
-	entries := buildEntries(groups)
-	totalItems := countItems(entries)
-	cursor := firstSelectableIndex(entries)
-	offset := 0
-
-	for {
-		w, h := screen.Size()
-		selectedCount := countSelected(entries)
-
-		drawPicker(screen, entries, groups, cursor, offset, w, h, selectedCount, totalItems)
-		screen.Show()
-
-		ev := screen.PollEvent()
-		switch ev := ev.(type) {
-		case *tcell.EventResize:
-			screen.Sync()
-		case *tcell.EventKey:
-			switch action := keyAction(ev); action {
-			case actionQuit:
-				return nil, nil
-			case actionConfirm:
-				if countSelected(entries) > 0 {
-					return collectSelected(entries), nil
-				}
-			case actionUp:
-				if cursor > 0 {
-					cursor--
-				}
-			case actionDown:
-				if cursor < len(entries)-1 {
-					cursor++
-				}
-			case actionTop:
-				cursor = 0
-			case actionBottom:
-				if len(entries) == 0 {
-					cursor = 0
-				} else {
-					cursor = len(entries) - 1
-				}
-			case actionPageUp:
-				listHeight := h - headerLines - footerLines
-				cursor -= listHeight
-				if cursor < 0 {
-					cursor = 0
-				}
-			case actionPageDown:
-				listHeight := h - headerLines - footerLines
-				cursor += listHeight
-				if cursor >= len(entries) {
-					cursor = len(entries) - 1
-				}
-			case actionToggle:
-				toggleAtCursor(entries, cursor)
-			case actionAll:
-				setAll(entries, true)
-			case actionNone:
-				setAll(entries, false)
-			}
-
-			// Keep the cursor visible
-			listHeight := h - headerLines - footerLines
-			if listHeight < 1 {
-				listHeight = 1
-			}
-			if cursor < offset {
-				offset = cursor
-			}
-			if cursor >= offset+listHeight {
-				offset = cursor - listHeight + 1
-			}
-		}
-	}
-}
-
-// --- actions ----------------------------------------------------------------
-
-type action int
-
-const (
-	actionNoop action = iota
-	actionQuit
-	actionConfirm
-	actionUp
-	actionDown
-	actionTop
-	actionBottom
-	actionPageUp
-	actionPageDown
-	actionToggle
-	actionAll
-	actionNone
-)
-
-func keyAction(ev *tcell.EventKey) action {
-	switch ev.Key() {
-	case tcell.KeyEscape, tcell.KeyCtrlC:
-		return actionQuit
-	case tcell.KeyEnter:
-		return actionConfirm
-	case tcell.KeyUp:
-		return actionUp
-	case tcell.KeyDown:
-		return actionDown
-	case tcell.KeyHome:
-		return actionTop
-	case tcell.KeyEnd:
-		return actionBottom
-	case tcell.KeyPgUp:
-		return actionPageUp
-	case tcell.KeyPgDn:
-		return actionPageDown
-	case tcell.KeyRune:
-		switch ev.Rune() {
-		case 'q':
-			return actionQuit
-		case ' ':
-			return actionToggle
-		case 'j':
-			return actionDown
-		case 'k':
-			return actionUp
-		case 'g':
-			return actionTop
-		case 'G':
-			return actionBottom
-		case 'a':
-			return actionAll
-		case 'n':
-			return actionNone
-		}
-	}
-	return actionNoop
-}
-
-// --- data helpers ------------------------------------------------------------
-
-func buildEntries(groups []PickerGroup) []entry {
-	var entries []entry
-	flat := 0
-	for gi, g := range groups {
-		entries = append(entries, entry{
-			label:    g.Label,
-			isHeader: true,
-			groupIdx: gi,
-			flatIdx:  -1,
-		})
-		for _, item := range g.Items {
-			entries = append(entries, entry{
-				label:    item,
-				isHeader: false,
-				selected: false,
-				groupIdx: gi,
-				flatIdx:  flat,
-			})
-			flat++
-		}
-	}
-	return entries
-}
-
-func firstSelectableIndex(entries []entry) int {
-	for i, e := range entries {
-		if !e.isHeader {
-			return i
-		}
-	}
-	return 0
-}
-
-func countItems(entries []entry) int {
-	n := 0
-	for _, e := range entries {
-		if !e.isHeader {
-			n++
-		}
-	}
-	return n
-}
-
-func countSelected(entries []entry) int {
-	n := 0
-	for _, e := range entries {
-		if !e.isHeader && e.selected {
-			n++
-		}
-	}
-	return n
-}
-
-func collectSelected(entries []entry) []int {
-	var result []int
-	for _, e := range entries {
-		if !e.isHeader && e.selected {
-			result = append(result, e.flatIdx)
-		}
-	}
-	return result
-}
-
-func toggleAtCursor(entries []entry, cursor int) {
-	if cursor < 0 || cursor >= len(entries) {
-		return
-	}
-	e := entries[cursor]
-	if e.isHeader {
-		// Toggle entire group: if all selected -> deselect all, else select all
-		allSelected := true
-		for _, e2 := range entries {
-			if !e2.isHeader && e2.groupIdx == e.groupIdx && !e2.selected {
-				allSelected = false
-				break
-			}
-		}
-		for i := range entries {
-			if !entries[i].isHeader && entries[i].groupIdx == e.groupIdx {
-				entries[i].selected = !allSelected
-			}
-		}
-	} else {
-		entries[cursor].selected = !entries[cursor].selected
-	}
-}
-
-func setAll(entries []entry, selected bool) {
-	for i := range entries {
-		if !entries[i].isHeader {
-			entries[i].selected = selected
-		}
-	}
-}
-
-// --- drawing ----------------------------------------------------------------
-
-const (
-	headerLines = 2 // title + blank line
-	footerLines = 2 // blank line + keybinds
-)
-
-var (
-	styleDefault    = tcell.StyleDefault
-	styleTitle      = tcell.StyleDefault.Bold(true)
-	styleGroup      = tcell.StyleDefault.Bold(true).Foreground(tcell.ColorTeal)
-	styleGroupCur   = tcell.StyleDefault.Bold(true).Foreground(tcell.ColorTeal).Reverse(true)
-	styleCheck      = tcell.StyleDefault.Foreground(tcell.ColorGreen).Bold(true)
-	styleUncheck    = tcell.StyleDefault.Dim(true)
-	styleItem       = tcell.StyleDefault
-	styleItemCur    = tcell.StyleDefault.Bold(true).Underline(true)
-	styleCheckCur   = tcell.StyleDefault.Foreground(tcell.ColorGreen).Bold(true).Underline(true)
-	styleUncheckCur = tcell.StyleDefault.Dim(true).Underline(true)
-	styleFooter     = tcell.StyleDefault.Dim(true)
-)
-
-func drawPicker(
-	screen tcell.Screen,
-	entries []entry,
-	groups []PickerGroup,
-	cursor, offset, w, h, selectedCount, totalItems int,
-) {
-	screen.Clear()
-
-	// Title
-	title := fmt.Sprintf(" Select traces to open (%d/%d selected)", selectedCount, totalItems)
-	drawLine(screen, 0, 0, w, title, styleTitle)
-
-	// List area
-	listHeight := h - headerLines - footerLines
-	if listHeight < 1 {
-		listHeight = 1
-	}
-
-	for i := 0; i < listHeight; i++ {
-		ei := offset + i
-		if ei >= len(entries) {
-			break
-		}
-		y := headerLines + i
-		renderEntry(screen, entries, groups, ei, cursor, w, y)
-	}
-
-	// Scrollbar hint
-	if len(entries) > listHeight {
-		drawScrollbar(screen, w-1, headerLines, listHeight, offset, len(entries))
-	}
-
-	// Footer
-	footerY := h - 1
-	footer := " ↑/↓ move  space toggle  a all  n none  enter open  q/esc quit"
-	drawLine(screen, 0, footerY, w, footer, styleFooter)
-}
-
-func renderEntry(screen tcell.Screen, entries []entry, groups []PickerGroup, ei, cursor, w, y int) {
-	e := entries[ei]
-	isCursor := ei == cursor
-
-	if e.isHeader {
-		groupSelected := 0
-		groupTotal := 0
-		for _, e2 := range entries {
-			if !e2.isHeader && e2.groupIdx == e.groupIdx {
-				groupTotal++
-				if e2.selected {
-					groupSelected++
-				}
-			}
-		}
-
-		label := fmt.Sprintf("  %s (%d/%d)", e.label, groupSelected, groupTotal)
-		style := styleGroup
-		if isCursor {
-			style = styleGroupCur
-		}
-		drawLine(screen, 0, y, w, label, style)
-		return
-	}
-
-	// Item row: "    [x] label" or "  > [x] label"
-	prefix := "    "
-	if isCursor {
-		prefix = "  > "
-	}
-
-	check := "[ ]"
-	cStyle := styleUncheck
-	iStyle := styleItem
-	if isCursor {
-		cStyle = styleUncheckCur
-		iStyle = styleItemCur
-	}
-	if e.selected {
-		check = "[x]"
-		cStyle = styleCheck
-		if isCursor {
-			cStyle = styleCheckCur
-		}
-	}
-
-	x := drawStr(screen, 0, y, w, prefix, iStyle)
-	x = drawStr(screen, x, y, w, check, cStyle)
-	drawStr(screen, x, y, w, " "+e.label, iStyle)
-}
-
-func drawScrollbar(screen tcell.Screen, x, top, height, offset, total int) {
-	if total <= height || height < 1 {
-		return
-	}
-
-	thumbSize := max(1, height*height/total)
-	thumbPos := top + offset*height/total
-
-	for y := top; y < top+height; y++ {
-		ch := '│'
-		style := styleDefault.Dim(true)
-		if y >= thumbPos && y < thumbPos+thumbSize {
-			ch = '┃'
-			style = styleDefault
-		}
-		screen.SetContent(x, y, ch, nil, style)
-	}
-}
-
-// drawLine fills an entire row starting at x=startX, padding to width w.
-func drawLine(screen tcell.Screen, startX, y, w int, s string, style tcell.Style) {
-	x := drawStr(screen, startX, y, w, s, style)
-	// Clear the rest of the line
-	for ; x < w; x++ {
-		screen.SetContent(x, y, ' ', nil, style)
-	}
-}
-
-// drawStr writes a string at (x, y) up to maxX and returns the next x position.
-func drawStr(screen tcell.Screen, x, y, maxX int, s string, style tcell.Style) int {
-	for _, ch := range s {
-		if x >= maxX {
-			break
-		}
-		screen.SetContent(x, y, ch, nil, style)
-		x++
-	}
-	return x
-}

web/package-lock.json

@@ -7901,9 +7901,7 @@
       }
     },
     "node_modules/anymatch/node_modules/picomatch": {
-      "version": "2.3.2",
-      "resolved": "https://registry.npmjs.org/picomatch/-/picomatch-2.3.2.tgz",
-      "integrity": "sha512-V7+vQEJ06Z+c5tSye8S+nHUfI51xoXIXjHQ99cQtKUkQqqO1kO/KCJUfZXuB47h/YBlDhah2H3hdUGXn8ie0oA==",
+      "version": "2.3.1",
       "license": "MIT",
       "engines": {
         "node": ">=8.6"
@@ -10703,9 +10701,7 @@
       "license": "MIT"
     },
     "node_modules/handlebars": {
-      "version": "4.7.9",
-      "resolved": "https://registry.npmjs.org/handlebars/-/handlebars-4.7.9.tgz",
-      "integrity": "sha512-4E71E0rpOaQuJR2A3xDZ+GM1HyWYv1clR58tC8emQNeQe3RH7MAzSbat+V0wG78LQBo6m6bzSG/L4pBuCsgnUQ==",
+      "version": "4.7.8",
       "dev": true,
       "license": "MIT",
       "dependencies": {
@@ -12559,9 +12555,7 @@
       }
     },
     "node_modules/jest-util/node_modules/picomatch": {
-      "version": "2.3.2",
-      "resolved": "https://registry.npmjs.org/picomatch/-/picomatch-2.3.2.tgz",
-      "integrity": "sha512-V7+vQEJ06Z+c5tSye8S+nHUfI51xoXIXjHQ99cQtKUkQqqO1kO/KCJUfZXuB47h/YBlDhah2H3hdUGXn8ie0oA==",
+      "version": "2.3.1",
       "dev": true,
       "license": "MIT",
       "engines": {
@@ -13887,9 +13881,7 @@
       }
     },
     "node_modules/micromatch/node_modules/picomatch": {
-      "version": "2.3.2",
-      "resolved": "https://registry.npmjs.org/picomatch/-/picomatch-2.3.2.tgz",
-      "integrity": "sha512-V7+vQEJ06Z+c5tSye8S+nHUfI51xoXIXjHQ99cQtKUkQqqO1kO/KCJUfZXuB47h/YBlDhah2H3hdUGXn8ie0oA==",
+      "version": "2.3.1",
       "license": "MIT",
       "engines": {
         "node": ">=8.6"
@@ -15009,9 +15001,7 @@
       "license": "ISC"
     },
     "node_modules/picomatch": {
-      "version": "4.0.4",
-      "resolved": "https://registry.npmjs.org/picomatch/-/picomatch-4.0.4.tgz",
-      "integrity": "sha512-QP88BAKvMam/3NxH6vj2o21R6MjxZUAd6nlwAS/pnGvN9IVLocLHxGYIzFhg6fUQ+5th6P4dv4eW9jX3DSIj7A==",
+      "version": "4.0.3",
       "license": "MIT",
       "engines": {
         "node": ">=12"
@@ -15899,9 +15889,7 @@
       }
     },
     "node_modules/readdirp/node_modules/picomatch": {
-      "version": "2.3.2",
-      "resolved": "https://registry.npmjs.org/picomatch/-/picomatch-2.3.2.tgz",
-      "integrity": "sha512-V7+vQEJ06Z+c5tSye8S+nHUfI51xoXIXjHQ99cQtKUkQqqO1kO/KCJUfZXuB47h/YBlDhah2H3hdUGXn8ie0oA==",
+      "version": "2.3.1",
       "license": "MIT",
       "engines": {
         "node": ">=8.6"

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

@@ -2,7 +2,6 @@
 
 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";
@@ -24,8 +23,9 @@ 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 } from "@opal/components";
+import { Button, Text } 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 mainContentEmphasis text05>
+          <Text font="main-content-emphasis" color="text-05">
             Image Generation Model
           </Text>
-          <Text secondaryBody text03>
+          <Text font="secondary-body" color="text-03">
             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 secondaryBody text03>
+            <Text font="secondary-body" color="text-03">
               {group.name}
             </Text>
             <div className="flex flex-col gap-2">
@@ -277,12 +277,13 @@ export default function ImageGenerationContent() {
           {needsReplacement ? (
             hasReplacements ? (
               <Section alignItems="start">
-                <Text as="p" text03>
-                  <b>{disconnectProvider.title}</b> is currently the default
-                  image generation model. Session history will be preserved.
+                <Text as="p" color="text-03">
+                  {markdown(
+                    `**${disconnectProvider.title}** is currently the default image generation model. Session history will be preserved.`
+                  )}
                 </Text>
                 <Section alignItems="start" gap={0.25}>
-                  <Text as="p" text04>
+                  <Text as="p" color="text-04">
                     Set New Default
                   </Text>
                   <InputSelect
@@ -329,22 +330,24 @@ export default function ImageGenerationContent() {
               </Section>
             ) : (
               <>
-                <Text as="p" text03>
-                  <b>{disconnectProvider.title}</b> is currently the default
-                  image generation model.
+                <Text as="p" color="text-03">
+                  {markdown(
+                    `**${disconnectProvider.title}** is currently the default image generation model.`
+                  )}
                 </Text>
-                <Text as="p" text03>
+                <Text as="p" color="text-03">
                   Connect another provider to continue using image generation.
                 </Text>
               </>
             )
           ) : (
             <>
-              <Text as="p" text03>
-                <b>{disconnectProvider.title}</b> models will no longer be used
-                to generate images.
+              <Text as="p" color="text-03">
+                {markdown(
+                  `**${disconnectProvider.title}** models will no longer be used to generate images.`
+                )}
               </Text>
-              <Text as="p" text03>
+              <Text as="p" color="text-03">
                 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 as OpalButton } from "@opal/components";
+import { Button } 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={
-        <OpalButton
+        <Button
           variant="danger"
           onClick={onDisconnect}
           disabled={
@@ -159,7 +159,7 @@ function WebSearchDisconnectModal({
           }
         >
           Disconnect
-        </OpalButton>
+        </Button>
       }
     >
       {needsReplacement ? (

web/src/app/app/interfaces.ts

@@ -75,14 +75,14 @@ export enum ChatFileType {
   IMAGE = "image",
   DOCUMENT = "document",
   PLAIN_TEXT = "plain_text",
-  TABULAR = "tabular",
+  CSV = "csv",
   USER_KNOWLEDGE = "user_knowledge",
 }
 
 export const isTextFile = (fileType: ChatFileType) =>
   [
     ChatFileType.PLAIN_TEXT,
-    ChatFileType.TABULAR,
+    ChatFileType.CSV,
     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.TABULAR);
+  const csvFiles = files.filter((file) => file.type === ChatFileType.CSV);
 
   const presentingDocument: MinimalOnyxDocument = {
     document_id: previewingFile?.id ?? "",

web/src/interfaces/settings.ts

@@ -37,7 +37,6 @@ export interface Settings {
   // User Knowledge settings
   user_knowledge_enabled?: boolean;
   user_file_max_upload_size_mb?: number | null;
-  file_token_count_threshold_k?: number | null;
 
   // Connector settings
   show_extra_connectors?: boolean;
@@ -69,12 +68,6 @@ export interface Settings {
 
   // Application version from the ONYX_VERSION env var on the server.
   version?: string | null;
-  // Hard ceiling for user_file_max_upload_size_mb, derived from env var.
-  max_allowed_upload_size_mb?: number;
-
-  // Factory defaults for the restore button.
-  default_user_file_max_upload_size_mb?: number;
-  default_file_token_count_threshold_k?: number;
 }
 
 export enum NotificationType {

web/src/providers/ProjectsContext.tsx

@@ -85,6 +85,8 @@ function buildFileKey(file: File): string {
   return `${file.size}|${namePrefix}`;
 }
 
+const DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB = 50;
+
 interface ProjectsContextType {
   projects: Project[];
   recentFiles: ProjectFile[];
@@ -339,20 +341,21 @@ export function ProjectsProvider({ children }: ProjectsProviderProps) {
       onFailure?: (failedTempIds: string[]) => void
     ): Promise<ProjectFile[]> => {
       const rawMax = settingsContext?.settings?.user_file_max_upload_size_mb;
+      const maxUploadSizeMb =
+        rawMax && rawMax > 0 ? rawMax : DEFAULT_USER_FILE_MAX_UPLOAD_SIZE_MB;
+      const maxUploadSizeBytes = maxUploadSizeMb * 1024 * 1024;
 
-      const oversizedFiles =
-        rawMax && rawMax > 0
-          ? files.filter((file) => file.size > rawMax * 1024 * 1024)
-          : [];
-      const validFiles =
-        rawMax && rawMax > 0
-          ? files.filter((file) => file.size <= rawMax * 1024 * 1024)
-          : files;
+      const oversizedFiles = files.filter(
+        (file) => file.size > maxUploadSizeBytes
+      );
+      const validFiles = files.filter(
+        (file) => file.size <= maxUploadSizeBytes
+      );
 
       if (oversizedFiles.length > 0) {
         const skippedNames = oversizedFiles.map((file) => file.name).join(", ");
         toast.warning(
-          `Skipped ${oversizedFiles.length} oversized file(s) (>${rawMax} MB): ${skippedNames}`
+          `Skipped ${oversizedFiles.length} oversized file(s) (>${maxUploadSizeMb} MB): ${skippedNames}`
         );
       }
 

web/src/refresh-components/Popover.tsx

@@ -142,6 +142,7 @@ 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]
@@ -226,7 +227,7 @@ export function PopoverMenu({
   });
 
   return (
-    <Section alignItems="stretch">
+    <Section alignItems="stretch" height="auto" className="flex-1 min-h-0">
       <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">
+    <div className="relative min-h-0 flex flex-col">
       <div
         ref={containerRef}
         className={cn("overflow-y-auto", className)}

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

@@ -25,7 +25,6 @@ import {
   SvgFold,
   SvgExternalLink,
   SvgAlertCircle,
-  SvgRefreshCw,
 } from "@opal/icons";
 import { ADMIN_ROUTES } from "@/lib/admin-routes";
 import { Content } from "@opal/layouts";
@@ -55,7 +54,6 @@ import * as ExpandableCard from "@/layouts/expandable-card-layouts";
 import * as ActionsLayouts from "@/layouts/actions-layouts";
 import { getActionIcon } from "@/lib/tools/mcpUtils";
 import { Disabled } from "@opal/core";
-import IconButton from "@/refresh-components/buttons/IconButton";
 import InputTypeIn from "@/refresh-components/inputs/InputTypeIn";
 import useFilter from "@/hooks/useFilter";
 import { MCPServer } from "@/lib/tools/interfaces";
@@ -83,10 +81,6 @@ interface ChatPreferencesFormValues {
   maximum_chat_retention_days: string;
   anonymous_user_enabled: boolean;
   disable_default_assistant: boolean;
-
-  // File limits
-  user_file_max_upload_size_mb: string;
-  file_token_count_threshold_k: string;
 }
 
 interface MCPServerCardTool {
@@ -191,173 +185,6 @@ function MCPServerCard({
   );
 }
 
-type FileLimitFieldName =
-  | "user_file_max_upload_size_mb"
-  | "file_token_count_threshold_k";
-
-interface NumericLimitFieldProps {
-  name: FileLimitFieldName;
-  defaultValue: string;
-  saveSettings: (updates: Partial<Settings>) => Promise<void>;
-  maxValue?: number;
-  allowZero?: boolean;
-}
-
-function NumericLimitField({
-  name,
-  defaultValue,
-  saveSettings,
-  maxValue,
-  allowZero = false,
-}: NumericLimitFieldProps) {
-  const { values, setFieldValue } =
-    useFormikContext<ChatPreferencesFormValues>();
-  const initialValue = useRef(values[name]);
-  const restoringRef = useRef(false);
-  const value = values[name];
-
-  const parsed = parseInt(value, 10);
-  const isOverMax =
-    maxValue !== undefined && !isNaN(parsed) && parsed > maxValue;
-
-  const handleRestore = () => {
-    restoringRef.current = true;
-    initialValue.current = defaultValue;
-    void setFieldValue(name, defaultValue);
-    void saveSettings({ [name]: parseInt(defaultValue, 10) });
-  };
-
-  const handleBlur = () => {
-    // The restore button triggers a blur — skip since handleRestore already saved.
-    if (restoringRef.current) {
-      restoringRef.current = false;
-      return;
-    }
-
-    const parsed = parseInt(value, 10);
-    const isValid = !isNaN(parsed) && (allowZero ? parsed >= 0 : parsed > 0);
-
-    // Revert invalid input (empty, NaN, negative).
-    if (!isValid) {
-      if (allowZero) {
-        // Empty/invalid means "no limit" — persist 0 and clear the field.
-        void setFieldValue(name, "");
-        void saveSettings({ [name]: 0 });
-        initialValue.current = "";
-      } else {
-        void setFieldValue(name, initialValue.current);
-      }
-      return;
-    }
-
-    // Block save when the value exceeds the hard ceiling.
-    if (maxValue !== undefined && parsed > maxValue) {
-      return;
-    }
-
-    // For allowZero fields, 0 means "no limit" — clear the display
-    // so the "No limit" placeholder is visible, but still persist 0.
-    if (allowZero && parsed === 0) {
-      void setFieldValue(name, "");
-      if (initialValue.current !== "") {
-        void saveSettings({ [name]: 0 });
-        initialValue.current = "";
-      }
-      return;
-    }
-
-    const normalizedDisplay = String(parsed);
-
-    // Update the display to the canonical form (e.g. strip leading zeros).
-    if (value !== normalizedDisplay) {
-      void setFieldValue(name, normalizedDisplay);
-    }
-
-    // Persist only when the value actually changed.
-    if (normalizedDisplay !== initialValue.current) {
-      void saveSettings({ [name]: parsed });
-      initialValue.current = normalizedDisplay;
-    }
-  };
-
-  return (
-    <div className="group w-full">
-      <InputTypeInField
-        name={name}
-        inputMode="numeric"
-        showClearButton={false}
-        pattern="[0-9]*"
-        placeholder={allowZero ? "No limit" : `Default: ${defaultValue}`}
-        variant={isOverMax ? "error" : undefined}
-        rightSection={
-          (value || "") !== defaultValue ? (
-            <div className="opacity-0 group-hover:opacity-100 group-focus-within:opacity-100 transition-opacity">
-              <IconButton
-                icon={SvgRefreshCw}
-                tooltip="Restore default"
-                internal
-                type="button"
-                onClick={handleRestore}
-              />
-            </div>
-          ) : undefined
-        }
-        onBlur={handleBlur}
-      />
-    </div>
-  );
-}
-
-interface FileSizeLimitFieldsProps {
-  saveSettings: (updates: Partial<Settings>) => Promise<void>;
-  defaultUploadSizeMb: string;
-  defaultTokenThresholdK: string;
-  maxAllowedUploadSizeMb?: number;
-}
-
-function FileSizeLimitFields({
-  saveSettings,
-  defaultUploadSizeMb,
-  defaultTokenThresholdK,
-  maxAllowedUploadSizeMb,
-}: FileSizeLimitFieldsProps) {
-  return (
-    <div className="flex gap-4 w-full items-start">
-      <div className="flex-1">
-        <InputLayouts.Vertical
-          title="File Size Limit (MB)"
-          subDescription={
-            maxAllowedUploadSizeMb
-              ? `Max: ${maxAllowedUploadSizeMb} MB`
-              : undefined
-          }
-          nonInteractive
-        >
-          <NumericLimitField
-            name="user_file_max_upload_size_mb"
-            defaultValue={defaultUploadSizeMb}
-            saveSettings={saveSettings}
-            maxValue={maxAllowedUploadSizeMb}
-          />
-        </InputLayouts.Vertical>
-      </div>
-      <div className="flex-1">
-        <InputLayouts.Vertical
-          title="File Token Limit (thousand tokens)"
-          nonInteractive
-        >
-          <NumericLimitField
-            name="file_token_count_threshold_k"
-            defaultValue={defaultTokenThresholdK}
-            saveSettings={saveSettings}
-            allowZero
-          />
-        </InputLayouts.Vertical>
-      </div>
-    </div>
-  );
-}
-
 /**
  * Inner form component that uses useFormikContext to access values
  * and create save handlers for settings fields.
@@ -374,7 +201,6 @@ function ChatPreferencesForm() {
   // Tools availability
   const { tools: availableTools } = useAvailableTools();
   const vectorDbEnabled = useVectorDbEnabled();
-
   const searchTool = availableTools.find(
     (t) => t.in_code_tool_id === SEARCH_TOOL_ID
   );
@@ -897,28 +723,6 @@ function ChatPreferencesForm() {
                   </InputLayouts.Horizontal>
                 </Card>
 
-                <Card>
-                  <InputLayouts.Vertical
-                    title="File Attachment Size Limit"
-                    description="Files attached in chats and projects must fit within both limits to be accepted. Larger files increase latency, memory usage, and token costs."
-                  >
-                    <FileSizeLimitFields
-                      saveSettings={saveSettings}
-                      defaultUploadSizeMb={
-                        settings?.settings.default_user_file_max_upload_size_mb?.toString() ??
-                        "100"
-                      }
-                      defaultTokenThresholdK={
-                        settings?.settings.default_file_token_count_threshold_k?.toString() ??
-                        "200"
-                      }
-                      maxAllowedUploadSizeMb={
-                        settings?.settings.max_allowed_upload_size_mb
-                      }
-                    />
-                  </InputLayouts.Vertical>
-                </Card>
-
                 <Card>
                   <InputLayouts.Horizontal
                     title="Allow Anonymous Users"
@@ -1058,21 +862,6 @@ export default function ChatPreferencesPage() {
     anonymous_user_enabled: settings.settings.anonymous_user_enabled ?? false,
     disable_default_assistant:
       settings.settings.disable_default_assistant ?? false,
-
-    // File limits — for upload size: 0/null means "use default";
-    // for token threshold: null means "use default", 0 means "no limit".
-    user_file_max_upload_size_mb:
-      (settings.settings.user_file_max_upload_size_mb ?? 0) <= 0
-        ? settings.settings.default_user_file_max_upload_size_mb?.toString() ??
-          "100"
-        : settings.settings.user_file_max_upload_size_mb!.toString(),
-    file_token_count_threshold_k:
-      settings.settings.file_token_count_threshold_k == null
-        ? settings.settings.default_file_token_count_threshold_k?.toString() ??
-          "200"
-        : settings.settings.file_token_count_threshold_k === 0
-          ? ""
-          : settings.settings.file_token_count_threshold_k.toString(),
   };
 
   return (

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

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

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

@@ -4,6 +4,14 @@ 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 }) => {
@@ -11,21 +19,33 @@ for (const theme of THEMES) {
     });
 
     test("should screenshot each settings tab", async ({ page }) => {
-      await page.goto("/app/settings");
-      await page.waitForLoadState("networkidle");
+      await page.goto("/app/settings/general");
+      await page
+        .getByTestId("settings-left-tab-navigation")
+        .waitFor({ state: "visible" });
 
       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();
-        await page.waitForLoadState("networkidle");
+
+        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 expectScreenshot(page, {
           name: `settings-${theme}-${slug}`,