* chore: bump Python backend dependencies, drop unused peewee
Minor/patch + reviewed major bumps across requirements.txt,
requirements-min.txt, pyproject.toml and uv.lock; playwright image bumped in
docker-compose.playwright.yaml. peewee/peewee-migrate removed (zero imports).
Security-relevant: cryptography 46->48, authlib 1.6.10->1.7.2, PyJWT 2.11->2.13,
requests 2.33.1->2.34.2, RestrictedPython 8.1->8.2, pillow 12.1.1->12.2.0.
Reviewed majors: redis 7->8, pymilvus ->2.6.14, azure-search-documents 11->12,
chardet 5->7, unstructured 0.18->0.22, pycrdt 0.12->0.13.
Testing:
- Resolution: `uv lock` resolves the full bumped set with no conflicts; uv.lock
regenerated to match (peewee dropped, every pin including
azure-search-documents==12.0.0 resolves).
- Per-dependency contract tests (external tests repo, unit/deps/): 105 files,
2205 passed / 6 skipped, ruff-clean. One file per dependency pins the symbols,
signatures and behaviour the backend actually uses, so an API removal/rename in
a bumped version fails loudly instead of at runtime. Offline/deterministic.
- End-to-end embed->retrieve test driving transformers + sentence-transformers +
chromadb together through Open WebUI's real RAG path (cached model, in-memory
chroma, semantic retrieval asserted).
- Install/startup/health resolution gate added to the dep-bump workflow and the
integration suite (uv/pip resolve + uvicorn /health + Playwright dev visibility).
- Bugs surfaced while testing each got an isolated fix branch + regression test:
Mistral OCR aiohttp FilePayload (#25779), chroma has_collection (#25780),
aiocache per-user model-cache key (security), otel semconv deprecation,
pydub/audioop <3.13 note.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Bump python-multipart 0.0.22 -> 0.0.27 (CVE-2026-42561, CVE-2026-40347)
0.0.22 is affected by two DoS CVEs in the multipart parser that
Starlette/FastAPI run for every multipart/form-data request, so any
authenticated user hitting an upload endpoint can trigger them:
- CVE-2026-42561: unbounded part-header count/size -> CPU exhaustion (fixed 0.0.27)
- CVE-2026-40347: large multipart preamble/epilogue DoS (fixed 0.0.26)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Add a security-reporting link to the issue chooser
With blank issues disabled, the New Issue chooser only offered Bug Report and
Feature Request, leaving security reporters no obvious route and nudging them
toward filing vulnerabilities as public issues. Add a contact_links entry that
points to the Security Policy (/security/policy), where the "Report a
vulnerability" button opens a private GitHub advisory — keeping security
reports out of public issues.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Update config.yml
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Replace the recursive parseJSONString with an equivalent iterative unwrap.
The recursion re-parsed its own already-parsed result until JSON.parse threw;
on scalar JSON values (e.g. "5" -> 5) that recursed until a stack overflow
which was then silently caught — wasted work on every complete tool-call
payload. The loop returns the identical value in all cases (verified
byte-identical across 29 inputs incl. double/triple-encoded and partial JSON)
without the stack churn.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
has_access_to_file() derives file access from the objects a file is attached to
(knowledge bases, workspace models). Those branches returned True for any access_type
whenever the user held that permission on the object, write/delete included. Since a
user can create their own KB or model and attach any file they can merely READ (KB
attach and the model meta.knowledge validator both gate on read access only), a user
with read access to a victim file could launder it into write/delete: attach it to an
object they own, then rename, overwrite or delete it via the write-gated file routes
(POST /files/{id}/rename, /data/content/update, DELETE /files/{id}). This is the
residual of GHSA-vjqm-6gcc-62cr (CVE-2026-54012) left open by the read-only attach
validator (CWE-863).
An object now confers write/delete on a file only when the object's owner owns that
file, so delegation originates from the file's own owner. Read is unchanged (RAG and
shared-object reads still work), and legitimate delegation is preserved: a write grant
on an object whose owner owns the attached file still confers write. Applied to all
three object branches: knowledge base, file home collection, and workspace model.
Co-authored-by: rexpository <30176934+rexpository@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The direct image-edit route was the only image-edit surface with no authorization: it ran
on get_verified_user alone, while POST /generations enforces ENABLE_IMAGE_GENERATION +
features.image_generation and the built-in edit_image tool enforces ENABLE_IMAGE_EDIT +
features.image_generation. A verified non-admin user could therefore reach the configured
image-edit provider (spending IMAGES_EDIT_OPENAI_API_KEY) even when the administrator had
globally disabled image editing (ENABLE_IMAGE_EDIT=False) or denied the user image
generation.
Split the route from the shared impl (mirroring generate_images/image_generations): the new
/edit wrapper enforces ENABLE_IMAGE_EDIT and the per-user image-generation permission, then
delegates to image_edits(). The internal callers (the edit_image tool and the chat
middleware) already gate themselves and call image_edits() directly, so they are unaffected.
Co-authored-by: jagstack <52110932+jagstack@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
ws_terminal() interpolated the path parameter session_id directly into the upstream
terminal WebSocket URL and then appended ?user_id=<caller>, with no encoding or
validation (the HTTP sibling proxy_terminal runs _sanitize_proxy_path; this path ran
nothing). An encoded '?'/'&' smuggled through session_id survives Open WebUI's single
decode and is re-decoded by the upstream, injecting an attacker-chosen user_id ahead
of the appended one. Query parsing binds the first occurrence, so the orchestrator
resolves the spoofed user's terminal scope, letting a normal authenticated user
present another user's identity to the upstream (CWE-116/863).
Encode session_id as an opaque path segment with urllib.parse.quote(session_id,
safe=''). This neutralises '?'/'#'/'&' at any decode depth (the upstream's single
decode reverses only the quote, leaving the original delimiters inert as path
content), while legitimate UUID session ids pass through unchanged. The appended
user_id is then the only query parameter the upstream binds.
The separate concern that the forwarded identity is a bearer claim with no integrity
binding spans Open WebUI and the upstream terminal server and is not addressed here.
Co-authored-by: rexpository <30176934+rexpository@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
These were the only two ydoc Socket.IO handlers missing the SESSION_POOL guard that
every sibling (join/state/update) enforces. With always_connect=True admitting
unauthenticated sockets, a client that knew a note's document_id could broadcast
spoofed awareness (cursors/presence) and fake ydoc:user:left events into a live
editing room, attributed to any client-supplied user_id.
Both handlers now require an authenticated SESSION_POOL session; the awareness handler
additionally requires prior ydoc:document:join (room membership); and the broadcast
user_id is fixed to the authenticated identity instead of the client-supplied value,
removing the impersonation. Document content was never reachable (ydoc:document:update
already enforces write permission).
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The SCIM update_user (PUT) and patch_user (PATCH) handlers mapped the SCIM active field
unconditionally onto the role column (role = 'user' if active else 'pending'), so a routine
IdP sync or a misconfigured IdP that marked a locally-provisioned admin inactive would
silently strip that admin's role and could lock an instance out of its own administration.
Gate both active->role assignments on user.role != 'admin' so SCIM provisioning can
activate/deactivate ordinary users but never demotes an existing admin; admin role changes
continue to go through the dedicated admin endpoints. SCIM already cannot promote to admin
(active only maps to user/pending), so this is symmetric.
Credit to @HOHK0923 for surfacing the admin-demotion footgun.
Co-authored-by: HOHK0923 <118590749+HOHK0923@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
is_string_allowed does endswith() matching and was called with the full URL
(retrieval/web/utils.py) against WEB_FETCH_FILTER_LIST, so a blocklisted host with any
path (https://blocked.example/x) ended with /x, not the host, and slipped through; the
allowlist direction false-rejected legitimate URLs and admitted attacker URLs ending in
an allowed string. The same endswith caused label confusion at the hostname call site
(retrieval/web/main.py): corp.com matched evilcorp.com, 10.0.0.1 matched 110.0.0.1.
Add is_host_allowed(host, ...) matching on DNS label boundaries (host == pattern or
host.endswith('.' + pattern)), called with the parsed hostname at both web-fetch call
sites. is_string_allowed is left unchanged for the unrelated function-name filters
(utils/middleware.py, utils/tools.py).
The separate is_global guard (validate_url / _ssrf_safe_new_conn, active when
ENABLE_RAG_LOCAL_WEB_FETCH is off) already blocks RFC1918/loopback/link-local, so this
restores the admin's intended blocking of specific public hosts.
Co-authored-by: addcontent <59762500+addcontent@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The connection-layer DNS-rebinding guard (_SSRFSafeResolver / _SSRFSafeAdapter, PR #24759) was
mounted only on SafeWebBaseLoader. Two user-reachable image fetches validate the URL then fetch
it through the shared get_session() pool with the default resolver, so a TTL-0 rebinding answer
that passed validate_url reaches an internal address at connect:
- get_image_base64_from_url (utils/files.py): user image_url on every chat completion.
- load_url_image (routers/images.py, POST /api/v1/images/edit): user-supplied image field.
Add get_ssrf_safe_session() (a one-off aiohttp session mounting _SSRFSafeResolver) and use it
for both fetches, so the connect-time IP is re-validated and a rebound loopback / RFC1918 /
metadata address is rejected. The shared pool is left untouched for the admin-configured
image-generation callers, which legitimately reach internal hosts.
Co-authored-by: dhyabi2 <32069256+dhyabi2@users.noreply.github.com>
Co-authored-by: geo-chen <2404584+geo-chen@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
process_uploaded_file auto-links an uploaded file to the knowledge base named in
client-supplied metadata.knowledge_id, but it called Knowledges.add_file_to_knowledge_by_id
directly with no authorization, while the dedicated POST /knowledge/{id}/file/add endpoint
gates the same operation on owner / admin / write grant. A read-only collaborator (or any
verified user who knows the KB id) could therefore attach arbitrary files to a knowledge base
they cannot write: the KnowledgeFile membership row was committed, and the later vector-index
write check failed closed but left the row in place.
Gate the auto-link on the same owner/admin/write check before inserting the membership row;
otherwise skip the link and log. The file still uploads as the user's own file, it just is
not linked to a KB they cannot write.
Co-authored-by: jagstack <52110932+jagstack@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
generate_chat_completion() checks model access on line 200 only when bypass_filter is
False. When an arena model reaches this function without a pre-resolved selected_model_id,
which is the task and background path (the /api/v1/tasks/* endpoints call
generate_chat_completion directly rather than through process_chat_payload), the fallback
resolves the arena to an underlying model and recurses with bypass_filter=True, so the
resolved model's access check is skipped. An authenticated user with access to an arena
could therefore reach a model they are denied directly, and for the default or exclude
arena, whose candidate pool is every non-arena model, any model on the instance (CWE-862).
The normal chat path resolves the arena in process_chat_payload before this function, so its
resolved model is checked on line 200; the task path was not, which is the inconsistency.
Enforce check_model_access() on the resolved model in the fallback, before the
bypass_filter=True recursion, mirroring the normal-path check. Admins and already-bypassed
recursive calls are unaffected, and legitimate arena use of accessible models is unchanged.
Co-authored-by: rexpository <30176934+rexpository@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Two lifecycle/authorization gaps let a deactivated (pending) account keep acting:
1. The background automation scheduler (execute_automation) rehydrated the owner by ID and
dispatched the chat pipeline without re-checking the owner. A user later set to pending,
or one whose features.automations permission was revoked, kept running scheduled
automations on the operator's provider credentials, even though the HTTP create/update/run
routes already gate on get_verified_user + features.automations. Re-gate the rehydrated
owner before dispatch: require role user/admin and, for non-admins, the features.automations
permission; otherwise record an error and skip the run.
2. check_model_access enforced per-model ACLs only for exactly role == 'user', so any other
non-admin role (a pending principal) fell through and was granted access. Enforce for every
non-admin role (admins still bypass), so the check fails closed (CWE-862, CWE-863).
Co-authored-by: rexpository <30176934+rexpository@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The decode-until-stable loops in _sanitize_proxy_path (terminals.py, cap 8) and
_safe_static_redirect_path (models.py, cap 2) proceed with whatever remains after the
cap instead of rejecting it. A path encoded more times than the cap therefore exits the
loop still percent-encoded, passes the literal '..' / prefix checks (the dots are still
%2E, not '..'), and is forwarded to the upstream terminal server, or emitted as a
redirect Location, which then decodes it once more and resolves the traversal. This is
the residual of the decode-until-stable hardening added for CVE-2026-54017: the cap is a
fixed depth, not a true stability guarantee.
Reject when the value is still not stable after the cap (unquote(x) != x), so anything
encoded more deeply than the cap fails closed rather than being forwarded. Legitimate
paths stabilize within a pass or two and are unaffected.
Co-authored-by: DavidCarliez <271374756+DavidCarliez@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix: block private-IP webhook URLs to close SSRF on caller-controlled URL
post_webhook(url, ...) in utils/webhook.py forwards the URL straight to
aiohttp.ClientSession.post with no SSRF gate. The URL is caller-controlled
on two surfaces:
- User notification settings under ENABLE_USER_WEBHOOKS=true — any
authenticated user can set the URL their notifications POST to.
- Automation notification triggers (calendar alerts, etc.).
Without a gate, the URL can target cloud metadata (169.254.169.254 /
fd00:ec2::254), localhost-bound services, RFC1918 internal hosts, or any
other private address reachable from the server process. Blind SSRF — no
response body returned to the caller — but enough to enumerate internal
services via response timing / status codes, and on cloud deployments
enough to issue requests against IMDSv1 if available.
Call validate_url() at the top of post_webhook. The function blocks
private/reserved IPs when ENABLE_RAG_LOCAL_WEB_FETCH is False (the
default), is the project's chosen SSRF gate, and is already applied to
the equivalent fetch surfaces (retrieval, image-load, OAuth profile
picture). Operators who legitimately need to webhook to private IPs
(internal monitoring, self-hosted Slack alternatives, etc.) can set
ENABLE_RAG_LOCAL_WEB_FETCH=True — same opt-out as the other gated
surfaces.
Scope intentionally limited to webhooks. The OAuth discovery and
external reranker paths cwanglab also flagged are admin-configured with
intentional private-IP defaults (reranker defaults to
http://localhost:8080/v1/rerank) and are out of scope per Rule 9 — the
admin owns the URL choice and the operator opt-out exists for them too.
Reported by cwanglab in GHSA-5x9f-85cg-w3hf (cluster canonical with six
closed siblings: g36v-23gj-j69x, 6j8f-h58v-xgmw, xpwv-52pm-p8hj,
v9gp-hv2c-9qv8, fw7w-jrw7-p3v9, x7xq-74rg-m8mf).
Co-authored-by: cwanglab <cwanglab@users.noreply.github.com>
* fix: also pass allow_redirects=False on webhook post_webhook session.post
Companion to the previous commit. validate_url() only validates the
initial URL; aiohttp's default allow_redirects=True would still follow
a 302 to a private-IP target. Same redirect-bypass class as the rh5x
cluster's five call sites, sixth call site to receive the same gate.
Co-authored-by: cwanglab <cwanglab@users.noreply.github.com>
---------
Co-authored-by: cwanglab <cwanglab@users.noreply.github.com>
The Qdrant backends implemented delete(ids=...) as a payload filter on
metadata.id, but points are stored with the item id as the Qdrant point id
(see _create_points), and not every point carries an id in its payload.
Memory points store only {created_at} in metadata (KB metadata embeddings
likewise), so deleting a single memory matched nothing and left an orphaned
vector that kept being injected into RAG context.
Delete by point id instead: PointIdsList for the standard backend, and a
tenant-scoped HasIdCondition for multitenancy (point ids are unique, so tenant
isolation is preserved). Filter-based deletion is unchanged.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix: gate chat_completion channel: branch on channel access + message scoping
When chat_id starts with 'channel:' the chat-completion handler skips
the chat ownership / storage block below it. Nothing replaced that
gate. The downstream channel emitter in socket/main.py:_make_channel_
emitter writes to Messages.update_message_by_id using a caller-supplied
message_id pulled from form_data['id'], with no membership check, no
write-access check, and no validation that the message_id belongs to
the channel.
Net effect: any authenticated user could submit
chat_id='channel:<any-channel-uuid>' + id='<any-message-uuid>' and
overwrite that message with attacker-controlled LLM output. Cross-
channel writes worked too — private channels, DMs, channels the
caller has no access to. Original author attribution stayed intact on
the overwritten row.
Add the missing checks at the channel: branch:
1. Channel must exist (404 otherwise).
2. Non-admin caller must have write access to the channel — membership
for group/dm channels, AccessGrants permission='write' for others.
3. The message_id (if supplied) must belong to the same channel — a
caller with write access to channel A cannot use this path to
overwrite a message in channel B.
Behaviour change is limited to callers who were exploiting the gap:
legitimate flows that supply a message_id under their own channel
membership continue to work unchanged.
Co-authored-by: sfwani <sfwani@users.noreply.github.com>
* chore: trim verbose comment on channel: branch gate
---------
Co-authored-by: sfwani <sfwani@users.noreply.github.com>
The history diff, delete, and version-restore routes authorize the URL
prompt_id but then act on a caller-supplied history/version id without
checking it belongs to that prompt (IDOR). Filter by prompt_id in
compute_diff and delete_history_entry, and reject a cross-prompt version_id
in update_prompt_version.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Firecrawl /search returns either `{"data": [...]}` (flat list — v1, and
what frost19k reported on #23966) or `{"data": {"web": [...]}}` (v2,
current production). The parser only handled the dict shape:
data = response.get('data') or {}
results = data.get('web') or []
On a list-shape response, `data.get('web')` raised AttributeError,
caught by the function's outer try/except, and `search_firecrawl`
silently returned []. Web search worked against v2 endpoints but is one
upstream-format-change away from failing closed again. Accept either.
The native function-calling tool resolver in utils/tools.py applies five
gates before exposing execute_code as a builtin tool: builtin-category
enable, ENABLE_CODE_INTERPRETER global config, model capability,
features.code_interpreter request flag, and the per-user
features.code_interpreter permission.
The legacy XML-tag detection path in streaming_chat_response_handler
applied only the request-flag gate. Brings the legacy path to parity by
running the same five-gate check before activating tag detection.
Behaviour change is limited to deployments that previously relied on
the asymmetry — admins who set ENABLE_CODE_INTERPRETER=False or revoked
the per-user permission, on the legacy tool-calling mode, with the
client supplying features.code_interpreter=true. Any of those three
conditions met now correctly disables tag detection.
Co-authored-by: sfwani <sfwani@users.noreply.github.com>
Open WebUI's collection ACL accepted any unknown name as a
legacy/ephemeral collection. In Milvus multi-tenancy mode that name
becomes the `resource_id` and is interpolated unescaped into a SQL-like
Milvus expression — `resource_id == '<name>'` — so a name like
x' or resource_id != '' or resource_id == 'x
turns the filter into a tautology and returns every tenant's chunks
from the shared collection.
All collection names Open WebUI generates are UUIDs, SHA-256 hex
digests, or fixed-prefix variants of those — they all fit
[A-Za-z0-9_-]. Add a strict format check in
filter_accessible_collections (utils.py) that drops any name outside
that set before any ACL or vector-store lookup, applied even on the
admin bypass path. _validate_collection_access then surfaces the dropped
name as a 403.
As defense in depth, MilvusClient now validates resource_id at every
expression-construction site and escapes single quotes / backslashes in
any other string interpolated into a filter (delete ids, metadata
filter values). Non-string filter values are typed-checked instead of
str()-formatted.
Co-authored-by: Claude <noreply@anthropic.com>
update_chat_by_id re-derived assistant content from `output` on every
save (serialize_output) so frontend edits to output items reflect in
content. But it ran unconditionally, so content set independently of
output — an __event_emitter__ {"type":"replace"} from an Action, or an
outlet filter footer — was reverted to the original output-derived text
on the next save. The reload reads chat.chat directly, so the change
vanished after navigating away (regression vs 0.9.2, which predates the
output mechanism).
Re-derive only when the message's `output` actually changed versus what's
stored, which still reflects genuine output edits but leaves
independently-set content intact.
Fixes#24585
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Symptom
On a fresh install (zero users) the frontend shows the mandatory
"Create Admin Account" onboarding screen, but POST
/api/v1/auths/signup returns 403 ACCESS_PROHIBITED ("You do not have
permission to access this resource."). Wiping the database does not
help when the config layer is backed by Redis (the value survives in
the Redis/valkey volume), or when only the user table is cleared (the
config row survives in Postgres). The instance is then unrecoverable
through the UI.
Root cause
signup_handler() auto-sets request.app.state.config.ENABLE_SIGNUP =
False immediately after the first admin is created. That value is
persisted by the config layer (the Postgres config table, and Redis
when REDIS_URL is set). On a later zero-user database the persisted
False is read back, so ENABLE_SIGNUP resolves False even though no
users exist. The old gate was:
if WEBUI_AUTH:
if not ENABLE_SIGNUP or not ENABLE_LOGIN_FORM:
if has_users or not ENABLE_INITIAL_ADMIN_SIGNUP:
403
ENABLE_INITIAL_ADMIN_SIGNUP defaults to False, so with zero users the
inner test (has_users or not ENABLE_INITIAL_ADMIN_SIGNUP) is True, and
a stale ENABLE_SIGNUP=False trips the outer test, producing a 403 on
the only UI path that can create the first admin. The frontend decides
to show onboarding purely from user_count == 0, so frontend and
backend disagree and the instance bricks.
Change
Split the gate by has_users. Subsequent signups (has_users True) are
unchanged: still gated by ENABLE_SIGNUP and ENABLE_LOGIN_FORM. The
first user (has_users False, the bootstrap admin the onboarding screen
invites) is gated only by the admin-chosen ENABLE_LOGIN_FORM (the
documented SSO-only hard-disable) unless ENABLE_INITIAL_ADMIN_SIGNUP is
set. It is no longer gated by ENABLE_SIGNUP, which in the zero-user
state is never an admin decision but the post-first-admin auto-disable
leaking across a database reset.
Why this is safe (full case analysis)
For WEBUI_AUTH the gate has 16 input combinations over (has_users,
ENABLE_SIGNUP, ENABLE_LOGIN_FORM, ENABLE_INITIAL_ADMIN_SIGNUP). Old and
new are identical in 15 of them:
* All 8 has_users=True cases: both reduce to "403 iff not
ENABLE_SIGNUP or not ENABLE_LOGIN_FORM". Unchanged.
* 7 of the 8 has_users=False cases: identical.
The only changed case is has_users=False, ENABLE_SIGNUP=False,
ENABLE_LOGIN_FORM=True, ENABLE_INITIAL_ADMIN_SIGNUP=False: old
behaviour 403, new behaviour allow. The new condition is a strict
subset of the old (new-403 implies old-403), so the change never newly
blocks any request that previously succeeded; it only stops blocking
that one bootstrap state.
That state has no legitimate deployment. With the login form enabled
and zero users the onboarding form is already served, and the only
operator-configurable way to keep the first signup closed (SSO-only:
ENABLE_LOGIN_FORM=False, optionally with ENABLE_INITIAL_ADMIN_SIGNUP)
is preserved byte for byte. ENABLE_SIGNUP=False with zero users is not
an operator choice, it is the automatic post-first-admin disable, so
the old behaviour there was purely a brick with no recovery path. No
security control is weakened: ENABLE_LOGIN_FORM and
ENABLE_INITIAL_ADMIN_SIGNUP keep their exact meaning, and the
WEBUI_AUTH=False path is untouched.
This is not Redis-specific: it reproduces with Redis disabled through
the Postgres config table alone (clear the user table, keep the config
row).
Verification
Drove the real signup endpoint across a 10-case matrix on freshly
migrated databases, including the full end-to-end first-admin creation
(returns role=admin, row persisted as admin) and the preserved
SSO-only, subsequent-signup and no-auth behaviours. All pass.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The startup banner uses Unicode box-drawing characters. On a stdout that
can't encode them (Windows cp1252, or redirected/headless/pythonw output)
print() raises UnicodeEncodeError and aborts startup. This blocks running
open-webui serve headless on Windows.
Guard the banner print and fall back to a plain ASCII line so startup
always proceeds regardless of the console encoding.
Fixes#24965
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
serve_cache_file gated the resolved path with file_path.startswith(os.path.abspath(CACHE_DIR)) without a trailing os.sep, so any path resolving to a sibling whose name starts with the cache-dir basename (e.g. cache_backup, cached_models) passed the prefix check. Authenticated users could read files from such siblings via /cache/../<sibling>/<file>. Appending os.sep to the prefix closes the bypass; deep traversal and absolute paths were already correctly blocked.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The proxy gathered _client_to_upstream and _upstream_to_client with
return_exceptions=True. When upstream sends a graceful CLOSE,
_upstream_to_client returns but gather keeps waiting on
_client_to_upstream, which is blocked in ws.receive() until the browser
disconnects. The handler stays pending and the finally: session.close()
cleanup is deferred, leaking a ClientSession and an open browser socket.
Use asyncio.wait(return_when=FIRST_COMPLETED) and cancel the pending
sibling, so the proxy unwinds as soon as either direction finishes. The
pumps' bare except Exception already lets CancelledError (a BaseException)
propagate, so cancellation is clean and they need no change.
Fixes#25464
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat: cap profile image data URI size to bound model/avatar bloat
validate_profile_image_url() validated data-URI format (MIME allowlist,
SVG rejection, scheme checks) but never its length, so a valid
data:image/...;base64,<huge> passed for both custom-model icons and user
avatars. Large inline images bloat Postgres and the Redis MODELS hash and
degrade model-list latency.
Add PROFILE_IMAGE_MAX_DATA_URI_SIZE (default 256 KiB, 0 disables) and
reject oversized data URIs in the shared validator, so both model meta
(ModelMeta.profile_image_url) and user avatars (UpdateProfileForm) are
bounded at one chokepoint. ModelMeta already clears invalid values to
None on read, so existing oversized icons stop propagating into the
MODELS hash on the next refresh.
Fixes#25468
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix: default PROFILE_IMAGE_MAX_DATA_URI_SIZE to None (no cap)
Per review: opt-in rather than a 256 KiB default. Unset leaves data URIs
uncapped; the validator already skips the check on a falsy value.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
getChatEventEmitter starts a setInterval emitting a `usage` socket event
every second; clearInterval ran only after sendMessageSocket resolved, so
any throw/reject left the interval firing for the page lifetime. Each
failed send added another orphaned interval, inflating server-side usage
accounting and growing CPU/memory over a session.
Wrap the send in try/finally so the interval is always cleared, on both
the happy path and any thrown/rejected path. The exception still
propagates unchanged.
Fixes#25465
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
_sanitize_proxy_path decoded the proxy path once before the '..' check, so a double-encoded payload (%252e%252e) survived the check as %2e%2e and was then re-decoded into '..' by the upstream terminal server, defeating the traversal guard. Decode until stable so no encoded traversal sequence can reach the upstream. Single-encoded payloads were already rejected; this closes the double (and deeper) encoding bypass.
Co-authored-by: sermikr0 <230672901+sermikr0@users.noreply.github.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
bypass_system_prompt is an internal flag used by utils/middleware.py and utils/chat.py to skip applying the model system prompt on recursive base-model calls, but it was still declared as a positional argument on the openai/ollama chat-completion route handlers, so FastAPI bound it from the query string. Move it to request.state so external clients cannot set it, matching how bypass_filter is handled.
Drop the argument from both route signatures and read getattr(request.state, 'bypass_system_prompt', False); utils/chat.py sets request.state.bypass_system_prompt alongside bypass_filter and drops the kwarg from the two route-handler calls (the recursive self-calls keep it). Mirrors c0385f60b.
Co-authored-by: anishgirianish <161533316+anishgirianish@users.noreply.github.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The 't0p-s3cr3t' default was dead code on every supported startup path:
start.sh, start_windows.bat and `open-webui serve` all set or
auto-generate WEBUI_SECRET_KEY before the backend imports env.py. It was
only ever reachable by invoking uvicorn directly, which is unsupported
and unsafe (the app would then sign tokens/cookies with a public,
hardcoded key). It also keeps getting reported as a vulnerability because
it looks dangerous, even though it is unreachable in practice.
Drop the fallback (default to '') so an unset key is caught by the
existing WEBUI_AUTH guard, and replace the vague error with a clear,
actionable message explaining that the key is a hard requirement and how
the supported start methods provide it. Exit cleanly via SystemExit
instead of raising a ValueError traceback.
WEBUI_AUTH=False keeps working unchanged (key defaults to '').
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
renderMermaidDiagram returned raw mermaid SVG, which FilePreview.svelte injects
via wrapper.innerHTML = svg. Mermaid runs with securityLevel: 'loose', so it
neither sanitizes click hrefs (formatUrl skips sanitizeUrl) nor DOMPurifies its
output; a .md file with a click X href "javascript:..." directive (or an
HTML-label payload) therefore executes script in the app origin when previewed.
The chat path was already safe because SVGPanZoom DOMPurifies before rendering;
file preview was not.
Sanitize at the source: renderMermaidDiagram now returns DOMPurify-cleaned SVG
via a shared sanitizeSvg helper (same policy as SVGPanZoom), so every consumer
including the FilePreview innerHTML sink receives safe output.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
insert_chat_files() stored any caller-supplied file_id with no ownership
check, so a user could attach another user's file to their own chat and
then read it through the shared-chat access path in has_access_to_file().
Filter file_ids to those the caller owns, is admin for, or can read.
Also repairs an UnboundLocalError introduced in 260ead64d: the existing
duplicate-check referenced `session` before it was assigned (db=session),
so the function threw on every call and no chat_file rows were persisted.
ModelMeta.profile_image_url now runs validate_profile_image_url, rejecting SVG/script data URIs (matching UserUpdateForm and ChannelWebhookForm). The /model/profile/image endpoint enforces the PROFILE_IMAGE_ALLOWED_MIME_TYPES allowlist and sets X-Content-Type-Options: nosniff, so an SVG data URI can no longer be served inline on-origin. Closes the fourth profile-image XSS sink missed by the user and webhook fixes.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When called without attached model knowledge and given a caller-supplied knowledge_id, search_knowledge_files passed it straight to Knowledges.search_files_by_id, which does not enforce ownership on knowledge_id. An authenticated user who happened to know a target UUID could enumerate file metadata (filename, file id, KB id, KB name, updated_at) from any knowledge base, bypassing the AccessGrants permission model.
Mirror the same admin/owner/AccessGrants check the attached-KB branch already uses, matching the sibling query_knowledge_files function.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The lookup-prompt-by-command endpoint's only frontend wrapper,
getPromptByCommand, is dead: nothing imports or calls it, the path is
referenced nowhere else, and the route handler has no internal caller
(slash-command resolution happens client-side from the loaded prompt
list). Removes the route handler, its section header, and the dead
wrapper.
The shared Prompts.get_prompt_by_command data-layer method is kept: it
is still used by create/update prompt validation (prompts.py:175, 275,
340). PromptAccessResponse and AccessGrants are untouched.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
_process_picture_url validated the initial picture URL with validate_url()
but then aiohttp followed 3xx redirects without re-validating the target,
so a validate_url-passing public URL could 302 to an internal address and
the body was base64-stored in the user's profile_image_url. This is the
sixth call site of the CVE-2026-45401 redirect-bypass cohort; the other
five already pass allow_redirects=AIOHTTP_CLIENT_ALLOW_REDIRECTS. Apply
the same.
validate_url() resolves DNS to check IPs but discards the result; the
HTTP client resolves again independently. Between those two lookups an
attacker can swap the DNS record from a public IP to an internal one
(DNS rebinding).
Push the IP-is-global check into the actual connection layer so the
validated resolution is the one used for the TCP connect:
- aiohttp (_fetch): _SSRFSafeResolver wraps DefaultResolver and rejects
non-global IPs at resolve time (zero TOCTOU window).
- requests (_scrape): _SSRFSafeAdapter mounts custom urllib3 connection
classes whose _new_conn resolves, validates, and connects to the
validated IP in one shot (zero TOCTOU window).
Both paths respect ENABLE_RAG_LOCAL_WEB_FETCH (skip validation when on).
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
has_access_to_file granted access whenever the file was attached to a
shared chat the user could read, ignoring the requested access_type. A
read-only shared-chat recipient therefore satisfied write and delete
checks and could delete or mutate the chat owner's attached file. Gate
the shared-chat branch on read access, matching the channels branch
directly above it.
Co-authored-by: oxsignal <oxsignal@users.noreply.github.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
SafePlaywrightURLLoader validated only the initially submitted URL and
then let the browser follow HTTP redirects and client-side navigations
without re-checking them, so a public URL could redirect into the
internal network (cloud metadata, RFC1918, loopback). Intercept
document-type requests, re-run validate_url on each, and apply the same
redirect policy as the requests loader (blocked unless
AIOHTTP_CLIENT_ALLOW_REDIRECTS). Sub-resource requests pass through
unchanged so page rendering performance is unaffected.
Co-authored-by: POV9en <POV9en@users.noreply.github.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
generateFollowUps in src/lib/apis/index.ts is dead: it appears only at
its own definition, nothing imports or calls it, and it targets a
non-existent path (/api/v1/tasks/follow_ups/completions, plural) while
the real route is /tasks/follow_up/completions (singular). Follow-up
suggestions are generated server-side in the chat-completion middleware
and delivered over the chat:message:follow_ups websocket event, so this
wrapper was never on the live path.
Removes only the dead wrapper. The backend POST /tasks/follow_up/completions
endpoint is intentionally kept: it is a member of the actively-used
/tasks/*/completions family (title, tags, emoji, queries, moa) and its
handler delegates to the core generate_follow_ups function.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
update_event only verified write access on the event's source calendar.
CalendarEventUpdateForm accepts a new calendar_id which the model layer
applies unconditionally, so a user with write access to their own calendar
could move (inject) an event into any other user's calendar. Mirror the
destination check create_event already performs.
The __main__ block called search_bing() with 4 positional arguments, but
the function requires 5 (subscription_key, endpoint, locale, query,
count). Running `python -m open_webui.retrieval.web.bing` raised a
TypeError and, before failing, silently misrouted every argument. Read
the key/endpoint from environment variables, matching config.py defaults.
Closes#24765
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The bulk-clear-chat-tags endpoint's only frontend wrapper,
deleteTagsById in src/lib/apis/chats/index.ts, is dead: nothing imports
or calls it, the path is referenced nowhere else, and the route handler
has no internal caller. Removes the route handler, the dead wrapper, and
the now-orphaned Chats.delete_all_tags_by_id_and_user_id model method
(its sole caller was this route). The shared
Chats.delete_orphan_tags_for_user method is untouched.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
POST /utils/markdown rendered a markdown string to HTML server-side. Its
only frontend wrapper, getHTMLFromMarkdown in src/lib/apis/utils/index.ts,
is dead: nothing imports or calls it, the route is hit by no other code
path, and the path string appears nowhere else in the repo (no direct
fetch, no test, no docs). Markdown is rendered client-side in the UI, so
this endpoint was redundant.
Fully self-contained removal: the endpoint, its MarkdownForm model, the
now-orphaned 'import markdown' in the utils router (used only here), and
the dead getHTMLFromMarkdown wrapper. Nothing else depends on any of them.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
GET /evaluations/feedbacks/all returned the entire feedback table in a
single response (flagged as a Medium OOM risk for admins in
open-webui#22206). Its only frontend wrapper, getAllFeedbacks in
src/lib/apis/evaluations/index.ts, is dead: nothing imports or calls it
anywhere in the codebase. The endpoint is a redundant view-only twin of
GET /evaluations/feedbacks/all/export, which is what the admin Feedbacks
UI actually uses.
Removes the endpoint, the now-unused FeedbackResponse import in the
evaluations router, and the dead getAllFeedbacks frontend wrapper. The
shared Feedbacks.get_all_feedbacks data-layer method is kept, since the
live /feedbacks/all/export endpoint still uses it.
Ref: open-webui#22206
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
These 19 exported wrappers are dead: each appears exactly once in the
codebase (its own definition), nothing imports or calls any of them, and
none has a corresponding backend route. They are leftovers from settings
that were consolidated server-side into /auths/admin/config,
/openai/config, /ollama/config and /api/config:
- index.ts: getModelFilterConfig, updateModelFilterConfig,
getCommunitySharingEnabledStatus, toggleCommunitySharingEnabledStatus,
getModelConfig, updateModelConfig (+ orphaned GlobalModelConfig type)
- auths: getSignUpEnabledStatus, toggleSignUpEnabledStatus,
getDefaultUserRole, updateDefaultUserRole, getJWTExpiresDuration,
updateJWTExpiresDuration
- openai: getOpenAIUrls, updateOpenAIUrls, getOpenAIKeys, updateOpenAIKeys
- ollama: updateOllamaUrls
- prompts: restorePromptFromHistory
- folders: updateFolderItemsById (+ orphaned FolderItems type)
Shared types (ModelConfig/ModelMeta/ModelParams) and all live wrappers
are untouched. Removal is import-safe: nothing referenced these.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Reported by bwgabrielsusai on #24719: granting a user only
`workspace.skills` doesn't show the Workspace menu, and visiting
`/workspace` directly bounces them to `/`.
The per-route guard in `/workspace/+layout.svelte` already covered
skills, but two earlier gates in the chain didn't:
* `Sidebar.svelte` case 'workspace' OR'd models/knowledge/prompts/tools
to decide menu visibility — skills was missing, so the entry never
rendered for skills-only users.
* `/workspace/+page.svelte` redirect chain picked the first available
section — skills was missing, so the fallback `goto('/')` fired.
Adding skills to both.
start.sh runs with `set -euo pipefail`, but three call sites added in
070ab2650 (refac: reorganize scripts and ci workflows) reference
optional env vars via bash's `,,` lowercase expansion without any
default. Containers that don't set these vars — the default for every
deployment that isn't explicitly opting into Playwright / bundled
Ollama / CUDA — crash on startup with:
start.sh: line 15: WEB_LOADER_ENGINE: unbound variable
(and the same for USE_OLLAMA_DOCKER, USE_CUDA_DOCKER once the first
were set in turn.) Reported in open-webui#24560 by urbenlegend.
The same refactor correctly defaulted every other optional env var
with `${VAR:-…}`. The three `,,` references slipped through because
bash can't combine `:-default` with `,,` in a single substitution —
`${VAR:-default,,}` makes the default literal `,,`, not what's wanted.
Fix: normalise the three vars in a one-line preamble with `${VAR:=}`,
which assigns an empty default if unset. The downstream `${VAR,,}`
expressions stay exactly as Tim wrote them, preserving the file's
visual style and matching the existing `${VAR:-…}` idiom for "this
variable is optional".
* fix: enforce features.direct_tool_servers on chat-completion tool_servers
The features.direct_tool_servers per-user permission was correctly
enforced on the storage path (routers/users.py user/settings/update,
which strips toolServers from saved settings when the caller lacks the
permission), but the inference path (/api/chat/completions) popped
tool_servers straight from the request body into metadata with no
permission check. The middleware (utils/middleware.py:2799) then
consumed direct_tool_servers to inject system_prompt into the message
array and register external tool specs that get invoked during the
completion. End result: any authenticated user could bypass the
admin-set per-user feature toggle and use inline tool_servers in their
chat-completion requests, even when admin had explicitly denied the
permission.
Default for USER_PERMISSIONS_FEATURES_DIRECT_TOOL_SERVERS is False
(config.py:2750), so under default config no regular user is supposed
to be able to use direct tool servers — making this a real boundary
bypass on out-of-the-box deployments rather than a corner case.
Mirror the storage-side behaviour at the inference entry point: pop
tool_servers from the request body, then silently drop the value if
the caller is non-admin and lacks features.direct_tool_servers. Admins
always pass; users with the explicit grant always pass; everyone else
gets None propagated into metadata, which the middleware already
handles as the no-tool-servers case.
Reported by berkant-koc in GHSA-f582-c373-jjf6.
Co-authored-by: berkant-koc <berkant-koc@users.noreply.github.com>
* chore: trim verbose comment on tool_servers permission check
---------
Co-authored-by: berkant-koc <berkant-koc@users.noreply.github.com>
POST /api/v1/chats/new and POST /api/v1/chats/{id}/folder accepted a
caller-supplied folder_id with no validation — neither ownership, nor
existence, nor UUID format. The row was persisted with the supplied
value verbatim, so the DB ended up with chat rows whose folder_id
referenced another user's folder, a non-existent UUID, or even a
non-UUID string.
No read path surfaces this across users — every chat-folder read is
user_id-filtered on both sides — so this is referential-integrity
hardening rather than a security boundary fix. But there's no reason
to accept dangling references either, and the downstream consumers
shouldn't have to assume the column is clean.
Add a Folders.get_folder_by_id_and_user_id() lookup at both writers:
if a folder_id is supplied, it must match a folder owned by the
caller. None remains allowed (chat-without-folder is the default).
Non-existent and non-UUID values fall through to 404.
Reported by ShigekiTsuchiyama in GHSA-4vrg-2vcq-q7jc.
Co-authored-by: ShigekiTsuchiyama <ShigekiTsuchiyama@users.noreply.github.com>
* fix: prevent mass-assignment user_id spoofing in POST /api/v1/evaluations/feedback
Two independent gaps in backend/open_webui/models/feedbacks.py let an
authenticated caller forge the `user_id` (and `id`, `version`) on a new
feedback record submitted to POST /api/v1/evaluations/feedback:
1. `FeedbackForm` declared `model_config = ConfigDict(extra='allow')`,
so Pydantic preserved any extra fields supplied in the request body —
including `user_id`, `id`, `version`. The form is the public input
boundary for the endpoint and should not accept unknown fields.
2. In `insert_new_feedback`, the dict literal placed
`**form_data.model_dump()` AFTER `'id': id`, `'user_id': user_id`,
`'version': 0`. Python dict-literal duplicate-key resolution is
last-wins, so any of those fields present in `form_data` overwrote
the server-derived values.
Combined effect: a regular user could POST a feedback record with an
arbitrary `user_id`, attributing the rating to any other user. The Elo
leaderboard at backend/open_webui/routers/evaluations.py computes model
rankings from these records, and the admin export
(GET /api/v1/evaluations/feedbacks/export) and admin list
(GET /api/v1/evaluations/feedbacks/all) display the spoofed attribution.
Two fixes, defense-in-depth:
- FeedbackForm: switch `extra='allow'` to `extra='ignore'` so Pydantic
drops unknown fields at parse time. Sub-models (RatingData / MetaData /
SnapshotData) intentionally keep `extra='allow'` because their contents
are deliberately schema-flexible — the spoofing surface was the form,
not the sub-payloads.
- insert_new_feedback: spread `form_data.model_dump()` first, then
overlay server-controlled fields (`id`, `user_id`, `version`,
`created_at`, `updated_at`) so the explicit keys win on duplicate-key
resolution regardless of what reaches the function. Matches the secure
pattern already used in backend/open_webui/models/functions.py:120.
Reported by yantongggg in GHSA-rjmp-vjf2-qf4g. Same root-cause class as
the prior published GHSA-hr43-rjmr-7wmm (folder mass-assignment, fixed
in v0.9.0); that fix did not generalize across the codebase, this fix
closes the feedback variant.
Co-authored-by: yantongggg <yantongggg@users.noreply.github.com>
* chore: trim comments
---------
Co-authored-by: yantongggg <yantongggg@users.noreply.github.com>
* fix: prevent redirect-based SSRF in get_image_base64_from_url
Cohort follow-up to PR #24491. That PR patched three call sites
(SafeWebBaseLoader._scrape, get_content_from_url, load_url_image) to
pass allow_redirects=False on the underlying HTTP client; this fourth
call site in utils/files.py was missed.
get_image_base64_from_url() is invoked from convert_url_images_to_base64
in utils/middleware.py on every /api/chat/completions request whose
message content includes an image_url part. validate_url() is called on
the originally-submitted URL only; the aiohttp session.get() call had
no allow_redirects argument and the shared session pool does not
override the aiohttp default (allow_redirects=True). An authenticated
user sending a chat message with image_url pointing at an attacker host
that 302-redirects to 169.254.169.254 / 127.0.0.1 / RFC1918 reached the
internal target. This is the most reachable variant in the redirect
cluster: no special endpoint, no admin permission, no feature flag.
Apply the same one-line fix as the other three call sites: pass
allow_redirects=AIOHTTP_CLIENT_ALLOW_REDIRECTS (defaults to False).
Reported by nayakchinmohan in GHSA-88jq-grjp-jx6f; consolidated under
GHSA-rh5x-h6pp-cjj6.
Co-authored-by: nayakchinmohan <nayakchinmohan@users.noreply.github.com>
* fix: enforce collection write access on process_file endpoint
Cohort follow-up to ba83613ff. That commit added _validate_collection_access
to process_text and process_web (the user-supplied collection_name path)
but missed process_file in the same router.
process_file accepts a user-supplied collection_name and writes the file's
embedded content into that collection via save_docs_to_vector_db. The
file_id is gated by file ownership (line 1562) but collection_name was
unchecked, so an authenticated user could append content from a file they
own into another user's knowledge-base collection by passing the victim's
KB UUID as collection_name. Identical pattern to the process_text and
process_web gaps that ba83613ff closed.
Apply the same one-line gate as the sibling endpoints: when
collection_name is user-supplied (not the default file-{file.id} fallback),
require write access via _validate_collection_access. The shared validator
delegates to filter_accessible_collections, which already correctly
handles file-* prefixes (via has_access_to_file) and KB UUIDs
(via Knowledges.check_access_by_user_id) — admins bypass.
Reported by tenbbughunters (Tenable) in GHSA-4g37-7p2c-38r9 (the
comprehensive write-path filing covering process_text / process_file /
process_web / process_youtube and the _validate_collection_access UUID
root cause), and independently re-identified for the missed process_file
call site by kodareef5 in GHSA-4m74-3cmc-293g.
Co-authored-by: tenbbughunters <tenbbughunters@users.noreply.github.com>
Co-authored-by: kodareef5 <kodareef5@users.noreply.github.com>
* fix: enforce collection write access on process_files_batch endpoint
Cohort follow-up to ba83613ff and the prior process_file fix on this
branch. process_files_batch (line 2604) is the third write endpoint in
the same router that accepts a user-supplied collection_name; it was
covered in the same Tenable filing as process_file and was missed by
the same cohort fix. The endpoint validates per-file ownership at line
2642 but does not check whether the caller has write access to the
target collection_name before save_docs_to_vector_db writes into it
at line 2683-2690 with add=True.
Apply the same one-line gate as the sibling endpoints. Validate only
when collection_name is user-supplied (truthy) so the existing fall
through behavior for the None case is unchanged.
Same Tenable / kodareef5 cohort as the previous commit.
Co-authored-by: tenbbughunters <tenbbughunters@users.noreply.github.com>
Co-authored-by: kodareef5 <kodareef5@users.noreply.github.com>
---------
Co-authored-by: nayakchinmohan <nayakchinmohan@users.noreply.github.com>
Co-authored-by: tenbbughunters <tenbbughunters@users.noreply.github.com>
Co-authored-by: kodareef5 <kodareef5@users.noreply.github.com>
* fix: gate tool content updates behind workspace.tools to match create endpoint
`update_tools_by_id` (routers/tools.py:452) authorizes a caller as long as
they are the tool's owner, hold a `write` access grant on the tool, or are
an admin. This means a verified user who has been given a write grant on
a tool — typically as part of a metadata-collaboration workflow (edit
description, adjust valves, manage access grants) — can also overwrite
the tool's Python source. Because `load_tool_module_by_id` further down
calls `exec(content, module.__dict__)` at module-import time, anything
the new content puts outside the `class Tools:` body executes immediately
on the server with the worker's privileges (root in the default Docker
deployment).
The `create_new_tools` endpoint already requires
`workspace.tools` (or `workspace.tools_import`) precisely because creating
a tool means submitting executable code. The update endpoint did not
mirror that check, producing an asymmetric authorization surface in which
a write-grantee with no workspace permission can still reach the same
exec sink as a workspace.tools-trusted creator. SECURITY.md frames
`workspace.tools` as the trust signal an admin uses to delegate
code-execution capability; the previous behavior let that signal be
bypassed by a per-resource share.
Fix: after the existing ownership / write-grant / admin gate, add a
content-change check. If `form_data.content != tools.content`, require
`workspace.tools` or `workspace.tools_import` (or admin role). Metadata
edits — `name`, `description`, valves config, access grants — continue
to flow through the existing gate, so the legitimate share-for-
collaboration workflow is unaffected.
Reported by KadirArslan in GHSA-p4fx-23fq-jfg6 with a working three-user
PoC (Alice trusted with workspace.tools creates a tool and shares write
to Bob; Bob updates content and the new code runs as root inside the
container, with Burp Collaborator confirming outbound exfiltration).
Co-authored-by: KadirArslan <KadirArslan@users.noreply.github.com>
* chore: trim comment
---------
Co-authored-by: KadirArslan <KadirArslan@users.noreply.github.com>
* fix: enforce message ownership in group/DM channel update + delete endpoints
`update_message_by_id` (channels.py:1348) and `delete_message_by_id`
(channels.py:1550) branch on `channel.type`. The `else` branch (standard
channels) correctly enforces `message.user_id != user.id` ownership before
mutating, but the `if channel.type in ['group', 'dm']` branch only checked
`is_user_channel_member` — channel membership alone, with no message
ownership verification.
Effect on group/DM channels: any verified member of the conversation could:
- overwrite another member's message content while the server preserved
`user_id=victim`, producing tampered content that renders to other
members as the original author's authentic post (integrity + authenticity);
- silently delete another member's messages, removing them from
conversation history without trace (integrity).
Reproduced end-to-end against v0.9.4 with three users (attacker, victim,
viewer) sharing a group channel: attacker overwrites victim's message and
deletes another, viewer reads the tampered content as victim-authored.
Two patches, identical shape, mirror the `else` branch's existing
ownership semantics:
- `update_message_by_id` group/DM branch: add
`if user.role != 'admin' and message.user_id != user.id: raise 403`
immediately after the `is_user_channel_member` check.
- `delete_message_by_id` group/DM branch: same.
The standard-channel branch is unchanged (it already enforced ownership).
Admins remain able to moderate any message, matching the existing semantic
in the standard-channel branch.
Reports consolidated under GHSA-wwhq-cx22-f7vv (earliest live filing of the
group/DM-specific variant). Same gap previously surfaced and partially
fixed under GHSA-jxwr-g6r6-j3fx (which addressed the standard-channel
branch only) — this completes the cohort.
* chore: trim comments
`pin_channel_message` (channels.py:1242) checked `permission='read'` on
the standard-channel branch before mutating `is_pinned` / `pinned_by` /
`pinned_at` via `Messages.update_is_pinned_by_id`. Pin/unpin is a write
operation; gating it on read access let any user with read-only channel
access pin or unpin any message in the channel, including admin posts.
One-character fix: change `permission='read'` to `permission='write'`.
Reported by kikayli in GHSA-5gc6-xhv4-2wg6.
Co-authored-by: kikayli <kikayli@users.noreply.github.com>
* chore: add validate_url() to get_image_data() for cohort consistency hardening
`get_image_data()` in `backend/open_webui/routers/images.py` fetches the
URL returned by the configured image generation API directly via
`session.get(data)` without first calling `validate_url()`. The sibling
`load_url_image()` in the same file (called from /images/edit) calls
`validate_url(data)` first — that gate was added under
GHSA-jgx9-jr5x-mvpv. The two functions handle structurally identical
input (an attacker-or-server-supplied URL string) and should enforce the
same SSRF gate as a matter of code hygiene.
In the current call graph, the URL passed to `get_image_data()` comes
from the admin-configured image generation API's response, so an
exploitable SSRF chain additionally requires admin-side trust delegation
(misconfigured/untrusted upstream image API, or a custom
OpenAI-compatible server that reflects user input into response URLs).
That makes the missing call a defense-in-depth gap rather than a
vulnerability per SECURITY.md Rule 9 — a position the GHSA-h7cc-wwjp-5xqh
advisory is being closed under. This change is hardening: it brings the
two image-fetch helpers into alignment so any future caller that begins
passing user-influenced URLs into `get_image_data()` is gated by the
same private-IP / loopback / metadata-IP filter the rest of the
codebase enforces.
Surface raised by brodmart in GHSA-h7cc-wwjp-5xqh.
Co-authored-by: brodmart <brodmart@users.noreply.github.com>
* chore: trim comment
---------
Co-authored-by: brodmart <brodmart@users.noreply.github.com>
GET /api/v1/models/model?id=<model_id> at routers/models.py:412
returned the full model.model_dump() to any caller with read access,
including the params dict that holds the admin-curated system prompt
and other behavior config. The user-facing /api/models endpoint
already strips this via utils/models.py:170,210 with the comment
"Remove params to avoid exposing sensitive info", and /api/v1/models/list
gates by write permission so non-curators don't see the model in their
workspace listing at all. The per-id endpoint missed the same gate, so
a user with read-only access (e.g. granted access to use the model in
chat) could open /workspace/models/edit?id=<not-mine> in the browser
and read the system prompt verbatim from the network response, even
though saving was correctly blocked.
Compute write_access once at the top of the handler so it can serve
both the response-shape decision and the response field. When the
caller lacks write access, replace params with an empty dict in the
serialised response. Owners, admins under BYPASS_ADMIN_ACCESS_CONTROL,
and explicit write-grant holders still get the full payload so the
workspace edit UI keeps working for users who legitimately curate the
model.
Read-permission users continue to receive everything else they need to
chat with the model — the chat path resolves prompt/params server-side
from the stored ModelModel and never echoes them back through this
endpoint.
Reported by destination-one in GHSA-h2cw-7qw9-56xr.
Co-authored-by: destination-one <destination-one@users.noreply.github.com>
urllib.parse.urlparse and requests/aiohttp disagree on how to split URLs
containing backslash, tab, CR, or LF in or around the netloc. urlparse
treats backslash as part of userinfo and uses what follows '@' as the
host; requests treats backslash as the start of the path and connects
to whatever precedes it. The same URL therefore passes the private-IP
filter (urlparse sees a public host) but reaches an internal target
(requests connects to e.g. 127.0.0.1). End result is an SSRF that the
existing IP block list cannot catch because it's evaluating the wrong
host.
PoC: http://127.0.0.1:6666\@1.1.1.1 — urlparse hostname is 1.1.1.1
(global, passes), requests reaches 127.0.0.1 (loopback).
Reject up front any URL containing one of the four documented parser-
confusing characters before either parser gets a chance to interpret
it. None of these characters is valid in an unencoded URL (\ should
always be %5C, whitespace should be %09 / %0A / %0D), so this is a
pure defensive rejection with no legitimate-input false positives.
Reported by Fushuling and RacerZ-fighting in GHSA-8w7q-q5jp-jvgx.
Co-authored-by: Fushuling <Fushuling@users.noreply.github.com>
Co-authored-by: RacerZ-fighting <RacerZ-fighting@users.noreply.github.com>
The `get_status()` handler at retrieval.py:263 (`@router.get('/')`) returned
the live RAG pipeline configuration (CHUNK_SIZE, CHUNK_OVERLAP, RAG_TEMPLATE,
RAG_EMBEDDING_ENGINE, RAG_EMBEDDING_MODEL, RAG_RERANKING_MODEL, etc.) without
any authentication dependency, while every adjacent endpoint on the same
router (/embedding, /embedding/update, /config, /config/update) requires
get_admin_user.
Exhaustive search of the repository confirms the endpoint has no callers:
- Frontend (src/): no `RETRIEVAL_API_BASE_URL}/'`-style fetch; the existing
`getRAGConfig()` in src/lib/apis/retrieval/index.ts targets `/config`,
not the root, and is the only consumer of admin-level retrieval state.
- Backend self-references: none.
- Cypress e2e (chat, documents, registration, settings): none.
- Backend tests (backend/open_webui/test/): none.
- Build/CI scripts (scripts/): none.
- Direct symbol import of `get_status` from this router: none.
The endpoint is dead code, almost certainly a relic from before the
/config GET split. Removing it has zero UX impact and eliminates the
unauthenticated-config-disclosure surface raised in advisory triage on
GHSA-65pg-qhhw-mxwg. External monitoring scripts that may have hit the
bare root will receive a 404 and can switch to the existing /config
endpoint, which returns the same fields plus the rest of the RAG config
under admin auth.
Surface raised by 0xRyuzak1 in GHSA-65pg-qhhw-mxwg. The advisory was closed
as not-a-vulnerability per SECURITY.md Rule 1 (no security boundary
crossed in default config — RAG_TEMPLATE default is a citation-format
instruction, not a system prompt; no integrity/availability impact); this
removal is independent code-hygiene that aligns the router cohort.
Reported-by: 0xRyuzak1 <https://github.com/0xRyuzak1>
The /create (L155-193) and /id/{id}/update (L248-297) endpoints in
routers/skills.py persisted form_data.access_grants directly to
AccessGrants.set_access_grants without filter_allowed_access_grants,
while every other shareable resource in the codebase (channels, knowledge,
models, notes, prompts, tools, calendars) and the dedicated
/id/{id}/access/update endpoint on this same router (L309-348) all do call
the filter. A user with workspace.skills permission (default False, but
admins can grant it to skill-creating users) could therefore attach
{"principal_type":"user","principal_id":"*","permission":"read"|"write"}
to the create or update payload and have it persisted unfiltered, bypassing
the sharing.public_skills gate that the rest of the cohort enforces.
Two changes:
- create_new_skill: call filter_allowed_access_grants with
'sharing.public_skills' immediately before insert, after the existing
permission check and ID-taken check.
- update_skill_by_id: call filter_allowed_access_grants with the same key
after the access check, before form_data.model_dump() flows into
Skills.update_skill_by_id. The pre-existing access check at L263-277 only
restricts WHO may modify the skill; the new filter restricts WHICH grants
they may set.
All supporting plumbing was already in place from prior PRs:
filter_allowed_access_grants is already imported at L22, the
USER_PERMISSIONS_WORKSPACE_SKILLS_ALLOW_PUBLIC_SHARING constant exists,
DEFAULT_USER_PERMISSIONS['sharing']['public_skills'] is wired up,
SharingPermissions.public_skills is in the Pydantic, and the admin UI
already renders the toggle. This is a pure 2-line router fix that closes
the cohort-consistency gap.
Same shape as the calendar fix in #24493, reported by Matteo Panzeri while
auditing the resource-cohort cohort during follow-up on #24493.
Co-authored-by: Matteo Panzeri <28739806+matte1782@users.noreply.github.com>
* fix: gate public sharing of calendars behind sharing.public_calendars permission
The calendar router did not call filter_allowed_access_grants on either the
create or update endpoint, while every other shareable resource in the
codebase (channels, knowledge, models, notes, prompts, skills, tools) does.
A verified non-admin owner could therefore attach
`{"principal_type":"user","principal_id":"*","permission":"read"|"write"}`
to their own calendar in the create or update payload and have it persisted
unfiltered. Any other verified user with the (default-on) features.calendar
permission could then read or, for write grants, write events on it via the
existing /events* endpoints, bypassing the per-user sharing.public_<X>
permission gate the rest of the resource cohort enforces.
Three changes:
- config.py: add USER_PERMISSIONS_CALENDAR_ALLOW_PUBLIC_SHARING (default
False, env-overridable) and surface it in DEFAULT_USER_PERMISSIONS
['sharing']['public_calendars'] so admins can grant it per group via the
same UI used for notes/models/etc.
- routers/calendar.py: import filter_allowed_access_grants and call it in
create_calendar with the new sharing.public_calendars key, identical to
the channel router's pattern.
- routers/calendar.py: call filter_allowed_access_grants in update_calendar
too. The pre-existing owner-only gate at L350 only restricts WHO may
change grants; the new filter restricts WHICH grants they may set, so a
non-admin owner cannot make their own calendar publicly readable or
writable without the corresponding sharing permission.
Same shape as GHSA-7rjh-px4v-5w55 (channels). Reported by Matteo Panzeri.
Co-authored-by: Matteo Panzeri <28739806+matte1782@users.noreply.github.com>
* fix: expose public_calendars + features.calendar through admin permissions surface
The earlier commit added DEFAULT_USER_PERMISSIONS['sharing']['public_calendars']
and the runtime filter call, but the new key was not yet plumbed through the
admin /users/default/permissions endpoint. Without these changes the toggle
would round-trip as silently dropped:
- routers/users.py SharingPermissions: any payload POSTed to
/default/permissions ran through `form_data.model_dump()`, and Pydantic
drops fields not declared on the model. The new public_calendars key
would have been stripped on every save, leaving admins unable to grant
the permission via the UI even though the runtime filter would honor it.
- src/lib/constants/permissions.ts: the frontend's DEFAULT_PERMISSIONS dict
is the seed shape used by the admin Groups Permissions panel; without
the new key it could not bind a Switch component to it.
- Permissions.svelte: add a Calendars Public Sharing toggle alongside the
Notes/Chats Public Sharing toggles, gated on the existing
features.calendar flag (matches the pattern used for notes/chats).
Also closes a pre-existing parity gap on features.calendar: DEFAULT_USER_
PERMISSIONS['features']['calendar'] has existed since the calendar feature
shipped, and Permissions.svelte already renders a Calendar feature toggle,
but FeaturesPermissions Pydantic and the frontend defaults never knew
about it. Adding it everywhere completes the round-trip so admin saves no
longer silently drop the calendar feature flag either.
---------
Co-authored-by: Matteo Panzeri <28739806+matte1782@users.noreply.github.com>
validate_url() in retrieval/web/utils.py only validates the initial URL.
The HTTP clients used downstream (sync requests, sync requests via the
parent WebBaseLoader._scrape, aiohttp via load_url_image) followed 3xx
redirects by default and did not re-validate the redirect target against
the private-IP / metadata-IP block list. An authenticated user could
submit a public URL that 302-redirected to an internal address (RFC1918,
127.0.0.1, 169.254.169.254, etc.) and the redirected response was returned
to them, enabling SSRF reads of internal services and cloud metadata.
Three call sites needed allow_redirects=False to match the policy already
enforced on the async _fetch() path:
- SafeWebBaseLoader: override requests_kwargs in __init__ so that the
inherited synchronous _scrape() path passes allow_redirects=False to
self.session.get() (the parent WebBaseLoader uses requests' default
allow_redirects=True).
- get_content_from_url (retrieval/utils.py): pass allow_redirects=False
on the streamed requests.get(...) call.
- load_url_image (routers/images.py, image-edits endpoint): pass
allow_redirects=False on the aiohttp session.get(...) call.
Reports consolidated under GHSA-rh5x-h6pp-cjj6:
- GHSA-rh5x-h6pp-cjj6 (tenbbughunters / Tenable) - sync _scrape
- GHSA-5vxg-6gmv-m2qr (YLChen-007) - load_url_image
- GHSA-hf76-c83f-63w2 (tempcollab) - aiohttp _fetch (already fixed)
- GHSA-h55f-h5fh-mvm4 (sneaXOR) - get_content_from_url
* refac(routers): reject external URLs in profile/model image handlers
* refac(ui): centralize image URL validation in safeImageUrl helper
* refac(auths): make signout POST-only
* refac: gate external profile image redirect behind ENABLE_PROFILE_IMAGE_URL_FORWARDING
Restore the 302 redirect for external http(s) profile image URLs in
the user and model profile-image endpoints, but gate it behind a new
ENABLE_PROFILE_IMAGE_URL_FORWARDING env flag (default: True).
Existing deployments that rely on external profile image forwarding
continue to work unchanged. Operators who want to suppress the
redirect (to prevent client-side IP/UA/Referer leaks) can set the
flag to False.
Non-admin GET /api/v1/prompts/tags went through get_prompts_by_user_id,
which loaded every active prompt with its full content/data/meta plus
owner records and all access grants, then ran one has_access query per
prompt that wasn't owned by the caller - all so the endpoint could
collapse the result to a sorted tag list. With 600 prompts this took
several seconds while the admin path (a single SELECT) returned in <1s.
Add Prompts.get_tags_by_user_id which selects only the tags column and
applies the same EXISTS-based access filter used by /list. Also tighten
the admin get_tags to project just the tags column instead of full rows.
The endpoint is now one DB query (plus one for groups), no row hydration,
no N+1.
Co-authored-by: Claude <noreply@anthropic.com>
get_prompts_by_user_id used to fetch every active prompt (with users +
all access grants), then call AccessGrants.has_access() once per prompt
that the user did not own. With 600+ prompts this issued ~600 extra
round-trips per request and explained the multi-second delay reported in
the GET /api/v1/prompts and /api/v1/prompts/tags endpoints for non-admin
users.
Push the access check into a single SQL query via the existing
AccessGrants.has_permission_filter (EXISTS subquery), so only accessible
rows come back from the DB. Users and access grants for the surviving
rows are still batch-fetched, no N+1 anywhere on this path.
Co-authored-by: Claude <noreply@anthropic.com>
Convert the /chats/all endpoint from loading all user chats into memory
at once to a streaming NDJSON response that fetches chats in batches of
100. This prevents Out-of-Memory crashes for users with large chat
histories.
Backend: Added async generator that paginates through chats with
short-lived DB sessions per batch (critical for SQLite lock release).
Frontend: Updated getAllChats to consume the NDJSON stream via
ReadableStream reader, accumulating results for the export file.
Ref: open-webui#22206
The transcription endpoint was async but called the synchronous transcribe() function directly, blocking the single-threaded uvicorn event loop for the entire duration of inference. This caused all HTTP and WebSocket connections to stall for every user on the instance during STT processing.
- Add asyncio import
- Use async UploadFile.read() instead of synchronous file.file.read()
- Offload the blocking transcribe() call via asyncio.to_thread()
Closes#24169
* perf(channels): batch user lookup in model_response_handler thread history
The thread-history builder in model_response_handler called
Users.get_user_by_id once per thread message (deduped via an intra-loop
dict), producing N individual SELECTs for a thread of N unique authors.
Replace with a single Users.get_users_by_user_ids call that returns all
authors in one WHERE id IN (...) query, matching the batch pattern
already used elsewhere in this file (lines 739, 804, 1320).
Behavior is preserved: deleted users still resolve to None and fall
through to the existing 'Unknown' fallback via .get().
* refac(channels): rename loop vars to full words per review
Address reviewer feedback to use descriptive names `message` and `user`
instead of single-letter `m` and `u` in the batch user-lookup
comprehensions.
---------
Co-authored-by: Claude <noreply@anthropic.com>
After set_access_grants, the handler was reloading the same knowledge
record via get_knowledge_by_id, which triggers an extra SELECT plus a
nested fetch of access grants. set_access_grants already returns the
newly-written grants and the local knowledge object is otherwise
unchanged, so update it in place and reuse it for the response.
https://claude.ai/code/session_01S18Lgqbih7Ry2JZUUv8TxF
Co-authored-by: Claude <noreply@anthropic.com>
Avoid loading the full Chat row (including the potentially large `chat`
JSON column) just to read tag IDs from `meta.tags`. Issue a narrow
SELECT on `Chat.meta` instead, which is much cheaper for chats with
large message histories.
Co-authored-by: Claude <noreply@anthropic.com>
Convert chat_message_file_ids from list to set so the membership test
in the comprehension is O(1) instead of O(m), turning the dedupe from
O(n*m) into O(n+m). Also replace the redundant set([...]) with a set
comprehension.
https://claude.ai/code/session_01Le3NnqNhcgaJvFrDZGqmwe
Co-authored-by: Claude <noreply@anthropic.com>
Pass the request-scoped AsyncSession into Models.get_model_by_id so the
endpoint no longer opens a fresh DB session on every call, avoiding an
extra connection acquisition per profile image request.
Co-authored-by: Claude <noreply@anthropic.com>
* perf(users): drop redundant get_user_by_id refetch in session-user endpoints
Five /user/* handlers refetched the user row via Users.get_user_by_id(user.id)
immediately after receiving an identical UserModel from Depends(get_verified_user).
Since get_verified_user already populated the user within the same request
microseconds earlier, the refetch is pure overhead. The dead else branches
(unreachable — get_verified_user raises 401 on missing user) are removed as
a natural consequence.
Affected endpoints:
- GET /user/settings
- GET /user/status
- POST /user/status/update
- GET /user/info
- POST /user/info/update
Eliminates one SELECT per request to each of these endpoints with no behavioral
change.
* fix(users): preserve USER_NOT_FOUND error on status update failure
update_user_status_by_id returns None when the target user is missing or
the update raises. The previous commit removed the pre-update existence
gate (get_user_by_id) and returned the update result directly, which
turned not-found/failure cases into 200 OK with a null body instead of
the expected 400 USER_NOT_FOUND.
Guard the update result explicitly to preserve the original API contract,
matching the equivalent pattern already applied in /user/info/update.
* docs(users): note lost-update tradeoff on /user/info/update
Make the concurrency tradeoff explicit: merging against the auth-time
snapshot slightly widens the lost-update window compared to the previous
pre-merge refetch, but the refetch only narrowed (did not eliminate) that
window. Real safety requires row locking or a version column.
---------
Co-authored-by: Claude <noreply@anthropic.com>
Loader.load() dispatches to the underlying langchain document loaders
(PyMuPDF, Unstructured, python-docx, Tika, …) which are all
synchronous and CPU/IO-bound. process_file() awaited it directly on
the event loop, so parsing a non-trivial PDF/DOCX would freeze the
entire FastAPI app for the duration of the parse — which is what users
experience as "the server hangs whenever I upload a file."
Add an `aload()` async wrapper on Loader that runs the sync load on a
worker thread via asyncio.to_thread, and update process_file() to
await it. The sync API is preserved so existing callers that already
run inside run_in_threadpool (e.g. save_docs_to_vector_db) are
unaffected.
https://claude.ai/code/session_01JSr4NZSskEUQvoJnavVXh8
Co-authored-by: Claude <noreply@anthropic.com>
* fix(retrieval): offload sync VECTOR_DB_CLIENT calls in async paths via AsyncVectorDBClient
The vector DB backends (Chroma, pgvector, Qdrant, Milvus, Pinecone,
Weaviate, …) are uniformly synchronous and their methods perform
blocking network or disk I/O. Multiple async route handlers and helpers
were calling them directly on the event loop — file processing,
memories, knowledge bases, hybrid search bookkeeping — so a single
upsert/delete/search would freeze every other in-flight request for the
duration of the call.
Introduce `AsyncVectorDBClient`, a thin async facade that wraps the
existing sync client and dispatches each method through
`asyncio.to_thread`. It mirrors `VectorDBBase` exactly and forwards
*args/**kwargs so backend-specific extra parameters keep working.
Update every async-context call site (routers/retrieval, routers/files,
routers/memories, routers/knowledge, retrieval/utils,
tools/builtin) to await `ASYNC_VECTOR_DB_CLIENT` instead of calling the
sync client directly. Two helpers that were sync-only also acquire
async siblings or are awaited via `asyncio.to_thread` at their async
call site (`remove_knowledge_base_metadata_embedding`,
`get_all_items_from_collections`, `query_doc`).
The original sync `VECTOR_DB_CLIENT` is unchanged, so callers that
already run inside `run_in_threadpool` (e.g. `save_docs_to_vector_db`
and the sync `query_doc`/`get_doc` helpers) are unaffected.
https://claude.ai/code/session_01JSr4NZSskEUQvoJnavVXh8
* fix(retrieval): restore explicit AsyncVectorDBClient signatures matching VectorDBBase
Per PR review: the original *args/**kwargs forwarding lost type
safety and IDE/static-analysis support. Restore explicit signatures
that mirror VectorDBBase exactly, so:
* Bad kwargs fail at the facade boundary instead of inside the
worker thread (where the resulting TypeError tends to be
swallowed by surrounding `try/except`).
* IDE autocomplete and static analysis work as expected.
* The stated intent ("mirror VectorDBBase exactly") now holds at
the API contract level, not just behaviourally.
While doing this, surface a pre-existing bug in
`delete_entries_from_collection` that the stricter typing flagged:
the call passed `metadata={'hash': hash}` which is not a parameter
on `VectorDBBase.delete` nor any backend. The TypeError raised
inside the sync delete was silently swallowed by `except Exception`
so the endpoint always reported `{'status': False}` for every
request instead of actually deleting matching vectors. Replace with
`filter=...` to do what the endpoint name promises.
The thorough review's other note (no concurrency/backpressure on
the shared default threadpool) is intentionally not addressed here:
asyncio.to_thread on the shared executor is the right primitive for
this use case; per-domain bounded executors would add lifecycle
complexity disproportionate to the problem and the loop is no
longer blocked, which was the actual bug.
https://claude.ai/code/session_01JSr4NZSskEUQvoJnavVXh8
* fix(retrieval): parallelize hybrid-search collection prefetch; document async facade contracts
Address PR review findings:
1. Hybrid-search prefetch was sequential
`query_collection_with_hybrid_search` previously awaited
`ASYNC_VECTOR_DB_CLIENT.get(name)` once per collection in a for
loop. Each call already off-loaded to a worker thread, but
awaiting them serially meant total prefetch latency scaled
linearly with the number of collections. Run them concurrently
with `asyncio.gather` so multi-collection queries actually
benefit from the threadpool. Per-collection exception handling
is preserved by wrapping each fetch in a small helper that
logs and returns `(name, None)` on failure, so a single bad
collection cannot poison the whole gather.
2. Document the thread-safety expectation explicitly
The facade now formally states what was always implicit: the
sync `VECTOR_DB_CLIENT` is shared across worker threads, so the
underlying backend driver must be thread-safe. This is not a
new exposure — `save_docs_to_vector_db` already called the sync
client from `run_in_threadpool`. Adding a global lock here
would defeat the responsiveness the facade exists to provide;
backends that cannot tolerate concurrent access should grow
their own internal serialization.
3. Document the API-surface choice and `.sync` escape hatch
The strict `VectorDBBase` mirror was a deliberate choice (the
previous `*args/**kwargs` revision let a `metadata=` typo
silently break an endpoint). Document it, and call out the
`.sync` escape hatch with an example for callers that genuinely
need a backend-specific parameter not on `VectorDBBase`.
https://claude.ai/code/session_01JSr4NZSskEUQvoJnavVXh8
* fix(retrieval): guard /delete against null file.hash and let HTTPException reach the client
Address PR review finding on the `metadata=` → `filter=` change in
`delete_entries_from_collection`.
The new `filter={'hash': hash}` query was correct for files that
have a hash, but did not handle `file.hash is None` (unprocessed,
failed, or legacy records). The match semantics of a null filter
value are backend-dependent — some ignore the key entirely, some
treat it as "metadata field absent" and match every such row — so
issuing the query risked deleting unrelated entries.
* Reject `hash is None` up front with a 400 explaining the file
has no hash to target.
* Narrow the surrounding `except Exception` so it no longer
swallows `HTTPException`. Without this fix the new 400 (and the
pre-existing 404 for missing files) would be silently re-shaped
into `{'status': False}` and the caller could not distinguish a
bad-request input from a backend error.
https://claude.ai/code/session_01JSr4NZSskEUQvoJnavVXh8
---------
Co-authored-by: Claude <noreply@anthropic.com>
* fix(middleware): replace BaseHTTPMiddleware HTTP middlewares with pure ASGI implementations
Starlette's BaseHTTPMiddleware (and the @app.middleware('http')
decorator that uses it) wraps the downstream app in an anyio task
group whose cancel scope tears down the inner task on every exit —
client disconnect, response complete, or any outer middleware bailing.
That CancelledError gets injected into whatever the inner task was
awaiting, so DB queries, embedding calls, and other long awaits get
killed mid-flight. Under aiosqlite the cleanup path then logs a
multi-page `terminate_force_close() not implemented` traceback at
ERROR for every cancelled DB call.
Open WebUI had four such middlewares stacked
(`commit_session_after_request`, `check_url`, `inspect_websocket`,
`RedirectMiddleware`) so a single cancellation would compound through
all four.
Move the four middlewares to a new `open_webui.utils.asgi_middleware`
module as plain ASGI classes (`__call__(scope, receive, send)`):
* `CommitSessionMiddleware` — was `commit_session_after_request`;
now also rolls back if commit fails
before releasing the connection.
* `AuthTokenMiddleware` — was `check_url`; sets request.state
token + enable_api_keys + stamps
X-Process-Time via a wrapped send.
* `WebsocketUpgradeGuardMiddleware`
— was `inspect_websocket`; rejects
/ws/socket.io HTTP requests that
claim transport=websocket without a
proper Upgrade/Connection header.
* `RedirectMiddleware` — was the BaseHTTPMiddleware subclass;
same /watch + share-target rewrites.
Pure ASGI does not introduce a cancel scope around the downstream app,
so client disconnects propagate via `receive()` (the way ASGI was
designed) instead of being injected as CancelledError. Middleware
ordering is preserved.
https://claude.ai/code/session_01JSr4NZSskEUQvoJnavVXh8
* fix(middleware): CommitSessionMiddleware — rollback on downstream error, never commit failed requests
The first cut put commit() in a finally block, which meant that even
when a downstream handler raised, the middleware would still commit
whatever partial sync writes that handler had made before the
failure. That regressed the previous BaseHTTPMiddleware semantics
where commit only ran on the success path.
Restructure the failure handling:
* Downstream raised → rollback any pending sync work, release the
connection, re-raise so the outer error middleware turns it into
an error response. We never commit a request that did not complete.
* Downstream returned → commit. On commit failure, log loudly,
rollback, and re-raise. ScopedSession.remove() always runs in
finally so the connection cannot leak.
Document the inherent pure-ASGI limitation explicitly: by the time
`await self.app(...)` returns the response messages have already
been emitted, so a commit failure can no longer change what the
client sees on the wire. Buffering the response to gate it on commit
success would break streaming responses (chat completions, SSE) which
are core to Open WebUI; the trade-off is intentional. Routes that
need commit-before-send must manage the sync session explicitly.
Also drop unused `typing` imports flagged by review.
https://claude.ai/code/session_01JSr4NZSskEUQvoJnavVXh8
---------
Co-authored-by: Claude <noreply@anthropic.com>
* fix: drop extra='allow' on FolderForm and FolderUpdateForm
These request models were configured to accept arbitrary extra fields,
which were then merged into the folder row via form_data.model_dump().
In insert_new_folder the server-assigned user_id is placed before the
form spread, so a client-supplied user_id in the request body would
override it and the folder would be persisted against another account.
Strictly typed inputs are the correct shape for these endpoints — the
client has no legitimate reason to send fields beyond the declared
ones, and dropping extra='allow' closes the mass-assignment sink at
the validation layer instead of relying on every callsite to merge
fields in the right order.
* fix: reject unknown fields on FolderForm and FolderUpdateForm
Address review feedback: dropping extra='allow' fell back to Pydantic
v2's default extra='ignore', which only silently drops unknown fields
instead of rejecting them. The intent for these request models is a
strict input contract — fail fast when a client sends anything the
server does not expect — so explicitly set extra='forbid'. This also
makes the hardening visible in the form definition rather than implicit
in the default.
These four Redis calls were the only places in the backend still using
bare cache keys. Every other call already namespaces through
REDIS_KEY_PREFIX, which is how multiple Open WebUI instances sharing a
Redis database stay isolated. Two instances with different
TOOL_SERVER_CONNECTIONS or TERMINAL_SERVER_CONNECTIONS would otherwise
clobber each other's cached OpenAPI specs.
* fix: replace brittle profile_image_url allowlist with safe-scheme validation
The previous validation used a hardcoded allowlist of specific static
paths and a single Gravatar prefix. This rejected OWUI's own internal
API paths (e.g. /api/v1/users/{id}/profile/image) and external OAuth
avatar URLs, making it impossible to save user profiles from the admin
panel.
Replace with scheme-based validation that allows relative paths,
HTTP(S) URLs, and data:image URIs while blocking dangerous schemes
like javascript:, file:, and ftp:.
Fixes open-webui#23387
* fix: harden profile image URL validation per review feedback
- Restrict data URIs to safe raster formats (png/jpeg/gif/webp);
SVG is excluded because it can carry embedded scripts.
- Block scheme-relative URLs (//host/path) which browsers resolve
against the current protocol, bypassing the relative-path check.
* fix: use structural validation instead of prefix checks
- Use urlparse for HTTP(S) URLs: gives case-insensitive scheme
matching and rejects bare schemes with no host (e.g. https://).
- Use a compiled regex for data URIs: enforces the ;base64, boundary,
restricts to safe raster formats, and is case-insensitive per spec.
- Removes the startswith-based prefix tuple in favour of proper
URL and data URI parsing.
* fix: validate hostname not netloc, fix misleading comment
- Use parsed.hostname instead of parsed.netloc so URLs like
http://:80/path (non-empty netloc but no actual host) are rejected.
- Update data URI comment to accurately state we validate MIME type
and structure, not base64 payload integrity.
* fix: constrain relative paths to known-safe prefixes
Accepting any relative path starting with / allowed a user to set
their profile_image_url to an arbitrary internal GET endpoint. When
another user (e.g. an admin) views that profile, the browser fires
the GET with the viewer's session cookies — an authenticated GET
trigger surface.
Constrain to known-safe prefixes (/api/v1/users/, /static/) and
exact matches (/user.png, /favicon.png) which are the only relative
paths OWUI itself generates.
* fix: use exact matches and anchored regex, eliminate all prefix wildcarding
Replace all startswith-based path checks with:
- frozenset exact matches for static assets (/user.png, /favicon.png,
/static/favicon.png)
- Anchored regex for the OWUI profile image API route that accepts
only /api/v1/users/{id}/profile/image (no trailing components,
no path traversal across segments)
This eliminates every prefix-based attack surface:
- /api/v1/users/{id}/anything-else is rejected
- /static/../../etc/passwd is rejected
- /api/v1/users/../../admin/config is rejected
- Arbitrary internal GET triggers are no longer possible
* fix: exclude query/fragment delimiters from user-ID regex segment
Change [^/]+ to [^/?#]+ so that inputs like
/api/v1/users/alice?x=1/profile/image are rejected — the browser
would interpret ? as the query string start, making the actual
request target /api/v1/users/alice instead of the intended route.
* Add ownership checks to global task endpoints
- Restrict GET /api/tasks and POST /api/tasks/stop/{task_id} to admin-only
- Add new scoped POST /api/tasks/chat/{chat_id}/stop endpoint with ownership
check so regular users can stop their own chat tasks
- Allow admins to access the scoped chat task endpoints alongside owners
- Update frontend to use the new scoped stop endpoint when a chatId is available
https://claude.ai/code/session_01K7zPDvvjRu8AxJ4Br2HhZc
* Handle temporary (local:) chat IDs in scoped task endpoints
Temporary chats use local:<socketId> as chat_id which doesn't exist in
the DB. The scoped endpoints now skip ownership checks for local: IDs
(they aren't enumerable) and use {chat_id:path} to handle the colon in
the URL path.
https://claude.ai/code/session_01K7zPDvvjRu8AxJ4Br2HhZc
* Verify session ownership for local: chat IDs and URL-encode chat_id
- For local:<socketId> chat IDs, look up the socket's owner in
SESSION_POOL and verify it matches the requesting user (or admin)
- URL-encode chat_id in frontend fetch calls to handle special
characters (colon in local: IDs) safely
https://claude.ai/code/session_01K7zPDvvjRu8AxJ4Br2HhZc
---------
Co-authored-by: Claude <noreply@anthropic.com>
