Adds LITELLM_SUPPRESS_SPEND_LOG_TRACEBACKS env var. When set to true and the
proxy log level is INFO or above, spend-tracking error paths emit a single
ERROR line without the full traceback. Stack traces are preserved at DEBUG
and the Sentry / proxy_logging_obj.failure_handler path is unchanged.
The new spend_log_error helper is wired through the spend write hot path:
- DBSpendUpdateWriter (update_database, _update_*_db, batch upsert,
redis-commit fallbacks)
- _ProxyDBLogger._PROXY_track_cost_callback
- get_logging_payload exception path
- update_spend / update_daily_tag_spend / spend logs queue monitor
Resolves LIT-2704.
Co-authored-by: Mateo Wang <mateo-berri@users.noreply.github.com>
* fix(proxy): re-validate user_id ownership after /user/info re-parses query
The route-level access check in `RouteChecks.non_proxy_admin_allowed_routes_check`
reads `request.query_params.get("user_id")`, which decodes literal `+` to
spaces. The endpoint then re-parses the raw query string with `urllib.unquote`
in `get_user_id_from_request` to preserve `+` characters (so plus-addressed
emails work as user_ids). Those two paths produce different ids: a caller
who registered a user_id containing a literal space could pass the route
check and then read another user's row by sending the encoded `+` form.
Add `_enforce_user_info_access` and call it after `_normalize_user_info_user_id`
returns the final id. Proxy admin / view-only admin still bypass; everyone
else must match the resolved user_id (or have no user_id, which falls back
to the caller's own id later in the handler).
Tests cover the admin bypass, owner-match path, and the cross-user lookup
that this change blocks.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(proxy): apply user_info ownership check to PROXY_ADMIN_VIEW_ONLY
`_enforce_user_info_access` was bypassing both PROXY_ADMIN and
PROXY_ADMIN_VIEW_ONLY, but the upstream route check in
`RouteChecks.non_proxy_admin_allowed_routes_check` only treats
PROXY_ADMIN as a true admin for the `/user/info` route — view-only
admins go through the `user_id == valid_token.user_id` enforcement
along with regular users. Mirroring that asymmetry left the same
encoded-`+` bypass open for view-only admins whose user_id contains a
literal space.
Drop the PROXY_ADMIN_VIEW_ONLY exemption so the post-decode re-check
matches the upstream rule. Update tests: a view-only admin must now
be blocked from cross-user lookups but still allowed to read their
own row.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: yuneng-jiang <yuneng@berri.ai>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(batches): count non-chat tokens and validate every model in batch file
Two security control bypasses on POST /v1/batches:
1. `_get_batch_job_input_file_usage` only summed tokens for
`body.messages` (chat completions). Embedding (`input`) and text
completion (`prompt`) batches reported zero, letting massive
non-chat workloads slip past TPM rate limits. Extend the counter
to handle string and list shapes for both fields.
2. The batch input file was forwarded to the upstream provider
without inspecting the models named inside the JSONL — only the
outer `model` query parameter was checked against the caller's
allowlist. A caller restricted to gpt-3.5 could submit a batch
targeting gpt-4o and the upstream would execute it under the
proxy's shared API key.
Add `_get_models_from_batch_input_file_content` (returns the
distinct `body.model` values) and call it from
`_enforce_batch_file_model_access` in the pre-call hook, which runs
each model through `can_key_call_model` so the same allowlist
semantics (wildcards, access groups, all-proxy-models, team aliases)
the proxy enforces on `/chat/completions` apply here too. Any
unauthorized model raises a 403 before the file is forwarded.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(batches): count pre-tokenized prompt/input shapes, classify 403 logs
Two follow-ups from the Greptile review on the batch validation PR:
1. P1 TPM bypass via integer token arrays. The OpenAI batch schema
accepts ``prompt`` and ``input`` as ``list[int]`` (a single
pre-tokenized prompt) or ``list[list[int]]`` (multiple) in addition
to the string and ``list[str]`` shapes. Pre-fix only the string
shapes were counted, so a caller could submit a batch with hundreds
of millions of pre-tokenized tokens and the rate limiter would
record zero. Extract the per-field logic into
``_count_prompt_or_input_tokens`` and count each int as one token.
2. P2 access-denial logs were indistinguishable from I/O failures.
``count_input_file_usage`` caught every exception under a generic
"Error counting input file usage" message, so an intentional 403
from ``_enforce_batch_file_model_access`` looked the same in the
logs as a missing file or a Prisma timeout. Catch ``HTTPException``
separately and log 403s at WARNING level with a security-relevant
message before re-raising.
Tests cover the new shapes: single ``list[int]``, ``list[list[int]]``
(the worst-case bypass vector), and embeddings ``input`` with
pre-tokenized arrays.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Reset the policy ID index during policy engine test cleanup so stale policy versions cannot leak between tests.
Co-authored-by: Cursor <cursoragent@cursor.com>
Mark runtime-created policies and attachments initialized so global policy attachments created from the policy builder apply immediately without requiring a restart.
Co-authored-by: Cursor <cursoragent@cursor.com>
When a team grants /key/list via team_member_permissions, non-admin members
should see all keys for that team — same as a team admin. Previously the
classification in list_keys() only checked admin status, so permitted
members fell into the service-account-only path and could not see other
members' personal keys. Routes those members into the full-visibility set.
Greptile P1: the unsafe-method branch of `_check_proxy_admin_viewer_access`
ended with a blanket `if route in management_routes: return`. That set is a
mix of reads (info/list — handled via the safe-method GET branch above) and
writes. The fallback let Admin Viewer POST to write endpoints not enumerated
in `_ADMIN_VIEWER_BLOCKED_WRITE_ROUTES`, including:
- /team/block, /team/unblock, /team/permissions_update
- /jwt/key/mapping/{new,update,delete}
- /key/bulk_update
- /key/{key_id}/reset_spend
Remove the fallback. The two remaining allow sets (admin_viewer_routes and
global_spend_tracking_routes) are both read-only, so removal does not affect
the legitimate POST-as-read cases (e.g. /spend/calculate, which is in
spend_tracking_routes ⊂ admin_viewer_routes).
Tests:
- 8 new parametrized cases pinning each previously-leaking management write
endpoint to 403 on POST for PROXY_ADMIN_VIEW_ONLY.
The reservation path (PR #26845) atomically pre-fills `spend:user:{user_id}`
and admits at the strict-`<` boundary. The legacy `_PROXY_MaxBudgetLimiter`
pre-call hook re-reads the same counter with `>=`, so a reservation that
fills the counter to exactly `max_budget` (e.g. a request without a
`max_tokens` cap that falls back to reserving the smallest remaining
headroom) is rejected by the hook even though the reservation already
admitted it.
Skip the hook when the request's active `budget_reservation` covers
`spend:user:{user_id}`. The reservation is the source of truth for that
counter cross-pod; the legacy `>=` path remains in place for requests
without a reservation (e.g. paths that bypass the reservation entirely).
Reproduces as `tests/otel_tests/test_prometheus.py::test_user_budget_metrics`
on a fresh user with `max_budget=10` calling `fake-openai-endpoint` without
`max_tokens`. Adds focused unit coverage in
`tests/test_litellm/proxy/hooks/test_max_budget_limiter.py`.
Greptile flagged two follow-ups on the OpenAPI/local-registry pre-call
check:
1. **P1 runtime crash via None proxy_logging_obj.**
`kwargs.get("proxy_logging_obj")` is `None` on the MCP entry path,
and `pre_call_tool_check` calls `proxy_logging_obj._create_mcp_request_object_from_kwargs`
unconditionally after the security checks, which would have crashed
every legitimate call with `AttributeError`. Source the logging
object from `litellm.proxy.proxy_server` the same way
`_handle_managed_mcp_tool` already does.
2. **P2 authorization-bypass window when mcp_server is None.**
Previously the new check was guarded by `if mcp_server is not None`,
so any local tool whose registry entry had no resolvable server (a
startup-race window before `_initialize_tool_name_to_mcp_server_name_mapping`
completes, or an orphaned registry entry) ran without the security
check. Tools registered via openapi_to_mcp_generator are always tied
to a server, so a missing one is a configuration/timing fault — fail
the call with 503 instead of dispatching unguarded.
Tests: existing two pass with an added assertion that
`proxy_logging_obj` is non-None at the call site, plus a new test that
covers the 503 deny branch when the tool→server mapping is missing.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
`execute_mcp_tool` dispatches in two ways: managed MCP servers go
through `_handle_managed_mcp_tool`, which calls
`MCPServerManager.pre_call_tool_check` to enforce allowed/banned tool
lists, key/team `object_permission` tool grants, and parameter
validation. OpenAPI-backed tools, however, were resolved via
`global_mcp_tool_registry` and dispatched directly to
`_handle_local_mcp_tool` — entirely skipping `pre_call_tool_check`.
A caller could invoke any registered OpenAPI tool regardless of their
key/team permissions, including administrative or destructive
operations on the upstream API.
Run `pre_call_tool_check` before the local-registry dispatch whenever
the resolved server is set (the same condition used to surface server
context to the managed path). Honor any guardrail-modified arguments
the hook returns. Errors raised by the hook propagate up before
`_handle_local_mcp_tool` runs.
Tests cover both directions: the pre-call hook fires when the local
tool resolves alongside a server, and a hook-raised HTTPException
prevents the local handler from being invoked.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The GitHub merge conflict resolver concatenated both test sets but left
`from litellm.proxy._lazy_openapi_snapshot import _normalize_operation_ids`
stranded between functions instead of at the top of the file.
A non-admin scoped to ["model-a"] could call /health?model_id=id-b
(where id-b belongs to a deployment outside their scope) and the
background-cache code path would return id-b's cached health entry. The
helper returned {model_id} unconditionally, so the cache filter was
driven by an unvalidated id and the global cache leaked the entry — the
ternary `targeted_ids if not None else allowed_model_ids` skipped any
intersection with the caller's allowed deployments.
Make _resolve_targeted_model_ids walk the supplied model_list for both
the model and model_id branches. Callers pass an already-scoped list
(filtered to allowed model_names for non-admins, full list for admins),
so an out-of-scope model_id resolves to an empty set and the cache
filter drops every entry — matching the live path's existing behavior.
Per review: `assert response.status_code != 503` is satisfied by 404,
500, or any other non-503 code, so a regression that returned the wrong
non-503 status would slip through. Switch to `== 200` so the assertions
verify the actual expected status, not just the absence of one specific
failure.
When use_background_health_checks is enabled, /health?model=foo returned
the full cached aggregate across every model — so an unhealthy foo
combined with any other healthy deployment kept healthy_count > 0 and
the targeted-503 path never fired.
Resolve the targeted model/model_id to a deployment-id set first
(mirroring perform_health_check's match-on-model_name-or-litellm_model
semantics) and narrow the cache to those IDs before _post_process
evaluates healthy_count, so the 503 contract holds for both the live
and cache code paths.
/health?model=foo and /health?model_id=foo previously returned HTTP 200
even when zero endpoints were healthy, forcing monitoring systems to
parse the JSON body to detect failure. /health/readiness similarly
returned 200 even when a configured Prisma DB was unreachable, leaving
unhealthy pods in rotation.
Both endpoints now flip to HTTP 503 in the failure case while keeping
the JSON response body identical, so existing parsers continue to work
and orchestrators can rely on the HTTP status alone.
If atomic_check_and_increment_by_n returns overall_code=OVER_LIMIT but no
status entry matches a descriptor key the dynamic limiter dispatcher knows
how to translate into a 429 (`model_saturation_check` or `priority_model`),
the for-loop previously exited cleanly and execution fell through to the
priority-tracking increment + the data["litellm_proxy_rate_limit_response"]
write — silently admitting an over-limit request.
This is the fail-open path a future contributor would hit by wiring a new
descriptor type into enforced_descriptors without updating the dispatcher.
Refuse the request with a generic 429 carrying the offending descriptor
metadata so the operator can see what slipped past, and emit an error log
to surface the wiring gap.
Adds a regression test (test_dynamic_rate_limiter_v3_fails_closed_on_unknown_descriptor)
that drives the limiter with a synthetic OVER_LIMIT response carrying an
unrecognized descriptor_key and asserts a 429 is raised.
Tests: 65 passed (1 skipped), 0 regressions.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix: add expected_base_url origin check to get_litellm_gateway_api_key
* fix: scope get_stored_api_key and save base_url on login
* fix: pass base_url to get_stored_api_key in CLI entrypoint
* fix: scope ProxyClient stored key to base_url
* test: add expected_base_url coverage for get_stored_api_key
* fix: initialize self.http with resolved api_key not raw param
* fix: black formatting in client.py and test_auth_commands.py
- Lua script now reads time via redis.call('TIME') instead of a client-supplied
timestamp. Prevents window-reset divergence across replicas with skewed
wall-clocks, which could otherwise reopen the cross-replica TOCTOU window.
- Per-descriptor window_size is now plumbed through both the Lua ARGV layout
and the in-memory fallback. Previously the in-memory path used the global
self.window_size while Lua honored the per-descriptor override, so a
descriptor with a custom window would be enforced inconsistently between
Redis-available and Redis-unavailable code paths.
- Lua-failure fallback path now logs at error severity and explicitly
documents the in-memory ↔ Redis state divergence risk so operators can
alert on it. Prior `warning` log understated the impact.
- Coarse-granularity lock is now documented inline with the conditions under
which a per-descriptor sharded lock would be worth introducing.
- New regression test: zero-token batch consumes RPM only and is properly
capped by the RPM ceiling (validates the asymmetric quota path that arises
from `inc_amount <= 0: continue`).
Tests: 64 passed (1 skipped), 0 regressions. Multi-instance Redis loadtest
re-verified: chat 20/80 success @ RPM=20, batches 3/20 @ TPM=200.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* refactor: scope /health response to caller's models and tidy display fields
Two small consistency changes to the /health response:
1. health_endpoint() now narrows _llm_model_list to deployments whose
model_name is in user_api_key_dict.models, matching how other model
listing endpoints already scope their output. The same narrowing applies
to the cached health_check_results dict when background_health_checks is
enabled, via a new _filter_health_check_results_by_model_ids helper.
2. ILLEGAL_DISPLAY_PARAMS in health_check.py picks up api_base and
api_version, which are provider routing fields and not part of the
health response shape.
Tests in tests/test_litellm/proxy/health_endpoints/test_health_endpoints.py
pin both behaviors so future changes do not widen the response shape.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* address greptile review feedback (greploop iteration 1)
- tests: extend background-cache test with model_id on cached entries plus
positive assertions that model-a's deployment is the one returned, so
the test is no longer satisfied by an empty result.
- _health_endpoints.py: add a verbose_proxy_logger.debug line when a scoped
key has accessible model_names but the matching deployments have no
model_info.id, so the empty cache-result case is observable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* make background-cache test's non-vacuity explicit
Restructure test_health_endpoint_filters_background_cache_by_user_access
so the assertions positively pin the post-scoping result (one entry,
model_id == "id-a", api_base == https://example-a.test) and add fixture
sanity checks that confirm the source cache had two entries and every
cached entry carries a model_id.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* gate api_base in /health response on proxy-admin role
Replace the blanket strip of api_base / api_version with a role-aware
post-processor:
- api_base is now left in the cleaned per-deployment dict that
_clean_endpoint_data produces (api_version stays in the denylist).
- health_endpoint() removes api_base from each endpoint entry before
returning when the caller's user_role is not PROXY_ADMIN /
PROXY_ADMIN_VIEW_ONLY. The strip uses a copy so the shared
health_check_results cache still carries api_base for subsequent
admin reads.
Net effect: a proxy admin can still see which Vertex region or Azure
resource is healthy in the /health output, while non-admin keys (and
read-only keys) only see model / model_id / status fields.
Tests:
- test_health_endpoint_admin_sees_api_base_non_admin_does_not pins both
branches and verifies the cache is not mutated.
- test_clean_endpoint_data_strips_credentials_but_keeps_api_base
replaces the previous mask/drop tests now that the cleaning helper
no longer touches api_base.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* address review feedback: api_version symmetry, missing-id warnings, deprecation header
Three blockers raised in review:
1. api_version asymmetry — api_base was role-gated for proxy admins, but
api_version was unconditionally stripped via ILLEGAL_DISPLAY_PARAMS.
Move api_version out of the credential denylist and into a new
ADMIN_ONLY_HEALTH_DISPLAY_PARAMS tuple alongside api_base, so admins
keep both routing fields and non-admins lose both. Useful for telling
apart Vertex regions or Azure api-versions from the /health response.
2. Silent empty results when scoped key's deployments lack model_info.id —
raise the existing log from .debug to .warning, and add a structured
"warnings" field to the response so the caller can distinguish "no
deployments configured" from "deployments excluded due to missing
model_info.id".
3. Migration signal for the api_base / api_version removal — when a
non-admin caller hits /health, set a "Litellm-Health-Field-Notice"
response header so existing dashboards or scripts that parsed those
fields can detect the change programmatically rather than silently
seeing absent keys.
Tests adjusted: existing background-cache test injects a Response stub,
admin-vs-non-admin test now asserts both api_base and api_version are
gated and asserts the notice header. New test covers the warnings field
when a scoped key's deployments are missing model_info.id.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* defensive copies + clarifying comments in /health filter
- _filter_health_check_results_by_model_ids now shallow-copies each
retained endpoint dict before returning. The shared module-level
health_check_results cache should never be mutated by downstream
transforms, even though _strip_admin_only_fields_from_health_result
already builds new dicts today.
- Document the live (model_name) vs cache (model_id) scoping asymmetry
so future readers do not have to derive it from the warnings field.
- Document why _PROXY_ADMIN_ROLES includes PROXY_ADMIN_VIEW_ONLY (read-
only operators need routing fields to diagnose health).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: yuneng-jiang <yuneng@berri.ai>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Greptile follow-ups on the prior commit:
- (P1) ``is_request_body_safe`` recursed into ``litellm_embedding_config``
with no depth bound, so a request body 1000 levels deep could exhaust
Python's call stack and surface a 500 ``RecursionError``. Refactored
the check to be iterative (single-level descent into a fixed list of
nested-config keys) and extracted the per-dict banned-param scan into
a helper that's shared between the root and the nested call sites.
Also fixes the ``recursive_detector`` CI job that was triggered by
the recursive-by-name pattern.
- (P2) ``assert_same_origin`` error messages identified the mismatching
component but echoed the ``expected`` host and the candidate
hostname back to the caller. In the SSRF threat model the caller is
the attacker, so reflecting that information was a secondary leak of
operator infrastructure. Messages now identify only *which*
component mismatched (scheme / host / port) without naming names.
- (P2) ``_NESTED_CONFIG_KEYS`` was defined after the function that used
it. Hoisted the constant (and the new ``_BANNED_REQUEST_BODY_PARAMS``
tuple) above the function for readability.
Adds a 1000-level-deep nested config test that asserts no
``RecursionError`` and a hostname-leak test that asserts no operator
host appears in the rejection message.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The async CustomStreamWrapper.__anext__ caught httpx.TimeoutException
and re-raised it raw, bypassing _handle_stream_fallback_error. The
Router's FallbackStreamWrapper only reacts to MidStreamFallbackError,
so stream_timeout firing mid-stream never triggered fallbacks the way
connection-phase timeout does.
Route timeouts through _handle_stream_fallback_error like every other
exception so they wrap into MidStreamFallbackError and the Router can
switch to a fallback model.
Two SSRF findings were OPEN with no in-flight fix; both are closed
now using narrow defenses that key off existing trust boundaries.
VERIA-6 (Milvus ``litellm_embedding_config``):
``is_request_body_safe`` already blocks ``api_base`` / ``api_key`` /
``langfuse_host`` / ``s3_endpoint_url`` / etc. at the *root* of the
request body, gated by an admin opt-in (``allow_client_side_credentials``
or per-deployment ``configurable_clientside_auth_params``). The bug is
that the Milvus vector-store transformer unpacks
``litellm_embedding_config`` into ``litellm.embedding(**embedding_config)``,
so a caller can smuggle the same banned params in via nesting and bypass
the check. Fix: ``is_request_body_safe`` now recurses into a known list
of nested-config dicts (``litellm_embedding_config`` for now) and applies
the same banned-param check with the same admin opt-in. Admin-side
vector-store config flows through ``litellm_params`` rather than the
request body, so it's unaffected.
VERIA-51 (polling URLs returned by upstream APIs):
Azure DALL-E 2, Azure Document Intelligence, and Black Forest Labs
all blindly fetched a polling URL returned by the upstream and
attached the operator's API key to the request. A compromised upstream
or a future API contract change could redirect credentials anywhere.
New ``url_utils.assert_same_origin(candidate, expected)`` helper checks
scheme, host (case-insensitive), and port (with default-port
normalization). Applied at all five polling sites: Azure DALL-E
sync+async, Azure DI sync+async, BFL image generation sync+async, BFL
image edit sync+async. Cross-origin polling URLs now raise rather than
forward credentials. The Azure DALL-E ``Expected 'status' in response``
exception no longer reflects the raw response body — that path turned
Blind SSRF into Full-Read SSRF for the limited window before the
origin check fully closed it.
Tests: 7 ``assert_same_origin`` unit tests, 6 ``is_request_body_safe``
nested-config tests, 5 polling-site rejection tests + 1 same-origin
sanity check.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
