A tools/call on a cold worker listed the target server once and then never
again, so a later caller whose credentials expose a wider upstream catalog
got 404 for tools the first caller never had. Gate the pre-call listing on
whether this worker already exposes the requested tool, so callers with
different catalogs no longer mask each other. Removing the per-server guard
also drops the empty-listing case that re-listed on every call.
The startup tool-name fill skips servers whose upstream wants the caller's
own token (true_passthrough, OAuth discovery), and mcp 2 no longer runs the
list handler before an uncached tools/call, so every uvicorn worker that had
not served tools/list answered 404 "Tool not found" for prefixed tools/call
and the REST server_id route on those servers.
On a resolution miss, execute_mcp_tool now lists the prefix-matched (or
server_id-requested) server once, with the caller's credentials, through the
existing tools/list path, then resolves as before. Listing failures fall
through to the existing 404, a worker that already listed the server never
re-lists it, and a server outside the caller's allowed set is never listed.
Format the ported files, annotate mutable wire payloads, give the e2e
OAuth client the SDK 2 httpx2/AuthorizationCodeResult API, tighten the
transport-streams alias to the two-stream SDK 2 shape, and add a
test-quality reason for the MockTransport factory injection.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Call handlers with ServerRequestContext and params models, seed the
litellm contextvar instead of the removed SDK request_ctx, forward
headers/auth through the httpx2 MockTransport factory, and add
regressions for handler registration, context propagation, and modern
protocol-version rejection.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
mcp_allowed_clients entries become {alias, value} objects: the value is what the JWT claim or header must equal, the alias is the name the dashboard and logs show. The Network Settings section is renamed Allowed Clients with one alias/value row per client
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Rename McpError/isError/inputSchema-style references to the SDK 2
spellings, parse the JSONRPCMessage union with a TypeAdapter, and drive
the SDK transports off httpx2 MockTransport injection where respx can no
longer intercept. Adjust for SDK 2 behavior: the initialize handshake
negotiates handshake-era protocol versions only, an empty SSE stream
surfaces CONNECTION_CLOSED, non-2xx tool responses surface INTERNAL_ERROR
MCPError instead of HTTPStatusError, and the SDK read timeout carries the
JSON-RPC REQUEST_TIMEOUT code.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Adds an admin-only DELETE /v1/mcp/sessions that terminates stateful MCP gateway sessions on the current worker by session id prefix and/or by the LiteLLM user that opened them, tombstones the terminated ids so a client reusing one gets 404 instead of a silently recreated stateless session, and lets PROXY_ADMIN name a user_id on the BYOK and OAuth credential delete routes. Full and view-only admins can list every user's stored credential metadata for a server (never the secret). The dashboard gains Disconnect controls on the Live Connections tab and a User Credentials tab with Revoke controls, both hidden from read-only admins.
Resolves LIT-8001
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
A permanent database fault (a missing or version-skewed query engine)
in the subject_token check was answered with the same "retry" wording
as a transient outage. The status stays 503 temporarily_unavailable,
the only OAuth error a client reads as the server's fault and what the
mint path already answers to the same fault, but the description now
says retrying will not help until the deployment is repaired, using
PrismaDBExceptionHandler.is_permanent_database_fault the way the mint
path does.
A subject token JWT auth could not check, because the IdP's JWKS was unreachable with no cached copy or the auth database was down, came back as 400 invalid_request with the same fixed message a bad token gets, so clients re-logged in instead of retrying the way they already do for a mint-time 503. Those checks now answer 503 temporarily_unavailable and log the reason, while real rejections stay 400 invalid_request.
With mcp_allowed_clients set the gateway used to read the whole POST body to
find clientInfo.name, so an authenticated client could make the proxy buffer an
arbitrarily large payload. Inspection is now capped at MCP_ALLOWLIST_PEEK_MAX_BYTES
and a sessionless POST that exceeds the cap is rejected with 403 before routing,
while posts on an admitted session stream through unchanged
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
The proxy stores both fields with a whole-row read-modify-write of
general_settings, so two concurrent writes from one save can drop one
of them. Also drops docstrings and suppressions the diff did not need
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Adds the mcp_allowed_clients general setting, enforced against the
clientInfo.name each MCP client sends in its initialize request. A client
not on the list, or one that does not identify itself, is rejected with
403 before any stateful session is created. The setting is configurable
from config.yaml and from the Admin UI MCP network settings page
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
An api_key server whose key lives in static_headers, the documented
shape for upstreams that expect a custom header name, dispatched fine
before the fail-closed check and was rejected as misconfigured after
it. The check now treats every static header the admin configured as a
credential slot for api_key mode, on both the MCP client path and the
OpenAPI tool path, with regression tests at all three layers.
JWT auth caches the user row before it adds the user to the JWT's team, and admission checks the credential's team against the cached row on whichever worker takes the next request. On a two-worker gateway the credential minted for a newly joined team answered 403 "not in your team memberships" until the management-object TTL ran out, because /team/member_add only evicted the membership spend sentinel. The add now evicts the added members' cached user rows and broadcasts the eviction to the other workers, the way /team/member_delete already did
The mint test now also covers a user SCIM deactivated after the cache last saw them active: the database read refuses the mint while the cached row still says active
The token exchange mint keeps reading the user row from the database, since JWT auth caches the user it creates before adding it to the JWT's team and a mint off that cached row refused the first exchange for a new user. Introspection and the refresh revalidation go back to the cache read, so a resource server calling /introspect per request pays no database read.
JWT auth caches the user it creates before it adds that user to the JWT's
team, and adding a team member never evicts the cached user row, so the mint
read a row with no teams and refused the very first token exchange for a
never-seen user as not a member. The loader now reads the row from the
database and leaves the fresh row in the cache for the requests the
credential makes next
