* Support CallArgument and Depends bindings from uncalled-for 0.4.0
uncalled-for 0.4.0 adds explicit argument references: CallArgument()
lets a dependency factory read an argument of the function it serves,
and Depends(factory, **bindings) supplies factory arguments at the
declaration site (https://github.com/chrisguidry/uncalled-for/pull/12).
FastMCP's resolver now opens a frame_scope() around dependency
resolution, with the sanitized user arguments as the frame's provided
values. A CallArgument can reference a tool call's public parameters,
but a caller-supplied value for a dependency parameter name is still
stripped before resolution. CallArgument and CycleError are re-exported
from fastmcp.dependencies, and the dependency-injection docs cover both
features.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Raise the pydocket floor to 0.24.0 outside Windows
pydocket 0.24.0 resolves TaskArgument and CallArgument through
uncalled-for 0.4.0's call-scoped frames. Windows keeps the 0.20.0
floor: the burner-redis<0.1.7 pin there transitively caps pydocket to
<0.20.2, and burner-redis has shipped no fixed release yet.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Bump the pydocket floor to 0.24.1 for reliable worker shutdown
docket 0.24.1 fixes a lost cancellation in worker shutdown on Python
3.10 and 3.11 (chrisguidry/docket#456): asyncio.wait_for swallowed a
cancellation delivered in the same event-loop tick that its inner future
completed, so cancelling run_forever during our lifespan teardown left
the worker running and hung the test session. That is what timed out the
Python 3.10 and lowest-direct jobs here. The floor stays platform-split;
Windows keeps >=0.20.0 under the burner-redis pin.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Drop the Windows burner-redis pin and unify the pydocket floor at 0.24.1
The pin blamed the wrong package. The Windows "interpreter crash" that
motivated it (#4618) was pydocket 0.23.1 losing an external cancellation
during worker teardown; pytest-timeout's hard kill of the hung xdist
worker discarded its stdout and looked like a native fault. Capping
burner-redis also dragged pydocket below 0.20.2, so the two variables
were never separated. The repro matrix on prefectlabs/burner-redis#7
shows the July environment failing as resolved, passing with only
pydocket rolled back, and passing with pydocket 0.24.1 alongside
burner-redis 0.1.7 on Windows. pydocket 0.24.1 carries the fix
(chrisguidry/docket#456), so every platform now shares one floor.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Co-authored-by: nate nowack <thrast36@gmail.com>
* feat: add Horizon authentication client and state
* fix: apply Windows state ACLs to existing descriptors
* fix: distinguish public route authorization failures
* fix: harden Horizon state boundaries
* Separate proxy protocol policy from client construction
🤖 Generated with OpenAI Codex
* Strip connection-owned request metadata at the proxy backend boundary
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Sanitize forwarded request metadata where the proxy copies it
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Forward hop-safe request metadata for proxied resources, templates, and prompts
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
* Fix upgrade static analysis
Generated with Codex
* Preserve concrete transport return types
Generated with Codex
* Avoid widening transport return types
Generated with Codex
* Model transforming transport return types
Generated with Codex
* Exclude standalone screenshot examples from ty
Generated with Codex
pytest-timeout falls back to its thread method on Windows, which os._exit()s
the process instead of failing the test. A single slow test therefore kills an
xdist worker and fails whichever unrelated test it was running.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
* Rewrite the v4 What's New page
Teach the headline features with code instead of asserting them, drop the
major-version throat-clearing and SEP list, and correct the elicitation
claim: ctx.elicit() is unchanged and handshake-only, while sampling and
roots are removed outright.
* Fix broken doc links and stale version references
Repoint five dead links and anchors, refresh v3-era version examples on the
v4 docs, and add the missing FastMCP 3 entry to the installation page's
upgrade section.
* Document server extensions
add_extension() shipped in v4 with no documentation page. Covers the
extension interface, request methods, tool-call interception, lifespan
ownership, and the client half.
* Link the FastMCP TypeScript library
* Address Codex review feedback
Gate the extension interceptor on the client's per-request opt-in rather
than claiming negotiation does it; show the v4 beta pin on the install
page instead of a version a reader cannot get; note that UserSession
requires authentication.
* Always emit a tool title, derived from name when unset
Some MCP clients (e.g. ChatGPT) drop tools with no `title` instead of
falling back to `name` for display as the spec allows. Deriving a
default title in Tool.to_mcp_tool() fixes this for every tool built on
top of it, including the search-transform, code-mode, and session
proxy tools that never set one explicitly.
Fixes#4414
* Derive fallback title from the overridden name, document it
Addresses Codex review on #4694.
* Resolve title precedence from effective overrides
* Normalize mapping annotations before deriving the title
* Add v4.0.0b1 changelog and updates entries
* Drop meta note from b1 intro; add #4682 under enhancements
* File #4682 under fixes
* Correct the camelCase rename claim: Python model fields, not the wire
* Baseline the b1 changelog on v3.4.5
* Simplify the beta banner
* Flatten OpenAPI discriminator subtypes into request bodies
* Resolve schema-name discriminator mappings and union conflicting variant fields
* Advertise discriminator values for propertyless variants and document the behavior
* Remove 3.x-era compatibility shims
* Require response_type in ctx.elicit()
* Name the utilities path for the two non-re-exported auth helpers
* Point sampling handler migration at its submodule
* Document the issuer_url identity change for upgraders
Adds an upgrade note covering the one-time re-authorization, fixes the MultiAuth examples that pointed issuer_url at the upstream IdP, and corrects the OCI docstring.
* Address review: valid docstring example, narrower reauth scope
* Scope the reauth checklist item to token-minting providers
* Add require_roles auth check
* Make role docs runnable standalone and fully annotated
* Treat a scalar role claim as one role; correct step-up docs
* Add v4 version badge to require_roles docs
* Use issuer_url for OAuth issuer identity, not base_url
* Apply ruff format to issuer identity tests
* Align ID-JAG audience docstring with issuer_url
* Make InMemoryOAuthProvider keyword-only like its parent
* Keep ID-JAG audience on base_url, out of scope for issuer identity
* Remove stray scratch script
* Make AuthorizationHandler keyword-only
* Bind ID-JAG audience to the issuer identifier
* Fix double slash in issuer_url well-known log hint
* Remove server-initiated sampling and roots from the server API
Deletes fastmcp/server/sampling/, Context.sample/sample_step/list_roots, and
FastMCP(sampling_handler=). The proxy's handshake-era relay now reaches the
front session through the SDK directly.
* Update tests for the removed sampling and roots server API
* Era-gate client.set_logging_level on modern connections
* Document that server-initiated sampling and roots are not in FastMCP 4
* Silence ty deprecation diagnostics and drop stale sampling doc mentions
* Baseline tools-call-sampling; fix removal leftovers flagged by ruff
* Document sampling handlers on both protocol routes; qualify log-level override
* Docs: sampling and roots work on modern via the guard pattern
The imperative ctx.sample()/ctx.list_roots() stay removed, but both
capabilities survive as input-required requests, as tests/conformance
exercises on 2026-07-28. Direct LLM calls remain the recommendation for
generation; roots has no round-trip-budget objection.
* Change register: record the guard route for sampling and roots
* Editorial pass on the sampling and roots docs
* Flag the sampling removal at the top of the page
* Restore the version badge and point sampling users at 3.x
* Keep the sampling conformance scenario live; fix roots example URIs
* Upgrade guide: staying on 3.x is an option for sampling servers
* Elicitation: state the era split once, not twice
Labels are bot-assigned from title/body/code; noting a "suggested" label
in the PR body was a leftover from an unrelated PR (#4392) and doesn't
match how this repo actually labels things.
The imperative ctx.sample()/ctx.list_roots() stay removed, but both
capabilities survive as input-required requests, as tests/conformance
exercises on 2026-07-28. Direct LLM calls remain the recommendation for
generation; roots has no round-trip-budget objection.
A proxy has no back-channel to the real user, so driving a backend ask inside
it failed outright. Surface it as a result for the parent, as ProxyTool does.
Partial fulfillment means two in-flight updates can carry different answers,
so acknowledging the one that loses the update lock stranded the task on a key
the client had already sent.
Prompt and resource asks carry no content, so caching one stored an empty
result and the client never saw the question. Bypass the cache on
continuation legs and return asks unwrapped, as tool calls already did.
Keep the final outstanding input marker until the next task leg is durable,
so a racing tasks/get cannot read a parked leg as complete. Let resources and
resource templates return InputRequiredResult like tools and prompts. Identify
parked requests by their question rather than sort order.
Deletes fastmcp/server/sampling/, Context.sample/sample_step/list_roots, and
FastMCP(sampling_handler=). The proxy's handshake-era relay now reaches the
front session through the SDK directly.
* Honor OAuth application_type in DCR (SEP-837)
* Simplify web redirect check per ruff SIM103
* Enforce application_type over HTTP, at auth time, and tighten native scheme rules
Recover the DCR application_type the SDK RegistrationHandler drops (P1), enforce the stored type on the authorization redirect path (P2), restrict native to loopback http + custom schemes (P2), and document the web/native rules (P2).
* Fix loopback range detection and use a positive scheme allowlist
Classify loopback hosts with ipaddress.is_loopback so all of 127.0.0.0/8 counts (a web client could bypass the non-loopback rule with 127.0.0.2). Replace the NON_REDIRECT_NETWORK_SCHEMES denylist with STANDARD_URI_SCHEMES: native now accepts only https, loopback http, and unregistered private-use schemes per RFC 8252, so smb/smtp/nfs and other unlisted standard schemes no longer pass.
* Vendor the IANA scheme registry and consolidate the loopback classifier
Replace the hand-picked STANDARD_URI_SCHEMES with a vendored snapshot of the IANA URI scheme registry (423 schemes), so registered transports nobody enumerated (coap, coaps, stun, turn, mqtt) fail closed instead of passing as private-use. Delete the stale duplicate _is_loopback_host in oauth_proxy/models.py and reuse the ipaddress-based classifier from redirect_validation, restoring loopback port flexibility across all of 127.0.0.0/8.
* Treat the reserved localhost namespace and absolute host forms as loopback
RFC 6761 6.3 reserves the whole localhost namespace for the local machine, so app.localhost and localhost. are loopback just as much as localhost. Previously a web client could register https://app.localhost/callback and bypass the non-loopback rule, while native clients were wrongly refused legitimate http://app.localhost:3000 dev callbacks. The suffix test is anchored on a leading dot so localhost.evil.com and notlocalhost stay non-loopback.
* Narrow scope: drop native scheme classification, keep the web rule
Registry membership cannot separate app-dispatch schemes from network transports (vscode is registered because it is an app scheme), so classifying a native client's scheme rejected callbacks that real MCP clients need. Remove the vendored registry and the private-use test; native now accepts any scheme outside the unsafe set, with cleartext http still limited to loopback. Also reject web registrations that omit redirect_uris rather than storing an unusable localhost placeholder.
* fix : canonical mime type mapping from formats to remove inconsistency
* Apply ruff format to _get_mime_type
---------
Co-authored-by: Jeremiah Lowin <153965+jlowin@users.noreply.github.com>
* Signal component-level scope shortfalls as insufficient_scope (SEP-2350)
* Fix ty type narrowing in scope step-up test
* Respect check short-circuit when reporting scope shortfall (P2)
* Report union of unmet scopes and document step-up contract
* Aggregate scope shortfall across the AuthMiddleware chain
* Stop chain scope aggregation at the first unevaluated gate
* Unpublish v4 development notes; prep docs for beta 1
* Nest development notes under dev-docs/
* Rewrite site-root links in dev notes as absolute URLs for GitHub rendering
2.1.216 regressed the bubblewrap sandbox the action wraps Bash in when
allowed_non_write_users is set, so every command failed and triage applied
zero labels while reporting success. Also fail the triage job on sandbox
errors, which the existing denial guard could not see.
The example called proxy.with_namespace("remote"), which is not defined
anywhere in the codebase and raises AttributeError. Namespacing a provider
is done via the add_provider() keyword argument.
* Pin burner-redis below the Windows-crashing 0.1.7 release
burner-redis 0.1.7 crashes the interpreter (native fault, no Python traceback)
running the memory:// task backend under pytest-xdist on Windows — reproduced on
GitHub Actions windows-latest via the 'Upgrade checks' workflow, confirmed
absent on macOS/Linux with the identical dependency versions.
pydocket only floors burner-redis at >=0.1.6, so capping pydocket's own version
is not enough: a resolver remains free to pick the newest burner-redis
satisfying that floor. fastmcp-tasks previously pinned pydocket>=0.20.0 with no
upper bound, so a fresh 'pip install fastmcp[tasks]' today can resolve straight
into the broken combination for a real Windows user on the default backend.
Pin burner-redis<0.1.7 directly, which in turn caps pydocket to <0.20.2 (the
last release that doesn't itself require burner-redis>=0.1.7). Verified the pin
holds under both locked and --upgrade (highest) resolution.
* Scope the burner-redis pin to Windows only
burner-redis 0.1.7 is confirmed fine on macOS/Linux (full suite green there with
the identical upgraded dependencies) - only Windows crashes. The previous
unconditional pin blocked every platform from newer pydocket/burner-redis
releases unnecessarily. Add sys_platform == 'win32' to the burner-redis
constraint so only Windows installs are capped.
Verified via uv pip compile --python-platform: macOS/Linux resolve to
burner-redis 0.1.7 / pydocket 0.23.0 (unblocked); Windows resolves to
burner-redis 0.1.6 / pydocket 0.20.1 (still capped).
Server-side completions (@mcp.completion) shipped in #4582 but the
What's New page didn't mention it. Adds it to the authoring-capabilities
cluster with a runnable example and a link to the servers/completions page.
* v4 docs quality pass: fix stale task/era claims, broken links, writing polish
* whats-new: add the client-side protocol negotiation story
The page told the server half of the era story (serves every era) but
never the client half — that a default Client(url) now negotiates the
modern era, where earlier versions pinned the handshake. Completes the
mental model and links to the client negotiation docs.
* Address review: drop 'complete' over-claim; link mounted-state to Session State
* Fix stale Mac/Windows-vs-Linux OAuth key/storage docs
#2223 replaced platform-aware keyring/MemoryStore defaults with
deterministic key derivation and an always-on-disk encrypted store,
but the docs update in that PR missed several spots.
* Fix OIDCProxy doc referring to internal upstream_client_secret name
Codex review: the public OIDCProxy constructor takes client_secret;
upstream_client_secret is only OAuthProxy's internal parameter name.
A task-augmented tools/call returns a CreateTaskResult up through the middleware
chain. Response caching and response limiting assumed a ToolResult and accessed
.content/.wrap(), crashing after the task was already enqueued (a client retry
could duplicate side effects). Both now pass any non-ToolResult through
untouched, alongside the existing InputRequiredToolResult bypass.
Five review fixes. ctx.session_id / get_state / set_state now work in a Docket
worker by falling back to the snapshotted session id. Task management wire calls
(submission, tasks/get/update/cancel) create client spans and propagate trace
context. TasksClientSettings loads .env like DocketSettings, and the docs use
its real env var name. A state-only guard round (request_state, no input
requests) fails with a clear error instead of silently completing wrong.
* Archive v3 docs under /v3 and publish v4 as the primary version
* Label primary docs version v4.0.0 (alpha 1)
* Add What's New in v4 page; fix upgrade-guide phrasing; point banner at What's New
* Rewrite What's New around v4's new capabilities, not the sampling deprecation
* Lead What's New with the SDK v2 engine swap and the SEPs it brings
* State ships now (link Session State); tasks arrive next alpha
* Exclude docs/v3 frozen snapshots from doc-example import validation
Lock in the tasks x stateless-session-state (#4604) integration: a
session: UserSession parameter resolves in a Docket worker via the task-aware
get_server() and the principal restored from the task snapshot, sharing state
across a principal's tasked calls and staying isolated between principals.
The example README pointed at github.com/PrefectHQ/docket (404); the canonical
repo is chrisguidry/docket. Point the docs' Docket-docs link at the canonical
docket.lol.
* Design doc: stateless session state
* Add stateless session-state primitives: Scope, SessionCodec, scoped state
* Add SessionProvider and Session() annotation for stateless session state
* Rewrite session-state design to final shape (Session object, two patterns, no seal)
* Rework stateless session state to final Session/SessionId design
Remove Scope, SessionCodec/sealing, and scoped ctx.get_state. Add the
Session object (get/set/delete/clear over one dict per (principal,
session_id) key), injected session: Session (keyed by principal, requires
auth), session_id: SessionId argument with auto-filled description, and
SessionProvider contributing create_session/end_session.
* Rename injected marker to UserSession; auto-wire SessionProvider on SessionId
* Document stateless session state as a v4 feature
* Require SessionProvider and make sessions create-then-validate
Remove the implicit SessionProvider auto-wiring; a SessionProvider must now be
registered explicitly. create_session records an owned session and get_session
validates the id, rejecting uncreated or foreign ids.
* Add Session.id (public id for session_id sessions, None for UserSession)
* Fix ty: narrow Tool | None and ToolResult.structured_content in session tests
* Fix session-provider enforcement gap for non-local tools; stop embedding raw principal in UserSession key
* Fix disabled session_id tools blocking listing; reject local tools shadowing SessionProvider lifecycle names
* Decouple SessionId description from lifecycle tool name so it survives namespaced mounts
* Remove SessionProvider enforcement; get_session validation is the guarantee
* Fix stale enforcement/key-format docs; document store-owned session TTL
* Dedup SessionId contract description; tighten context.mdx session-state lead
* Make session store/description resolution work in Docket task workers and for partial tools
* Expose get_session as a standalone task-safe function; drop foreground-only Context.get_session
* Move get_session to dependencies alongside the other request accessors
* Reframe context state docs as Request State; cross-request persistence points to Session State
* Address UserSession injection edge cases from review
- inject a UserSession instance (not bare Session) so isinstance holds
- support session: UserSession | None = None (inject None when unauth)
- detect SessionId params past a partial's positional binding
A resumed leg that runs longer than its pointer's wall-clock TTL stranded
_lookup_task on the base leg (false completion / not found). Each poll now
refreshes the routing keys' TTL (sliding expiration), so an actively-polled
task keeps them alive regardless of execution duration, and the resumed-leg
write uses the same buffered TTL as creation. Separately, remote-worker server
resolution now respects the requested tool version, so two versions of the same
mounted tool name resolve to their own child server.
asyncio.wait_for raises asyncio.TimeoutError, a distinct type from the builtin
before Python 3.11, so an elicitation-callback timeout leaked an uncaught type
on 3.10. Convert it to the builtin TimeoutError the rest of the drive raises.
DocketSettings now loads the same dotenv source as core settings, so a
FASTMCP_DOCKET_* value in .env configures the backend instead of silently
using memory://. The root fastmcp publish waits for the matching fastmcp-tasks
to appear on PyPI before uploading, so the [tasks] extra is never installable
but unresolvable. And the example README uses the real worker entry point
(python -m fastmcp_tasks.worker_cli worker).
Three review fixes. A Docket worker may reuse an asyncio context across tasks,
so snapshot restore now always resets auth and headers to the current task's
state — an anonymous task following an authenticated one no longer inherits the
prior caller's identity. A stalled in-task elicitation handler is now bounded by
the call's remaining timeout, like polling and sleeps. And call_tool_task takes
a version= to task a specific component version rather than the highest.
Two remote-worker fixes. A separate worker process cannot reach the submitting
process's server map, so a mounted task's ctx.fastmcp/CurrentFastMCP() fell back
to the root; the worker now re-resolves the owning child from the root using the
snapshotted tool name. And restoring headers no longer fabricates a live Request
— get_http_headers() reads a dedicated task-headers context var while
get_http_request()/CurrentRequest() correctly keep raising inside a task.
The fastmcp[tasks] extra pins fastmcp-tasks=={version}, but no workflow
published it — pip install "fastmcp[tasks]" would fail to resolve. Mirror the
fastmcp-remote workflow: build on release, wait for the matching fastmcp-slim to
appear on PyPI, then publish.
Three review fixes: transparent call_tool(timeout=N) now enforces one deadline
across the whole poll loop (not per-request), matching the sync timeout; the
tools/call interceptor resolves the client-requested component version instead
of the highest; tasks/cancel runs under the per-task update lock and re-resolves
the live leg, so it can't cancel a stale leg while an update enqueues the next.
Server runs over HTTP on the default memory:// backend (no Redis needed); the
client drives it transparently, via an explicit handle, and with a parallel
command that fires several tasks at once to show them overlap. A 1s poll
interval keeps the demo snappy.
A guard task parked on input has an already-COMPLETED Docket execution, so
docket.cancel on it was a no-op: tasks/get reported input_required forever and
tasks/update could still resume it. Record a durable logical-cancellation
marker that tasks/get reports as cancelled and tasks/update refuses to resume,
and clear the parked leg's outstanding requests on cancel.
A queued task can outlive its submitter's token expiry: install the snapshot
token only if still valid, matching the SDK bearer check, so a delayed task
never runs under credentials a live request would reject. ToolTask.wait now
bounds each tasks/get by the remaining deadline so a stalled poll cannot block
past the caller's timeout.
Resolve the error-masking policy via the worker-server resolver instead of
the active Context: a task tool that raises without requesting a ctx param
has no active context, so the old lookup leaked unmasked error text past
mask_error_details=True. Also route custom Tool subclasses through the same
error-conversion wrapper as FunctionTool.
- Client task support is opt-in via importing fastmcp_tasks (drop the core
auto-load of companion packages); a plain Client never advertises tasks.
- A worker restores the submitting caller's auth token and headers from the
task snapshot into the standard ambient context, so get_access_token() /
get_http_headers() work in a distributed worker with no new core hooks.
- worker_cli validates the loaded extension's resolved backend, not env defaults,
so a constructor-configured Redis worker starts.
- Thread the per-call read timeout through task polling; bound ToolTask.wait by
its deadline; set_elicitation_callback rebuilds internal extensions so a
later-set handler answers in-task input.
- README imports TaskConfig from fastmcp.utilities.tasks.
Co-Authored-By: Claude <noreply@anthropic.com>
Server (servers/tasks.mdx) and client (clients/tasks.mdx) docs rewritten for
the extension model: add_extension(TasksExtension()), the guard pattern for
in-task input (no imperative ctx.elicit()), tools-only, and the modern-protocol
requirement (the inverse of the old SEP-1686 legacy-only note). Mechanical
fixes elsewhere for the same reason: telemetry.mdx's tasks/{operation} method
list (get/update/cancel, not result/list), client.mdx's legacy-only feature
list (tasks moved to modern-only) and extension-composition paragraph
(describes the tasks ClientExtension, not the removed notification binding),
and stale SEP-1686 references in the FastMCP 2 upgrade guide. v4-notes status
lines updated to Shipped (#4602, #4603).
- tasks/get|update|cancel now return -32003 when the client did not declare the
tasks extension for the request (SEP-2663 MUST).
- A task tool that raises is a completed task with an is_error result, not a
failed task; failed is reserved for protocol faults, matching a live tools/call.
- A per-task lock serializes concurrent tasks/update so two racing answers cannot
each enqueue a next leg (double execution).
Co-Authored-By: Claude <noreply@anthropic.com>
A FastMCP client now transparently completes tasked tools/call: the tasks
ClientExtension advertises the capability and claims the CreateTaskResult, and
the resolver drives the tasks/get poll loop to completion, answering in-task
input through the client's elicitation handler and returning the tool's real
result. call_tool is transparent, call_tool_mcp exposes the raw result, and
call_tool_task yields a Task handle. The client half moves to fastmcp-tasks;
the [tasks] client extension auto-wires into Client (ProxyClient opts out).
Co-Authored-By: Claude <noreply@anthropic.com>
A task tool that returns InputRequiredResult now ends its leg (freeing the
worker) and stores the ask as durable state; tasks/update enqueues a fresh
Docket execution (the next leg) with accumulated request_state/input_responses
injected via ctx. No worker ever blocks on input, so a parked task no longer
holds up shutdown. Imperative ctx.elicit() inside a task is removed and raises
with guard-pattern guidance.
Co-Authored-By: Claude <noreply@anthropic.com>
Widen the tools/call result serialization (via a refcounted, modern-gated wrap
installed by TasksExtension) so a CreateTaskResult reaches the client instead of
being stripped by the CallToolResult|InputRequiredResult surface — the SDK ships
claim consumption but no production. Emit the resultType discriminator the
protocol requires (task on CreateTaskResult, complete on the tasks/* results);
the draft schema forbids it (additionalProperties:false), a contradiction
reported upstream. Closes compliance gaps G1/G4/G5.
Co-Authored-By: Claude <noreply@anthropic.com>
TasksExtension serves io.modelcontextprotocol/tasks on the extension API:
a decide-and-task tools/call interceptor (era-gated to modern connections),
tasks/get with inlined results and inputRequests, tasks/update delivering
poll-based in-task elicitation, tasks/cancel, durable creation, and
auth-scoped task isolation. Wire models validate against the vendored
ext-tasks schema. Worker-side Context hooks are refcounted so sibling
servers cannot strand each other's workers.
Co-Authored-By: Claude <noreply@anthropic.com>
Engine modules (keys, context snapshot, docket lifespan, worker CLI,
client handles) move intact; SEP-1686 wire modules park in _legacy_wire
for adaptation to SEP-2663. Core keeps task=True declaration on tools
only and raises at serve time until the tasks extension is registered.
Co-Authored-By: Claude <noreply@anthropic.com>
* Fix percent-encoded skill file names unreadable in resources mode
Encode supporting-file paths explicitly (quote/unquote) when building
and resolving skill:// resource URIs, instead of relying on AnyUrl's
implicit encoding. This also closes the ambiguity where a file literally
named "setup%20guide.md" would collide with "setup guide.md" once both
were percent-encoded.
Fixes#4545
* Quote main_file_name when building its resource URI
Keeps the main-file URI on the same explicit quote/unquote round-trip
as supporting files, so a custom main_file_name containing a literal
'%' still resolves after the shared unquote() in _get_resource().
Preserve explicit suffixes when building data-backed File resource URIs, while keeping the inferred-extension fallback for names without a suffix.
Closes#4530
Co-authored-by: chatgpt-codex-connector[bot] <199175422+chatgpt-codex-connector[bot]@users.noreply.github.com>
* Add M2M client credentials auth providers
Wrap the SDK's client_credentials and private_key_jwt OAuth providers as
FastMCP-idiomatic ClientCredentialsOAuthProvider and PrivateKeyJWTOAuthProvider,
enabling browser-free client authentication via Client(auth=...).
* Fix M2M token cache collision and explicit-scope drop
Namespace the token cache by client_id so distinct clients sharing one store don't overwrite each other's tokens; pin caller-supplied scopes so the token request keeps them; fix CodeQL URL-substring check in tests; drop unused logger.
* Preserve step-up scope union, scope-aware token cache, restore token expiry
Only pin the caller's explicit scopes on initial authorization, leaving the SDK's step-up scope union intact; namespace the token cache by requested scopes as well as client_id; restore persisted absolute expiry on init so an expired stored token is re-fetched.
* Skip expiry restore for non-expiring reloaded tokens
* Distinguish expires_in=0 from omitted when restoring expiry
* Scope step-up flag to the flow via ContextVar; runnable JWT signing example
* Add server-side argument completion (@mcp.completion)
* Reference CompletionValues directly in cast so the import reads as used
* Import completion types from mcp_types, not the fastmcp.types mirror
* Fix test imports after dropping the fastmcp.types mirror
* Fix change-register example import after dropping the types mirror
* Enforce 100-value completion cap; make docs example runnable
* Document completion authorization contract
* Offload sync completion handlers to threadpool
* Exclude bare str from completion return type
* Pass Any-typed value in bare-string rejection test
* Point completion authoring types to mcp_types in v4 notes
* Trim fastmcp.types to FastMCP-unique types only
fastmcp.types re-exported 29 mcp_types symbols verbatim, which was
pointless indirection users had to discover. It now holds only Textarea,
the one type FastMCP actually defines; everything else imports from
mcp_types directly. These mirrors were added during unreleased SDK v2
migration work and never shipped, so this is not a breaking change.
* Keep historical mcp.types import in v2/v3 migration examples
fastmcp.types re-exported 29 mcp_types symbols verbatim, which was
pointless indirection users had to discover. It now holds only Textarea,
the one type FastMCP actually defines; everything else imports from
mcp_types directly. These mirrors were added during unreleased SDK v2
migration work and never shipped, so this is not a breaking change.
* Restore upgraded dependency checks
Generated with Codex
* Clarify settings loading and teardown logging
Generated with Codex
* Align ty checks on the upgraded version
Generated with Codex
* Preserve simultaneous caller cancellation
Generated with Codex
* Delegate forked client protocol helpers to the SDK
Replace FastMCP's copies of _fold_extensions, _evicting_message_handler,
and _synthesize_discover with imports from mcp.client.client. The fork had
drifted: it was missing validate_extension_identifier, so non-reverse-DNS
extension identifiers were silently accepted.
Full lifecycle composition over mcp.Client stays blocked upstream —
mcp.Client hardcodes ClientSession construction (no session_class hook)
and forbids reentry.
* Drop duplicate local helpers reintroduced by the merge; use SDK versions
Resource/prompt error detail and proxy instructions/connection-error
surfacing now work on the modern protocol era, so the tests pinned to
mode="legacy" with a TODO(defect)/TODO(mode="legacy" pin) marker run
on the default auto mode again.
Three defects hidden by tests pinned to the handshake era, where a raw
exception reaches the wire as str(exc). At 2026-07-28 the runner masks
anything that is not an MCPError/ValidationError as "Internal server error".
- _on_read_resource / _on_get_prompt now translate FastMCPError through
to_mcp_error, mirroring _on_call_tool. Masking is unchanged.
- FastMCPProxy registers a server/discover handler so upstream instructions
reach modern clients; on_initialize only fires for the handshake.
- ProxyProvider's list methods normalize transport failures into MCPError.
Removes pins added while making the auto-default suite pass that weren't
actually testing older-protocol-only behavior, and keeps (with a stated
reason) the ones that are. Along the way, fixes two real defects the audit
surfaced in the modern protocol path: PingMiddleware could leak a
_active_sessions entry when a connection's exit_stack closed before its
keepalive task got its first scheduler turn, and FastMCP(experimental_
capabilities=...) was silently dropped from server/discover responses
(it only ever reached the legacy initialize handshake).
Most pins in tests/server/providers/proxy/ were added only to keep tests
green while unpinning changed which protocol era the proxy's backend
connection used, not because the test's subject cared about the era. With
proxy era-mirroring (#4573) landed, a front client on auto correctly moves
the whole chain to the modern protocol, so plain tool/resource/prompt calls
through a proxy no longer need a pin.
Kept pins fall into three buckets, each commented at the call site: tests
whose subject is genuinely handshake-only (sampling, roots, elicitation
push-forwarding, ping, initialize handshake mechanics); tests whose backend
is a directly-constructed ProxyClient/StatefulProxyClient, which always
defaults to legacy independent of the front era; and two tests left pinned
with a TODO documenting a real defect this audit surfaced (upstream
instructions not forwarded to a modern-era client through a proxy, and
ProxyProvider.list_tools leaking an unwrapped connection error instead of
an MCPError).
Remove 4 unjustified pins (proxy header passthrough, connect timeout,
two response_title validation tests that fail before any request is
dispatched). Keep 81 pins that genuinely exercise older-protocol-only
behavior (ctx.elicit back-channel, sampling, roots, ping, session IDs,
initialize handshake, client.set_logging_level).
Flags a real defect: _on_read_resource/_on_get_prompt only catch
(DisabledError, NotFoundError), unlike _on_call_tool which catches
FastMCPError broadly. A ResourceError/PromptError escapes as a raw
exception and the modern protocol's generic exception ladder masks it
as "Internal server error", losing the detailed message tool errors
still get. Left pinned with a TODO in test_client.py and
test_error_handling.py rather than hidden.
* Mirror front protocol era onto proxy backend connection
A proxy created from a non-Client target now negotiates, on its backend,
whatever era its front client negotiated, instead of pinning one era.
Explicit create_proxy(mode=...) still overrides. Guards the eager backend
initialize() so an explicit modern pin behind a handshake front no longer
crashes.
* Carry the mirrored proxy era into multi-server config backends
A multi-server MCPConfig target mounts one proxy per configured server on a
composite router, so setting the era on the outer client stopped at the router
and every real backend stayed on its default era. TransportOptions.backend_mode
carries it down, resolved per request alongside the outer mirroring.
The router is also sealed under a policy held on the transport rather than a
fresh per-router ephemeral key, so a guard tool's request_state survives the
router being rebuilt between rounds.
* Fix: gather() eagerly creates coroutines before scheduling them
AggregateProvider fans out Provider.get_tool() (and sibling calls) across
child providers via gather(*[p.get_tool(x) for p in providers]). The list
comprehension builds every coroutine up front, then gather()'s scheduling
loop hands them to an anyio task group one at a time. If that loop is
interrupted partway through - e.g. by pytest-timeout's SIGALRM-based
per-test timeout, which can fire between any two bytecode instructions,
unlike normal async cancellation - any coroutine not yet scheduled is
abandoned and silently garbage collected later, producing a "coroutine
'Provider.get_tool' was never awaited" warning attributed to whatever
unrelated test happens to be running when the GC gets to it.
Change gather() to take a single iterable consumed lazily, one awaitable
at a time, right before each is scheduled, and close any awaitable that
was just retrieved if scheduling it raises. Update call sites to pass
generator expressions instead of eagerly-built lists so coroutine
creation and scheduling stay tightly coupled.
* Close unscheduled awaitables from eager callers; make get_tasks lazy
The guard failed on any denial, so an agent falling back to an unlisted tool
during a GitHub outage tripped it — and the error blamed the allowlist, which
was intact. It now fails only when a command the workflow actually grants is
refused, which is the signal that a pattern was mangled.
mcp__github__get_pull_request was never granted, so on a PR the agent could
only read via get_issue and reached for denied fallbacks when that failed.
Client negotiates the newest mutual protocol era by default (probe
server/discover, fall back to the initialize handshake). ProxyClient and the
inspect utility explicitly pin the handshake era so proxy forwarding and
server_info reads are unchanged. SSE and multi-server config transports are
legacy-only. extensions= and result_claims= (SEP-2133) are thin passthroughs
to the SDK session.
* Make the SDK seam the root of FastMCP middleware dispatch (D3)
Notifications, cancellations, and malformed/unroutable messages now reach
on_message/on_request/on_notification at the SDK seam. Component methods keep
their interior dispatch (typed hooks, tool-exception visibility) unchanged; the
seam covers only messages the interior never dispatches, so each hook fires once.
* Document the middleware seam coverage and suspend semantics (D3)
* Align seam docs and ask-visibility test with the result-cycle MRTR model
An InputRequiredResult is the full result of a complete request->response
cycle, not a suspension: component hooks observe an asking round's
InputRequiredToolResult as an ordinary return value.
* Replace 'seam' language with plain dispatch terminology
* Keep the raw middleware __call__ signature; forward middleware message edits
* Cover fires-once across an MRTR continuation round
* Align cherry-picked coverage test with renamed recorder
* Rewrite only the message, never the dispatch destination
* Add guard-mode MRTR server support (SEP-2322)
* Add server-side MRTR guard tests
* Add MRTR guard docs, exports, and output-schema handling
* Apply formatting to MRTR guard changes
* Fix MRTR review round 1: middleware-safe suspend, Annotated strip, stable audience
- ToolInputRequired subclasses BaseException (CancelledError precedent) so
error middleware's broad except Exception cannot swallow a suspension
- Strip InputRequiredResult arms inside Annotated return types
- Reject a custom RequestStateSecurity without a stable audience (random
per-replica server names would break shared-key verification)
* Fix static analysis: rewrite tuple([...]) as tuple literal (C409)
* Recognize InputRequiredResult inside Annotated union arms
_is_input_required_type now peels Annotated first, so a metadata-carrying
guard arm (str | Annotated[InputRequiredResult, Field(...)]) is stripped
and the data arm's output schema survives.
* docs: frame multi-round tools as elicitation on the modern protocol
Fold multi-round-tools.mdx into elicitation.mdx as two eras of one
capability; drop pause/suspend framing for the stateless per-round model.
* Transport MRTR asks as InputRequiredToolResult, not a raised signal
An input-required result is the full result of a stateless MRTR leg, so it
flows through the middleware chain as an ordinary ToolResult subclass instead
of a raised ToolInputRequired(BaseException). Middleware observes it, caching
skips it, and response-limiting leaves it untouched.
* Document MRTR middleware interaction and the isinstance pattern
* Update MRTR change-register verify note to InputRequiredToolResult
* Align test module docstring with result-cycle framing
* Fix MRTR review: bypass cache on continuation legs; soften audience guard
- ResponseCachingMiddleware skips read AND write on continuation legs:
the cache key is name+arguments only, so a continuation's final result
would be served to later fresh calls, which would never be asked
- The stable-audience check is a warning, not an error: a policy object
cannot reveal whether its keys are shared, and single-process
customization (ephemeral ttl, custom codec) is legitimate unnamed
* Treat state-only rounds as continuations in the response cache
A round carrying request_state but no questions retries with
input_responses=None; request_state alone must bypass the cache or its
terminal result is stored under the fresh-call key.
* Fix MRTR review round: preserve asks through transforms, empty-name audience, docs predicate
- TransformedTool.run returns an InputRequiredToolResult intact instead of
reshaping it into an empty ToolResult for non-object output schemas
- audience warning uses a falsy-name check (empty string also autogenerates
a per-replica name)
- the elicitation docs continuation predicate checks request_state too
* Add create_proxy(mode=) opt-in for guard round-tripping through proxies
An auto-created proxy client stays handshake-era by default (a dual-era
backend serves both, and one proxy session is one era; handshake preserves
server-initiated push forwarding). Pass create_proxy(target, mode="auto")
to negotiate modern so an upstream guard's InputRequiredResult round-trips —
the two are mutually exclusive per session.
* Wrap raw InputRequiredResult returned by a transform_fn
A custom transform function may return the raw ask directly, like any tool
body — wrap it into InputRequiredToolResult so it survives output
normalization and reaches the wire, not only pre-wrapped forwarded guards.
* Reject input-required results from background tasks
* Unwrap type aliases before stripping guard arms
* Apply ruff format
* Recursively strip guard arms through nested and composed aliases
* Reflect MRTR continuation fields on the middleware message
* Suppress output schema for InputRequiredResult subclasses
* Forward progress on modern proxy tool calls
* Suppress output schema for bare aliased guard returns
* Suppress output schema for any surviving guard return wrapping
* docs: align server component docs
Generated with Codex.
* docs: clarify resource return shapes
Generated with Codex.
* docs: clarify initialize middleware response
Generated with Codex.
* docs: lead visibility filtering with names, scope keys to version targeting
* docs: correct initialize result semantics, template mime type, docket scope, visibility tip
* Warn when a visibility key omits the @ version delimiter
* Honor a resource template's declared mime_type and meta on read
* Strip internal visibility meta from resource content; document filter intersection
---------
Co-authored-by: Jeremiah Lowin <153965+jlowin@users.noreply.github.com>
* Add OAuthProxy issuer response parameter
* Cover OAuthProxy issuer error redirects
* Relax host origin guard defaults (#4439)
* Use exact issuer in authorize errors
* Restore HTTP host guard compatibility (#4472)
* Hugging Face Auth Integration (#4385)
Co-authored-by: Jeremiah Lowin <153965+jlowin@users.noreply.github.com>
* Docs: add v3.4.4 changelog entries (#4473)
* Explain unnormalized issuer; cover consent-denial path base_url
* Revert "Merge remote-tracking branch 'origin/release/3.x' into codex/oauth-proxy-rfc9207-issuer"
This reverts commit 9e34b1686c, reversing
changes made to 640dc60fe0.
* Preserve callback query bytes when appending iss/code/state params
add_query_params previously decoded the existing query with parse_qsl
and re-encoded it, mutating opaque or signed query strings (a valueless
?flag became ?flag=, non-UTF-8 percent-encoded bytes got replaced).
Append the newly-encoded params to the existing query string instead of
round-tripping it through parse/encode.
Also fixes a stray bare `httpx` reference in a test that should use
httpx2 following the SDK v2 migration.
* Attach RFC 9207 iss to authorize() success redirects too
AuthorizationHandler only added iss to error redirects from the SDK's
base handler, not to code redirects returned directly by authorize()
overrides that bypass consent/upstream (as GitHub's mocked test does).
Since metadata now unconditionally advertises
authorization_response_iss_parameter_supported, any client-facing
redirect missing iss hard-fails RFC 9207-aware clients.
Also fixes HeadlessOAuth, which parsed code/state from the redirect
but silently dropped iss, so the same regression would have masked
itself across every other provider integration test too.
* Carry RFC 9207 iss through the production OAuth callback path
OAuthProxy advertises authorization_response_iss_parameter_supported and
sends iss on every authorization redirect, but the client's production
callback chain (CallbackResponse -> OAuthCallbackResult -> OAuth.callback_handler)
had no iss field, so it was silently dropped and the SDK's
validate_authorization_response_iss rejected the callback. HeadlessOAuth
already carried iss through, which is why CI stayed green while real
clients failed.
Add iss to CallbackResponse and OAuthCallbackResult, thread it through
store_result_once for both success and error branches, and pass it into
AuthorizationCodeResult in OAuth.callback_handler.
* Don't duplicate iss when a provider redirect already carries one
* Consolidate RFC 9207 iss handling into a single redirect helper
Every client-facing authorization redirect must carry exactly one iss.
That invariant was being enforced by hand at five separate call sites,
each building its own params dict -- which is how the success-redirect
path shipped without iss in the first place, and how a registered
redirect_uri that already carries its own iss could end up duplicated.
Route all five sites through build_client_redirect(), which owns the
idempotent replace-or-append behavior so no caller can get it wrong.
---------
Co-authored-by: shaun smith <1936278+evalstate@users.noreply.github.com>
* Reapply span attributes after creation to survive non-forwarding samplers
Tracer.start_span builds the span from sampling_result.attributes, not
the attributes kwarg — a custom Sampler that returns
SamplingResult(RECORD_AND_SAMPLE) without forwarding attributes
silently drops everything FastMCP passed at creation time. Reapply the
same attributes immediately after span creation (guarded by
is_recording()) so on_start hooks and samplers still see them, while
the finished span is guaranteed to carry FastMCP's telemetry
regardless of sampler behavior.
* Restore only missing span attributes, not a blanket reapply
Reapplying all attributes after span creation overwrote values a
sampler deliberately set (e.g. a redacted mcp.method.name) and
inflated dropped-attribute counts when the SDK's attribute limit was
hit. Compare against the span's existing attributes and restore only
the keys a non-forwarding sampler actually dropped, via a shared
restore_missing_attributes() helper in fastmcp.telemetry.
* Gate attribute restore on all-or-nothing, not per-key
Restoring only missing keys reinserted attributes the SDK's bounded
attribute map had already evicted under a low
OTEL_SPAN_ATTRIBUTE_COUNT_LIMIT, evicting a different retained key and
inflating dropped_attributes beyond what the sampler actually dropped.
Gate on none of our attributes being present (plus dropped_attributes
== 0) instead — the regression this exists to fix is a sampler
dropping everything, and eviction under a limit always leaves some.
Renamed restore_missing_attributes to restore_dropped_attributes to
match.
* Gate attribute restore on empty span, not per-key presence
A sampler that intentionally supplies only its own attributes (e.g. to
strip component names or resource URIs for privacy/cardinality
control) left none of FastMCP's keys on the span, so the previous
all-or-nothing gate treated it identically to a bare non-forwarding
sampler and restored everything, defeating the filter. Key off the
span having no attributes at all instead — a bare sampler leaves it
empty, a filtering sampler doesn't.
* Add FASTMCP_SSRF_TRUST_PROXY to allow SSRF fetches through a corporate proxy
🤖 Generated with Claude Code
* Make SSRF fetch client trust_env explicit for proxy routing
🤖 Generated with Claude Code
* Warn when SSRF proxy trust is enabled without a configured proxy
🤖 Generated with Claude Code
* Warn when NO_PROXY would send an SSRF-trust-proxy fetch direct
🤖 Generated with Claude Code
* Refuse SSRF-trust-proxy fetches when no proxy would route the target
🤖 Generated with Claude Code
* Fix TestProxyMode mocks to patch httpx2.AsyncClient
main's httpx -> httpx2 migration (#4503) landed after these tests were
written; ssrf.py's fetch path already uses httpx2.AsyncClient, but
TestProxyMode still patched the old httpx module, so the mock silently
stopped intercepting and requests escaped to the real network.
* Fix port-qualified NO_PROXY bypass in SSRF proxy-trust guard
proxy_bypass(hostname) discarded the port, so a NO_PROXY entry like
127.0.0.1:8443 went undetected while httpx2 honored it and sent the
request direct with the blocklist already disabled. Pass host:port
instead, except for IPv6 literals, where httpx2 ignores port when
matching NO_PROXY and neither bracketed nor unbracketed host:port
reliably matches through proxy_bypass()'s own parser.
* Replace NO_PROXY prediction with explicit proxy control in SSRF trust-proxy mode
Predicting httpx2's proxy routing (via proxy_bypass(), then via httpx2's own
get_environment_proxies()/URLPattern internals) kept diverging from its real
NO_PROXY handling — three rounds, three different divergences, always in the
unsafe direction. Read HTTPS_PROXY/ALL_PROXY directly and pass it to httpx2
explicitly with trust_env=False, so the request provably goes through that
proxy instead of being predicted to. NO_PROXY is no longer evaluated in this
mode: a NO_PROXY'd host is now routed through the proxy rather than refused,
since that's strictly safer than the alternative (direct with the blocklist
already off).
---------
Co-authored-by: Jeremiah Lowin <153965+jlowin@users.noreply.github.com>
* Fix typos
* Document Cachable* -> Cacheable* rename as v4 breaking change
Adds the response-cache model rename to the change register, per
maintainer decision to skip compatibility aliases in favor of clear
documentation.
* Skip invalid Before import in doc test; use inline codespell ignore
The Cachable* -> Cacheable* breaking-change entry showed the old,
now-invalid import for contrast, which the doc-example test picked up
as a real import and flagged as a regression. Comment out the
deliberately-broken "Before" line (matching the McpError entry just
above it) so only the working "After" import is exercised.
Also swap the blanket codespell ignore-words-list entry for a
narrower inline `codespell:ignore` directive on the one line that
needs it, so codespell keeps flagging "cachable" everywhere else.
---------
Co-authored-by: Jeremiah Lowin <153965+jlowin@users.noreply.github.com>
* Add subject field to AccessToken initialization
Fixes#4266
Add `subject` property to the AccessToken.
```python
return AccessToken(
token=access_token_as_dict["token"],
client_id=access_token_as_dict["client_id"],
scopes=access_token_as_dict["scopes"],
subject=access_token_as_dict["subject"],
# Optional fields
expires_at=access_token_as_dict.get("expires_at"),
resource=access_token_as_dict.get("resource"),
claims=access_token_as_dict.get("claims") or {},
)
```
* Populate AccessToken.subject across all token verifiers
Closes#4266. get_access_token().subject was always None: the SDK's
AccessToken.subject wasn't carried into FastMCP's AccessToken by the
dependency-layer conversion, and none of the built-in TokenVerifiers
(JWT, introspection, and the OAuth-provider verifiers for Discord,
Clerk, Google, WorkOS, HuggingFace, GitHub, and Cognito) populated it
from the sub claim/field they already extract.
---------
Co-authored-by: Jeremiah Lowin <153965+jlowin@users.noreply.github.com>
* Add Auth0MCPProvider for Auth0 Auth for MCP
* Document Auth0 MCP provider integration
* Add Auth0MCPProvider scope and auth rejection tests
Cover permissions-based required_scopes enforcement and unauthenticated MCP 401 responses.
* fixed documentation
* Narrow Auth0 docs to integration guide only
* Fix Auth0 provider: use httpx2 instead of httpx
httpx is a dev-only transitive dependency in this repo; runtime installs
declare httpx2 exclusively. The module-level 'import httpx' in auth0.py
broke import on a clean install of fastmcp or fastmcp-slim[server].
---------
Co-authored-by: Jeremiah Lowin <153965+jlowin@users.noreply.github.com>
#4466 added examples/ to [tool.ty.src] include, but two filters still
excluded it: the ty prek hook's files: scope and the static-analysis
workflow's push path triggers. An examples-only commit or direct push
to main could skip the gate entirely.
* Document icon theme support and add round-trip tests
MCP SDK v2 added a `theme` field to `Icon` (light/dark), letting a
server ship complementary icon variants for clients that render in
different UI themes. Document the field on the icons page and cover
it with round-trip tests through the server/client protocol.
* docs: add 4.0.0 version badge to Theme Variants section
* Add server-side SEP-990 identity assertion (ID-JAG)
* Test SEP-990 identity assertion token endpoint
* Format identity assertion test
* Document SEP-990 identity assertion
* Thread identity_assertion through OIDCProxy
* Harden ID-JAG: authoritative scopes and grant-type enforcement
Scopes for the issued token now derive only from the signed assertion (or server policy when it omits scopes); the client-supplied request scope may narrow but never widen them. Enforce the registered grant-type constraint the SDK check bypassed, and have proxy DCR add the jwt-bearer grant to registered clients when identity assertion is enabled.
* Harden ID-JAG: honor nbf, reject non-object payload, bound jti cache, preserve required_scopes
* Document per-process ID-JAG replay limitation and nbf check
* Harden ID-JAG round 3: resource indicator, non-object header, algorithm config
- Honor RFC 8707 resource on the jwt-bearer grant (invalid_target on
mismatch), mirroring authorize()'s invariant incl. skip-when-unconfigured
- Reject JSON-array JOSE headers with invalid_grant instead of a 500
- Add IdentityAssertion.algorithm so ES256/PS256 issuers can be verified
(JWTVerifier otherwise defaults to RS256)
* Bind ID-JAG exchange to the assertion's signed client_id and resource
SEP-990: the IdP signs which client and which resource the assertion was
minted for. With public proxy clients the presented client_id is
self-asserted, so the signed binding is what stops client B redeeming
client A's leaked assertion — and the signed resource claim stops an
assertion for server A being redeemed at server B behind the same IdP.
* Harden ID-JAG round 4: check bindings before jti consumption; validate temporal claims, algorithm, and discovery body
- Move the client_id/resource binding checks into the validator itself,
before jti is recorded as consumed, so an assertion presented with the
wrong binding is rejected without burning replay protection for whoever
it actually belongs to
- Reject non-numeric exp/iat/nbf with invalid_grant instead of a 500
- Validate IdentityAssertion.algorithm at config time (must be an
asymmetric JWS algorithm verifiable via JWKS)
- Reject a non-object OIDC discovery body with invalid_grant instead of a 500
- Centralize the resource-URL comparison helpers used by both the
validator and OAuthProxy.authorize()
* Rebase onto httpx2/SDK b2 and harden ID-JAG round 5
- Migrate identity assertion + tests to httpx2 and the local httpx2_mock
(legacy httpx is now banned; pytest-httpx no longer intercepts)
- Add is_optional to the shared httpx2_mock, mirroring pytest-httpx
- Tighten the algorithm allowlist to JWTVerifier's exact supported set
(prefix check accepted typos like RS999 -> 500 on first exchange)
- Reject non-string jti before the cache lookup (unhashable -> 500)
- Track revocation for self-contained ID-JAG tokens: revoke_token records
the jti and load_access_token rejects it until natural expiry
- Dedupe resource-URL helpers: proxy now imports the shared
normalize_resource_url/server_url_has_query from identity_assertion
* Advertise 'none' token-endpoint auth method when ID-JAG is enabled without CIMD
DCR clients are public, so metadata consumers must see 'none' to use the
advertised jwt-bearer grant; previously only the CIMD path added it.
* Document 2026-07-28 protocol support as a distinct feature catalog
SEP-990 identity assertion leads: the SDK provides the wire contract and
provider hook; FastMCP provides the complete server-side implementation.
Inventories the full modern-era capability set for v4.
* Harden ID-JAG round 6: lazy re-export, dual-form audience, per-issuer algorithms, discovery backoff
- IdentityAssertion re-exported lazily from server.auth (the eager import
bypassed the package's documented lazy-import boundary)
- Accept the ID-JAG aud both with and without base_url's trailing slash;
metadata advertises the slashed form, so IdPs echoing it verbatim work
- algorithms={issuer: alg} per-issuer override, mirroring jwks_uris
- OIDC discovery serializes per-issuer and backs off 30s after a failure
(discovery runs pre-signature, so garbage could amplify into HTTP floods)
* Add examples/ to ty static-analysis gate
* Fix example type errors and stale SDK idioms for ty
* Use typing_extensions.TypedDict for the quiz tool-param type
Question is a take_quiz parameter, so FastMCP builds a Pydantic schema
for it; typing.TypedDict raises PydanticUserError on Python 3.10/3.11
(only 3.12+ accepts it). ty and 3.12 runs miss this, so it slipped in.
* Guard get_access_token() None case in huggingface_oauth example
Caught by the ty gate this PR adds: the example, merged separately,
had never been type-checked against examples/. Matches the existing
aws_oauth/keycloak_oauth pattern.
* Print actual YAML text in custom serializer example
* Improve DescopeProvider scope discovery and well-known URL support
Co-authored-by: Cursor <cursoragent@cursor.com>
* Simplify DescopeProvider scope and URL handling
Co-authored-by: Cursor <cursoragent@cursor.com>
* Make DescopeProvider scope discovery async and lazy
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
🤖 Generated with Claude Code
* Use generic scope in Descope tests
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
🤖 Generated with Claude Code
* Address Descope discovery edge cases
* Deduplicate Descope metadata fallback
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Co-authored-by: Jeremiah Lowin <153965+jlowin@users.noreply.github.com>
* Add ResourceSecurity screening for templated resources (defaults on)
* Add tests for resource path-security screening
* Document resource path-security; fix ty in tests
* Carry child template security policy through provider mount
Preserve a mounted template's explicit ResourceSecurity (per-param
exemptions or a deliberate opt-out) through FastMCPProviderResourceTemplate.wrap
so the parent read chokepoint honours it instead of the parent default.
* Defer mcp SDK import so fastmcp.resources loads without the [mcp] extra
* Make resource path-security docs examples self-contained and runnable
* Match exempt_params under both hyphen and underscore spellings
Template placeholders like {git-ref} extract as git_ref, so an exemption
written with the natural URI-template spelling never matched.
* Docs: describe net-depth traversal rule accurately; make example runnable
The screening only rejects .. segments that escape the starting depth
(foo/../bar passes) — saying any standalone .. is rejected overstated
the guarantee. Also define DOCS_ROOT so the example runs.
* Turn OpenTelemetry instrumentation on by default with explicit off-switch
Add FASTMCP_ENABLE_TELEMETRY setting (default true) and mcp.protocol.version
span attribute for SDK parity.
* Make disabled telemetry a transparent pass-through, not a NoOpTracer
The stock NoOpTracer.start_as_current_span attaches a NonRecordingSpan, hijacking the current OTel context from any enclosing application span. When telemetry is disabled, get_tracer() now returns a non-attaching pass-through tracer so trace.get_current_span() inside handlers still resolves to the caller's span.
* Add regression test: HTTP lifespan fires once per process across sessions
* Drop redundant enter-count assertion at teardown (CodeQL)
* Assert session-manager lifespan entry directly, not user-lifespan count
Replaces FastMCP.as_proxy() helper calls with create_proxy(), rewrites the
mount() as_proxy=/prefix= kwarg tests to plain mount() (the params are gone),
and deletes deprecation-only tests for as_proxy() and remove_tool().
Tier 2 aggressive-window removal: the parameter was a deprecated no-op
on the streamable-HTTP transport (the SDK v2 client no longer supports
it). SSETransport still accepts sse_read_timeout.
Tier 2 aggressive-window removal: these shims were deprecated in 3.2,
a shorter deprecation window than the usual policy. Canonical imports
are fastmcp.apps / fastmcp.FastMCPApp.
The era-gate blocked every ctx.sample/sample_step on a 2026-07-28
connection, but a server-configured sampling handler answers server-side
without the client back-channel. Gate only when the request would hit the
removed client path; force the handler path (client_available=False) on
modern so "fallback" goes straight to the handler instead of a bare
client-attempt failure.