Commit graph

2,968 commits

Author SHA1 Message Date
Bill Easton
8e1f662d93
Bump py-key-value-aio to >=0.4.0,<0.5.0 (#3143)
Co-authored-by: Bill Easton <strawgate@users.noreply.github.com>
Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
Co-authored-by: marvin-context-protocol[bot] <225465937+marvin-context-protocol[bot]@users.noreply.github.com>
2026-02-11 10:03:45 -05:00
Chris Guidry
efcc12bc76
Merge pull request #3136 from jlowin/task-elicitation-relay
Relay task elicitation through standard MCP protocol
2026-02-10 16:06:37 -05:00
marvin-context-protocol[bot]
a53030a806 chore: Update SDK documentation 2026-02-10 21:00:34 +00:00
Chris Guidry
6e22914360 Move relay logic to elicitation.py, fix related-task metadata key
Moves relay_elicitation() into elicitation.py so it can reuse
handle_task_input() for the Redis push instead of duplicating that logic.
notifications.py just detects the trigger and calls it.

Also fixes the related-task metadata key from modelcontextprotocol.io/ to
io.modelcontextprotocol/ to match the current spec:
https://modelcontextprotocol.io/specification/2025-11-25/basic/utilities/tasks

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-10 16:00:01 -05:00
Chris Guidry
a95fed3041
Merge branch 'main' into task-elicitation-relay 2026-02-10 15:41:25 -05:00
marvin-context-protocol[bot]
95b4271b3b chore: Update SDK documentation 2026-02-10 20:36:17 +00:00
Chris Guidry
aa4db3d00e Address review: set status key on cancel fallback, prevent relay task GC
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-10 15:35:50 -05:00
Chris Guidry
95816c5ae2
Merge pull request #2906 from gfortaine/fix/statusMessage-not-forwarded
feat: distributed notification queue + BLPOP elicitation for background tasks
2026-02-10 15:32:55 -05:00
Chris Guidry
361eb08f42 Relay task elicitation through standard MCP protocol
When a background task calls ctx.elicit(), the notification subscriber now
detects the input_required notification and sends a standard elicitation/create
request to the client via session.elicit(). The client's elicitation_handler
fires, and the relay pushes the response to Redis for the blocked worker.

This means clients can respond to background task elicitation using the same
elicitation_handler they'd use for any other elicitation — no need to interact
with Redis or call handle_task_input() directly.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-10 14:58:19 -05:00
Chris Guidry
3809f0c77b
Merge branch 'main' into fix/statusMessage-not-forwarded 2026-02-10 14:22:58 -05:00
Jeremiah Lowin
1d0c0adeab
Add validate_output option for OpenAPI tools (#3134)
Co-authored-by: marvin-context-protocol[bot] <225465937+marvin-context-protocol[bot]@users.noreply.github.com>
2026-02-10 13:29:45 -05:00
Jeremiah Lowin
a715176499
Fix unhandled exceptions in OpenAPI POST tool calls (#3133)
Co-authored-by: marvin-context-protocol[bot] <225465937+marvin-context-protocol[bot]@users.noreply.github.com>
2026-02-10 13:13:45 -05:00
Jeremiah Lowin
40e80d60e5
Fix session visibility marks leaking across sessions (#3132) 2026-02-10 11:42:43 -05:00
Chris Guidry
f7c47f498c
Merge branch 'main' into fix/statusMessage-not-forwarded 2026-02-10 11:34:05 -05:00
Bill Easton
cdad99583e
Fix Windows test timeouts in OAuth proxy provider tests (#3123)
Co-authored-by: Bill Easton <strawgate@users.noreply.github.com>
Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
Co-authored-by: Jeremiah Lowin <153965+jlowin@users.noreply.github.com>
2026-02-09 20:54:16 -05:00
Jeremiah Lowin
eeb17855a6
docs: add pre-registered OAuth clients to v3-features (#3129) 2026-02-09 20:48:31 -05:00
marvin-context-protocol[bot]
a1cf2aff6e
chore: Update SDK documentation (#3116)
Co-authored-by: marvin-context-protocol[bot] <225465937+marvin-context-protocol[bot]@users.noreply.github.com>
2026-02-09 20:45:52 -05:00
Jeremiah Lowin
81a7c83c67
CI: Commit generated artifacts back to PR branch instead of opening separate PRs (#3128) 2026-02-09 20:45:28 -05:00
Bill Easton
5bab188106
Add concurrent tool execution with sequential flag (#3022)
Co-authored-by: Bill Easton <strawgate@users.noreply.github.com>
Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
Co-authored-by: Jeremiah Lowin <153965+jlowin@users.noreply.github.com>
2026-02-09 20:43:53 -05:00
Martim Santos
939cf5fcf2
feat: add Static Client Registration (#3085) (#3086)
Co-authored-by: Jeremiah Lowin <153965+jlowin@users.noreply.github.com>
2026-02-09 20:40:06 -05:00
Jeremiah Lowin
45af482e73
Add Azure OBO dependencies, auth token injection, and documentation (#2918) 2026-02-09 20:06:57 -05:00
Jeremiah Lowin
dfb857aa08
Scope Martian triage to bug-labeled issues for jlowin (#3124) 2026-02-09 10:35:53 -05:00
Jeremiah Lowin
3b0660a686
Rename ui= to app= and consolidate ToolUI/ResourceUI into AppConfig (#3117)
* Rename ui= to app= and consolidate ToolUI/ResourceUI into AppConfig

* Remove backward-compat aliases for ToolUI/ResourceUI/ui_to_meta_dict

* Add extra=allow to AppConfig model_config for forward compatibility
2026-02-08 20:26:47 -05:00
Jeremiah Lowin
f7cdd20a42
generate-cli: auto-generate SKILL.md agent skill (#3115)
* generate-cli: auto-generate SKILL.md alongside CLI script

generate-cli now produces a SKILL.md agent skill file next to the CLI
script, documenting every tool's exact invocation syntax, parameter
flags, and types. Agents can use the CLI immediately without discovery.

* Use uv run --with fastmcp in generated SKILL.md invocations

* Fix skill generation issues from review

- Escape pipe chars in union type labels so markdown tables render
- Boolean params omit <value> placeholder in example invocations
- Quote YAML frontmatter values to handle special chars in names
- Match cyclopts camelCase→snake_case in flag derivation
- Use four-backtick fence for nested code block in docs

* Replace --skill/--no-skill with just --no-skill

* Escape quotes in YAML frontmatter description

* Strip newlines from param descriptions in skill table rows

* Detect boolean union types for flag placeholder
2026-02-08 17:09:28 -05:00
Guillaume FORTAINE
3d665d48c0 fix: stabilize task notification integration tests
🤖 Generated with Codex
2026-02-08 15:20:22 +01:00
Guillaume FORTAINE
1f84e5055e test: rewrite as integration tests with zero mocks
Replace 1300+ lines of mock-heavy unit tests with 391 lines of integration
tests using real Client(mcp) connections and memory:// Docket backend.

- test_context_background_task.py: 17 tests covering report_progress delta
  tracking, elicitation flow, edge cases, and fail-fast on push failure
- test_notifications.py: 2 E2E tests for notification queue lifecycle
2026-02-08 01:59:07 +01:00
Guillaume FORTAINE
5dafe19a9d fix: store progress in Redis and use BLPOP for elicitation
- context.py: report_progress uses delta tracking via increment() instead
  of set_current() (which doesn't exist), stores progress in Redis for
  background tasks
- elicitation.py: replace polling with BLPOP for efficient blocking wait,
  fail-fast on notification push failure, use get_task_context() for
  authoritative session_id
- handlers.py: subscriber cleanup on session disconnect via
  _exit_stack.push_async_callback()
2026-02-08 01:59:04 +01:00
Guillaume FORTAINE
ecbce07636 feat: distributed notification queue for background task elicitation
Add Redis-backed notification queue (LPUSH/BRPOP) enabling the MCP server
to notify clients about background task events like elicitation requests.

- notifications.py: subscriber management with weakref tracking, retry
  logic, TTL expiration, and graceful shutdown
- __init__.py: export ensure_subscriber_running, push_notification,
  stop_subscriber
2026-02-08 01:59:01 +01:00
Jeremiah Lowin
806aa8c579
Update docs to reference beta 2 (#3112) 2026-02-07 10:33:12 -05:00
SrzStephen
1ed9770892
Updated deprecation URL (#3108)
For V3 this should be https://gofastmcp.com/servers/dependency-injection#using-depends
For V2 this should be https://gofastmcp.com/v2/servers/context#using-depends

current url fails for both v2 and v3 documentation

https://gofastmcp.com/servers/dependencies
https://gofastmcp.com/v2/servers/dependencies
2026-02-07 08:18:39 -05:00
Jeremiah Lowin
25f3b0878e
Add missing beta2 features to v3 release tracking (#3105) v3.0.0b2
generate-cli, goose integration, response limiting middleware,
background task context, require_auth removal
2026-02-06 20:27:42 -05:00
marvin-context-protocol[bot]
3e3ed76a8c
chore: Update SDK documentation (#3089)
Co-authored-by: marvin-context-protocol[bot] <225465937+marvin-context-protocol[bot]@users.noreply.github.com>
2026-02-06 20:27:29 -05:00
Jeremiah Lowin
d12d46b049
Exclude content-type header from get_http_headers() to prevent HTTP 415 errors (#3104)
Fixes #3097

When using FastMCP.from_openapi() with APIs that require specific
Content-Type headers (e.g., application/vnd.api+json), the transport
connection's content-type: application/json was being injected into
downstream API requests, causing HTTP 415 (Unsupported Media Type) errors.

This change adds content-type to the exclude_headers set in get_http_headers(),
similar to how accept is already excluded. The MCP transport's content type
has no relevance to downstream API calls and should not be forwarded.

Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
Co-authored-by: Jeremiah Lowin <jlowin@users.noreply.github.com>
2026-02-06 20:19:36 -05:00
Jeremiah Lowin
931d6f878c
Remove require_auth; fix auth docs re: component-level enforcement (#3103)
🤖 Generated with Claude Code

https://claude.ai/code/session_01WWzwcBfLWnxoN9XNs5Fhxr

Co-authored-by: Claude <noreply@anthropic.com>
2026-02-06 20:08:50 -05:00
Jeremiah Lowin
ad3b1b9d1b
Fix CIMD redirect allowlist bypass and cache revalidation (#3098)
* Harden CIMD redirect and cache handling

* Preserve CIMD cache policy on 304 revalidation

* Refresh 304 cache expiry from cached lifetime
2026-02-06 20:08:23 -05:00
Jeremiah Lowin
85eff33b81
Infer MIME types from OpenAPI response definitions (#3101)
* Infer mime_type from OpenAPI response content types for resources

🤖 Generated with Claude Code

https://claude.ai/code/session_01FZD5ZT8WiQqfBu39ybuQis

* Handle media types without schemas in MIME inference

🤖 Generated with Claude Code

https://claude.ai/code/session_01FZD5ZT8WiQqfBu39ybuQis

---------

Co-authored-by: Claude <noreply@anthropic.com>
2026-02-06 20:08:08 -05:00
Jeremiah Lowin
6a358902f2
Fix compress_schema to preserve additionalProperties: false for MCP compatibility (#3102)
Changes:
- Changed default of prune_additional_properties from True to False in compress_schema
- Added test demonstrating MCP client compatibility requirement
- Updated existing tests to explicitly enable pruning when needed
- Added additionalProperties: false to manually constructed schemas in tool_transform
- Updated inline snapshots to reflect new behavior

Fixes #3008

Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
Co-authored-by: Jeremiah Lowin <jlowin@users.noreply.github.com>
2026-02-06 18:52:21 -05:00
Jeremiah Lowin
b8d789c1b4
Document token passthrough security in OAuth Proxy docs (#3100) 2026-02-06 18:20:17 -05:00
Jeremiah Lowin
32c6826e13
Add note about output_schema incongruity when responses are truncated (#3099) 2026-02-06 18:15:57 -05:00
Diogo Santos
30832ced1c
Add ResponseLimitingMiddleware for tool response size control (#3072) 2026-02-06 18:13:26 -05:00
Jeremiah Lowin
880d835ccc
Add CIMD (Client ID Metadata Document) support for OAuth (#2871) 2026-02-06 13:44:52 -05:00
Jeremiah Lowin
db650ca7cb
Update Anthropic and OpenAI clients to use Omit instead of NotGiven (#3088)
* Fix ty 0.0.15 type errors: use Omit/omit instead of NotGiven/NOT_GIVEN

🤖 Generated with Claude Code

https://claude.ai/code/session_01Fs5vHiWaUebe826pGq4eCN

* Use kwargs dict to avoid NotGiven/Omit sentinel type issues across SDK versions

🤖 Generated with Claude Code

https://claude.ai/code/session_01Fs5vHiWaUebe826pGq4eCN

* Bump ty minimum to 0.0.15

🤖 Generated with Claude Code

https://claude.ai/code/session_01Fs5vHiWaUebe826pGq4eCN

---------

Co-authored-by: Claude <noreply@anthropic.com>
2026-02-05 11:40:25 -05:00
marvin-context-protocol[bot]
5e0211dd5e
chore: Update SDK documentation (#3069)
Co-authored-by: marvin-context-protocol[bot] <225465937+marvin-context-protocol[bot]@users.noreply.github.com>
2026-02-04 18:31:07 -05:00
Diogo Santos
b776089ecc
Add @handle_tool_errors decorator for standardized error handling (#2885)
* Add @handle_tool_errors decorator for standardized error handling

* Add tests for @handle_tool_errors decorator

* Add documentation for @handle_tool_errors decorator

* Fix type checking: use getattr for func.__name__ with fallback

* Add @overload declarations for proper async/sync type checking

* Fix type checking: reorder overloads and use Coroutine for async typing

* Update lockfile and fix test formatting

* Improve error_handling module: add docstrings, fix logging, handle cancellation, and update documentation

* Add auth error mappings, doc tweaks, and doc fix

* Pivot to hybrid approach

* Remove decorator, keep only core 429/timeout handling

---------

Co-authored-by: Jeremiah Lowin <153965+jlowin@users.noreply.github.com>
2026-02-04 18:30:31 -05:00
Richard Michael
d5f5300e63
Add server version to banner (#3076) 2026-02-04 17:40:09 -05:00
Jeremiah Lowin
422384e576
Fix --reload port conflict when using explicit port (#3070)
* Fix --reload port conflict by killing entire process group

* Gate start_new_session on Unix (no-op on Windows)
2026-02-04 17:35:39 -05:00
Nathan
76f054e957
fix: enforce redirect URI validation when allowed_client_redirect_uris is supplied (#3066)
* fix: enforce redirect URI validation when patterns are explicitly configured

Security fix: When allowed_redirect_uri_patterns is explicitly set, reject redirect URIs that don't match the patterns instead of falling back to parent validation. This prevents unauthorized OAuth clients from bypassing the allowlist and accessing protected resources.

* Update models.py

no need to return twice

* fix redirect uri access issue

* update style

* feat: add unit test to enforce fallback not applied when redirect uri's supplied

* fix: improve test case

* apply linter

* refactor: simplify logic and do not exposed allowed redirect patterns

---------

Co-authored-by: Nathan <2381793w@student.gla.ac.uk>
2026-02-04 17:33:33 -05:00
Jeremiah Lowin
4262cfc16a
Add fastmcp generate-cli command (#3065)
* Add `fastmcp generate-cli` command

Connects to any MCP server, reads its tool/resource/prompt schemas,
and writes a standalone Python CLI script with typed subcommands.

* docs: add generate-cli documentation

* docs: add generate-cli documentation; skip Windows executable test

* fix: address PR review feedback

- Sanitize tool and parameter names to valid Python identifiers
- Replace bare except Exception with specific exception types
- Escape server name in generated string literals
- Handle trailing colon edge case in _derive_server_name
- Clarify in docs that generated CLI is a client, not a bundled server

* Fix string escaping issues in generate-cli

- Use single-quoted docstrings to avoid triple-quote escaping issues
- Escape quotes in app_name derived from server_name
- Add tests for descriptions with quotes and server names with quotes

Addresses CodeRabbit review comments about insufficient escaping.

* Implement smart parameter handling for generate-cli

- Simple types (str, int, float, bool): Direct typed flags
- Arrays of simple types (list[str], list[int]): Repeatable flags via cyclopts
- Complex types (objects, nested arrays): Accept JSON strings with parsing
- JSON schema shown in help text for complex parameters
- Proper escaping of newlines and quotes in help text
- Filter out None and empty list defaults when calling tools

This gives typed, discoverable CLIs for common cases while handling
complex schemas via JSON input.

* Update generate-cli docs to explain smart parameter handling

- Document simple types as direct typed flags
- Document arrays of simple types as repeatable flags
- Document complex types as JSON strings with schema in help
- Add examples showing all three patterns

* Fix Codex review issues in generate-cli

High priority fixes:
- Complex type defaults: Serialize dict/list defaults to JSON strings
- List params: Preserve help metadata with Annotated wrapper
- Name collisions: Detect and error on sanitized name conflicts
- JSON parsing: Use isinstance check for safety with defaults

Added tests for:
- Complex types with default values
- Parameter name collision detection
- Updated existing tests to match new format

* Use pydantic_core.to_json for consistency

- Generator now uses pydantic_core.to_json() instead of json.dumps()
- Consistent with rest of fastmcp codebase
- Generated CLI still uses plain json module (standalone script)

* Move local imports to module level in generate-cli

* Handle union item types and Python keyword collisions in generate-cli
2026-02-03 21:08:51 -05:00
marvin-context-protocol[bot]
f6988c9206
chore: Update SDK documentation (#2995)
Co-authored-by: marvin-context-protocol[bot] <225465937+marvin-context-protocol[bot]@users.noreply.github.com>
2026-02-03 21:08:20 -05:00
Jeremiah Lowin
5fd41b2e15
Remove OpenAPI timeout parameter, make client optional, surface timeout errors (#3067)
* Remove OpenAPI timeout param, make client optional, surface timeout errors

* Close auto-created httpx client via provider lifespan
2026-02-03 21:08:09 -05:00