`FastMCPToolset` does not work with Temporal

[DouweM]

We need to wrap FastMCPToolset to use activities for listing and calling tools, like we do with MCPServer (See TemporalMCPServer)

[Pavanmanikanta98]

Hi @DouweM,

I'd like to work on this issue.

Root Cause: FastMCPToolset's list_tools() and call_tool() methods perform direct network I/O, violating Temporal's deterministic execution model. This toolset currently lacks the activity-based wrapping seen in TemporalMCPServer.

Solution: Create TemporalFastMCPToolset following the TemporalMCPServer pattern:

• Wrap I/O operations in Temporal activities (get_tools_activity, call_tool_activity)
• Handle RunContext serialization/deserialization (serialize before activity, return only ToolDefinition objects, reconstruct ToolsetTool
  after)
• Integrate into temporalize_toolset() in _toolset.py (add type check after MCPServer handling)
• Use workflow.in_workflow() check to delegate to activities when in workflow context

Files:

• Create: _fastmcp_toolset.py
• Modify: _toolset.py
• Tests: Add to test_temporal.py

This is a low-risk solution that mirrors the proven TemporalMCPServer pattern. Could you assign this to me?

[DouweM]

@Pavanmanikanta98 I just started working on this in #3413 :)

[Pavanmanikanta98]

@DouweM! I see you've already created PR #3413 to fix this.

I'm new to OSS contributions, so it took me some time to understand the root cause, but this was a great learning experience. I'll keep an eye on the PR to learn from your implementation approach. If you need any help with testing or review, I'm happy to contribute.

In the meantime, are there any other open issues that would be good for a first-time contributor? I'm interested in issues beyond just "good first issue" labels - something with moderate complexity would be great.

[DouweM]

@Pavanmanikanta98 If you're using FastMCPToolset with Temporal it'd be great if you could check if it works now that #3413 has been merged to main! If not, please file new issues.

As for good first issues, check out https://github.com/pydantic/pydantic-ai/issues?q=sort%3Aupdated-desc+state%3Aopen+label%3A%22good+first+issue%22