[Frontend] Add MCP type support infrastructure to Responses API

[daniel-salib]

Add MCP type support infrastructure to Responses API

Summary

Add MCP type support infrastructure to Responses API

Changes

• Import and use McpCall type from openai library
• Add _parse_mcp_call() function to parse MCP tool calls into properly typed output items
• Update parse_output_message() to route unknown recipients (not browser.*, functions.*, or built-in tools) to MCP parsing
• Update parse_remaining_state() to handle in-progress MCP calls
• Add server_label field to McpCall objects (extracted from recipient name)

Test Plan

Unit tests:

python3 -m pytest tests/entrypoints/test_harmony_utils.py -k "mcp" -v

Manual test - Before this change (missing server_label field):

curl -X POST "http://localhost:8000/v1/responses" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer dummy-api-key" \
  -d '{
    "model": "default",
    "input": "Use the filesystem tool to list files in /tmp directory",
    "tools": [{
      "type": "mcp",
      "server_label": "filesystem",
      "headers": {"test": "test"}
    }],
    "enable_response_messages": true
  }'

Response: {"error":{"message":"unhandled errors in a TaskGroup (1 sub-exception)","type":"BadRequestError","param":null,"code":400}}

After this change:
Same curl command now returns:

{
  "output": [
    ...reasoning items...,
    {
      "id": "mcp_bc8c1958cd96823c",
      "arguments": "{\"path\": \"/tmp\"}",
      "name": "filesystem",
      "server_label": "filesystem",
      "type": "mcp_call",
      "status": "completed"
    }
  ]
}

[gemini-code-assist]

Code Review

This pull request adds support for MCP (Multi-turn Conversation Protocol) tool calls to the Responses API. The changes introduce parsing logic for MCP calls in both complete and streaming responses, update the protocol definitions, and add corresponding tests. My review identified a critical issue in the streaming logic that could lead to incorrect behavior for certain built-in tools, and a high-severity issue regarding code duplication that impacts maintainability. I've provided suggestions to address both. The rest of the changes, including the new tests and protocol updates, are well-implemented.

[chaunceyjiang]

/cc @yeqcharlotte @qandrew PTAL

[mergify]

Hi @daniel-salib, the pre-commit checks have failed. Please run:

uv pip install pre-commit
pre-commit install
pre-commit run --all-files

Then, commit the changes and push to your branch.

For future commits, pre-commit will run automatically on changed files before each commit.

[mergify]

This pull request has merge conflicts that must be resolved before it can be
merged. Please rebase the PR, @daniel-salib.

https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork

[yeqcharlotte]

LGTM @qandrew @daniel-salib can we create a follow-up PR to update examples?

[mergify]

Documentation preview: https://vllm--30054.org.readthedocs.build/en/30054/