Skip to content

Conversation

@pane2004
Copy link

Implements MCP server integration for Flare AI Kit, enabling agents to connect to external tools and services via MCP within TEE environments.

  • Add MCPSettings and MCPServerConfig for flexible server configuration via MCP__SERVERS environment variable
    • Servers are not manually added through github.py, etc, and instead dynamic in a Cursor/IDE json format
  • Add MCPManager that creates and manages MCP toolsets with support for three transport types:
    • stdio: Local process-based servers (e.g., filesystem, git)
    • SSE: Remote servers using Server-Sent Events
    • HTTP: Remote servers using streamable HTTP
  • Integrate with Google ADK's McpToolset for seamless agent tool composition
  • Support tool filtering, server enable/disable, timeout configuration, and authentication headers
  • Add comprehensive unit tests and documentation

Test plan

  • Run uv run pytest tests/unit/mcp/ to verify all unit tests pass
  • Configure an MCP server via MCP__SERVERS env var and verify toolset creation.
    Example config for OpenSea:
    MCP__SERVERS='{"OpenSea": {"url": "https://mcp.opensea.io/mcp","headers":{"Authorization": "Bearer <token>"}}}'

Closes #5

@pane2004 pane2004 changed the title feat: Implement MCP Integration Connectors feat(mcp): Implement MCP Integration Connectors Dec 27, 2025
@pane2004 pane2004 changed the title feat(mcp): Implement MCP Integration Connectors feat(mcp): implement MCP integration connectors Dec 27, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Agent Framework] [Optional] Implement MCP Integration Connectors for TEE Operation

1 participant