Skip to content

Conversation

@taylorwilsdon
Copy link

@taylorwilsdon taylorwilsdon commented Oct 18, 2025

This is a first pass - significant compatibility gains with relatively little overhead

  • centralize configuration: added ServerSettings with cached env loading, transport validation, configurable API timeout, and workspace configuration so all components share the same source of truth.
  • hardened API client: reuse authenticated httpx clients, propagate endpoint-specific failure context, and avoid ad-hoc header construction across tools.
  • modernized tool registration: tightened type hints/docstrings, standardised import structure, and updated object-store uploads to use the shared client context.
  • improved CLI entrypoint: introduced run_server, argparse CLI with transport/host/port/log-level flags, mapped the http alias to FastMCP's streamable-http, and added the quantconnect-mcp console script hook. Allows you to run via python rather than having to build docker and makes development easier.
  • refreshed documentation: documented non-Docker usage, migration notes, environment variables (including QUANTCONNECT_API_TIMEOUT), and transport listing command.
  • expanded tests: coverage for settings validation/cache reset, transport aliasing, API timeout/error paths, and CLI argument parsing.
  • stabilized project tooling: normalized QuantConnect's empty parameter arrays before validation to keep list_projects responses compatible with generated models.

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.

1 participant