brain-trust

# Changelog All notable changes to this project will be documented in this file. The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html). ## [0.1.1] - 2025-10-04 ### Changed - **BREAKING**: Refactored authentication architecture to pass OpenAI API key per-request instead of at server startup - `phone_a_friend` now requires `api_key` parameter and accepts optional `model` and `max_tokens` parameters - `review_plan` now requires `api_key` parameter and accepts optional `model` and `max_tokens` parameters - `health_check` no longer returns `openai_configured` field - Updated README with comprehensive documentation of new per-request authentication flow - Updated MCP client configuration examples to reflect API key passing from client ### Removed - Removed `OPENAI_API_KEY`, `OPENAI_MODEL`, and `OPENAI_MAX_TOKENS` from Docker environment variables - Removed global OpenAI client initialization at server startup - Removed API key requirement from `.env.example` file - Removed `OPENAI_*` environment variables from `fastmcp.json` deployment configuration ### Security - **Improved**: API keys are no longer stored in Docker containers or environment files - **Improved**: Per-request authentication allows different MCP clients to use different API keys - **Improved**: API key management handled securely by MCP client configuration - **Improved**: Server no longer requires access to OpenAI credentials at startup ### Fixed - Server now starts successfully without requiring OpenAI API key configuration - Cleaned up unused imports (`asyncio`, `os`) from server.py ## [0.1.0] - 2025-10-04 ### Added - Initial release of brain-trust MCP server - `phone_a_friend` tool for AI Q&A with optional context - `review_plan` tool with review levels (quick, standard, comprehensive, expert) - `health_check` endpoint for server status - Docker setup (Dockerfile, docker-compose, nginx reverse proxy) - FastMCP HTTP transport on `http://localhost:8000/mcp` - Comprehensive README with setup, usage, and configuration - Open-source documentation: LICENSE, CONTRIBUTING, CODE_OF_CONDUCT, SECURITY, CHANGELOG - GitHub workflows: Python lint/type-check (black, isort, flake8, mypy) and Docker build - GitHub templates: Issue templates (bug report, feature request) and PR template - Dependabot configuration for pip and GitHub Actions - Smoke test script (`scripts/smoke.sh`) for health endpoint validation [0.1.1]: https://github.com/bernierllc/brain-trust-mcp/releases/tag/v0.1.1 [0.1.0]: https://github.com/bernierllc/brain-trust-mcp/releases/tag/v0.1.0