MCP Server Whisper

# MCP Server Whisper - Development Guidelines ## Commands - Run tests: `uv run pytest` - Run single test: `uv run pytest tests/path_to_test.py::test_function_name -v` - Run with coverage: `uv run pytest --cov=src` - Type checking: `uv run mypy src` - Linting: `uv run ruff check src` - Format code: `uv run ruff format src` - Run MCP server: `mcp dev src/mcp_server_whisper/server.py` - Install MCP server: `mcp install src/mcp_server_whisper/server.py` - Setup pre-commit: `pre-commit install` - Run pre-commit: `pre-commit run --all-files` ## Code Style - Line length: 120 characters - Use type hints for all functions and variables - Import order: standard library → typing → third-party → local - Use async/await for I/O operations - Use Pydantic models for data validation - Function/variable naming: snake_case - Class naming: PascalCase - Constants: UPPER_SNAKE_CASE - Follow Google docstring style for documentation ## Error Handling - Use specific exceptions with descriptive messages - Propagate exceptions with `raise ... from e` - Handle errors at appropriate abstraction levels - Log errors with context information ## MCP Implementation - Tools are defined with `@mcp.tool()` decorator - Resources are defined with `@mcp.resource()` decorator - Use batch processing with lists of inputs for parallel operations - Process multiple files concurrently using `asyncio.gather()` - Return structured data (typically Dict[str, Any]) for consistent handling ## Audio Processing - Supports multiple formats: mp3, wav, mp4, mpeg, mpga, m4a, webm - GPT-4o direct access supports: mp3, wav - Files over 25MB are automatically compressed - Parallel processing for batch transcription jobs - Transcription options: basic Whisper, GPT-4o with prompt, enhanced transcription