Built on the Model Context Protocol (MCP), this server provides intelligent code querying and global database management capabilities.
- Code Ingestion: Support for both local codebases and GitHub repositories
- Intelligent Querying: Search and analyze code using semantic understanding
- Multi-Format Support: Handles various code file formats and structures
- Reference Tracking: Maintains relationships between classes and methods
- Multi-Modal Storage: Store and retrieve both text and images
- Vector Search: Semantic search capabilities using LanceDB
- Base64 Support: Handle base64 encoded images
The server provides the following MCP tools:
-
ingest_codebase
: Add a new codebase to the vector database- Supports local paths and GitHub repository URLs
- Automatically processes and indexes code structure
-
codeqa
: Query and analyze ingested codebases- Natural language queries about code
- Returns relevant code snippets and context
-
list_codebases
: List all available ingested codebases
globaldb_ingest
: Store text and images in the global databaseglobaldb_query
: Query stored data using semantic search
pip install basicmcp
{ "mcpServers": { "codeqa": { "command": "uv", "args": [ "--directory", "/path/to/your/project", "run", "--python", "3.12", "basicmcp-codeqa" ] } } }
Location:
- MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json
- Windows: %APPDATA%/Claude/claude_desktop_config.json