Model Context Protocol (MCP) server for researching Git repositories using semantic search This MCP server enables developers to research external Git repositories and influence their code generation without having to clone repositories to local projects. It provides tools to index, search, and explore Git repositories using semantic search powered by Amazon Bedrock and FAISS.
This server provides the following tools for AI assistants:
Indexes a Git repository (local or remote) using FAISS and Amazon Bedrock embeddings.
Performs semantic search within an indexed repository.
Searches for GitHub repositories based on keywords, scoped to AWS organizations.
Accesses file or directory contents within repositories or on the filesystem.
Create searchable FAISS indexes from local or remote Git repositories
Query repository content using natural language and retrieve relevant code snippets
Get directory structures and identify key files like READMEs
Find repositories in AWS-related organizations filtered by licenses and keywords
Access repository files and directories with support for both text and binary content
Deletes an indexed repository.