study8677/repobrain

What it solves

RepoBrain provides a cross-IDE repository knowledge engine that solves the problem of AI agents forgetting coding styles, hallucinating architecture, or struggling to navigate large codebases. Instead of relying on a single massive documentation file or simple grep searches, it creates a structured, portable knowledge layer that allows AI tools to get grounded, factual answers with exact file paths and line numbers.

How it works

RepoBrain uses a multi-agent cluster to autonomously analyze a codebase. During a refresh process, it groups files by import graphs and directory structure, assigning a dedicated agent to each module to generate a detailed Markdown knowledge document. A Map Agent then indexes these documents into a routing map. When a user asks a question, a Router identifies the relevant modules and feeds the corresponding agent documents to a synthesizer to produce a grounded answer. This process is language-agnostic as it relies on LLM-driven analysis rather than AST parsing.

Who it’s for

It is designed for developers using AI-powered IDEs and CLI tools such as Claude Code, Cursor, Windsurf, Codex CLI, and VS Code + Copilot who want more accurate and faster codebase Q&A.

Highlights

• Cross-IDE Compatibility: Uses a shared .repobrain/ folder and context files (like .cursorrules or CLAUDE.md) to maintain consistency across different AI tools.
• Multi-Agent Analysis: Deploys specialized agents to self-learn each module of a project to prevent information loss in large repos.
• Grounded Answers: Provides responses backed by real code evidence, including file paths and line numbers.
• MCP Support: Can be registered as a Model Context Protocol (MCP) server, allowing agents like Claude Code to call ask_project as a live tool.
• Language Agnostic: Works with any programming language by analyzing directory structures and source code directly via LLMs.