📚 RepoBrain Documentation

Welcome to the documentation for RepoBrain — a cross-IDE repository knowledge engine for grounded codebase Q&A.

🎯 Quick Navigation

Getting Started

• Quick Start Guide — Installation, local development, and first steps
• Usage Reference — Commands, automation, optional context files, and MCP server setup
• Project Philosophy — Product thesis and support boundaries

Core Features

• Zero-Config Features — Automatic tool and context discovery
• MCP Integration — Connect to external tools and data sources
• Sandbox Execution — Trusted-local boundary and Microsandbox opt-in
• Multi-Agent Swarm — Orchestrate specialist agents for complex tasks

Planning & Vision

• Development Roadmap — Completed phases and architectural vision

🌟 Key Features

🧠 Repository Knowledge Base

rb-refresh builds .repobrain/ so AI tools can reason over a fresh model of the repository.

🛠️ Routed Codebase Q&A

rb-ask routes questions to the right module context and answers with file evidence.

🎓 Native Plugin Channels

Claude Code and Codex CLI get slash commands for setup, refresh, ask, and workspace bootstrapping.

🔌 Compatible Integrations

Other IDEs can use shared context files, raw CLI commands, or the optional MCP server.

🚀 Common Tasks

I want to...

Task	Documentation
Get started with RepoBrain	Quick Start
Build a custom tool	Zero-Config Features
Initialize a new project from this template	Zero-Config Features
Connect to an MCP server	MCP Integration
Use multiple agents	Multi-Agent Swarm
Understand the architecture	Project Philosophy
See what's coming	Development Roadmap
Query project context	rb-ask "question" / rb-refresh

📊 Project Structure

.
├── cli/                         # rb CLI, templates, offline helpers
├── engine/repobrain_engine/    # Knowledge engine, hub, MCP server, sandbox
├── commands/                    # Shared slash-command definitions
├── skills/                      # Plugin-facing skills
├── docs/                        # Documentation in multiple languages
├── artifacts/                   # Plans, reports, and benchmark outputs
├── memory/                      # Markdown interaction memory
└── .repobrain/                # Generated knowledge base in target repos

🎓 Documentation by Role

For Developers

1. Start with Quick Start
2. Learn Zero-Config tool discovery
3. Explore the swarm protocol

For DevOps/Deployment

1. Read Quick Start Docker section
2. Understand Sandbox Execution safety boundaries
3. Configure MCP servers in MCP Integration

For Architects

1. Understand Project Philosophy
2. Study Multi-Agent Swarm architecture
3. Review Development Roadmap vision

For Contributors

1. Read Project Philosophy
2. Read the repository root CONTRIBUTING.md
3. Open an issue to propose ideas

🔗 External Resources

• 🌐 RepoBrain Official Docs
• 📘 MCP Protocol Specification
• 🐍 Python Documentation
• 🐳 Docker Documentation
• 🧪 Pytest Documentation

❓ FAQ

Q: Which LLM providers are supported?
A: Run rb-setup and choose one of the OpenAI-compatible providers it offers: OpenAI, DeepSeek, Groq, DashScope, NVIDIA NIM, Ollama, or a custom endpoint. The command writes OPENAI_BASE_URL, OPENAI_API_KEY, and OPENAI_MODEL.

Q: How do I add a custom tool?
A: Drop a Python file in engine/repobrain_engine/tools/ with your functions. No registration needed! See Zero-Config Features.

Q: How do I initialize a fresh project from this template?
A: Use the agent-repo-init skill in quick or full mode, or run skills/agent-repo-init/scripts/init_project.py. See Zero-Config Features.

Q: How do I deploy the MCP server?
A: Use Docker or run rb-mcp --workspace <project>. See Quick Start Docker section.

Q: Can I use multiple agents?
A: Yes! Use the swarm system. See Multi-Agent Swarm.

Q: How do I add context/knowledge? A: Create files in .context/ directory. They're automatically loaded! See Zero-Config Features.

Q: What is the Knowledge Hub? A: The Knowledge Hub (rb-ask, rb-refresh, rb report, rb log-decision) maintains project context in .repobrain/, making all AI IDEs smarter. See the main README.

Q: What languages does module detection support? A: Python, TypeScript/JavaScript, Go, Rust, Java, Kotlin, Swift, C/C++, and C#. The scanner uses a unified extension list to detect modules across all supported languages.

Q: What are structured facts? A: Since April 2026, rb-refresh produces structured JSON claims with source evidence (file path + line range) per module. rb-ask verifies these claims against live source before answering, reducing hallucination and improving traceability.

🤝 Contributing

We welcome contributions at all levels:

Report Issues

Found a bug? Open an issue

Suggest Ideas

Have an architectural idea? Ideas are contributions too!
Propose your thought

Submit Code

Ready to code? Check the Roadmap to understand the current architecture.

Improve Docs

See a typo or unclear section? Submit a PR to improve the docs!

📞 Support

• 📖 Documentation: You're reading it! (or check README.md)
• 🐛 Bug Reports: GitHub Issues
• 💡 Feature Requests: GitHub Discussions
• 👥 Community: Star the repo to stay updated

👥 Contributors

• @devalexanderdaza — First contributor. Implemented demo tools, enhanced agent functionality, helped shape the early roadmap and completed MCP integration.
• @Subham-KRLX — Added dynamic tools and context loading (Fixes #4) and the multi-agent cluster protocol (Fixes #6).
• @SunkenCost — Added rb clean command and __main__ entry-point guard (#37).
• @aravindhbalaji04 — Unified instruction surface around AGENTS.md (#41).
• @xiaolai — Provided NLPM audit feedback that improved skill frontmatter and dependency hygiene (#51, #52, #53).

📄 License

This project is licensed under the MIT License. See LICENSE for details.

---

Latest Update: August 2026 Current Architecture: Generative knowledge hub + local host-runner + incremental agent-group refresh + structured evidence verification

Friendly Link: LINUX DO