🤖 RageBot MCP

Intelligent Project Context Engine with full MCP Server support. Index your codebase, query it in natural language, and expose it as tools to any AI client.

---

📦 Installation

To install RageBot as a package, run:

pip install ragebot

---

✨ Features

• 🔍 Semantic Code Search: Powered by FAISS and Sentence-Transformers for deep codebase understanding.
• 💬 Interactive AI Chat: Multi-turn persistent sessions with context preservation.
• 🤖 Multi-Provider LLM: Support for Google Gemini, Groq, and local Ollama models.
• 🔌 Full MCP Server: Expose your codebase as tools to Claude Desktop, Cursor, Zed, and more.
• 📖 Code Intelligence: Auto-generate documentation, unit tests, and symbol explanations.
• 📸 Snapshot Management: Save, list, and restore project index states instantly.
• 🔒 Secure Key Storage: API keys are stored in the OS keyring (macOS Keychain, Windows Credential Manager).
• 🎨 Modern CLI: Rich terminal output with interactive menus and progress indicators.

---

🚀 Quick Start

1. Authenticate

Configure your preferred AI provider:

# Interactive setup for Gemini, Groq, or Ollama
rage auth 

2. Initialize & Index

Prepare your project for analysis:

cd /path/to/your/project
rage init
rage save

3. Start Querying

Ask questions or enter the interactive REPL:

# Ask a quick question
rage ask "How does the authentication logic work?"

# Start interactive REPL
ragebot

---

📋 Command Reference

Core Commands

Command	Description
rage init	Initialize RageBot in the current directory.
rage save	Index the project and save a snapshot.
rage ask	Ask a one-shot question about the codebase.
rage chat	Start an interactive multi-turn chat session.
rage search	Perform a semantic search across files.
rage explain	Get a detailed explanation of a file or symbol.
rage docs	Generate Markdown documentation for a file.
rage test	Generate pytest test cases for a file.
rage status	Show project indexing and LLM connectivity status.
rage context	Show project overview (use --tree for file structure).

Management Commands

Command	Description
rage auth	Manage API keys and active providers.
rage config	View or edit configuration settings.
rage model	Switch between available models for the active provider.
rage snapshot	List, restore, or delete project snapshots.
rage history	List and resume previous chat sessions.

---

🔌 MCP Server

RageBot is a compliant Model Context Protocol server. You can use it as a toolset for AI IDEs and clients.

Tools Provided:

• ragebot_ask: AI question & answer with citations.
• ragebot_search: Semantic/keyword search.
• ragebot_save: Index or re-index the project.
• ragebot_explain: Explain files or symbols.
• ragebot_file_tree: Get project structure.
• ragebot_generate_docs: Auto-generate documentation.
• ragebot_generate_tests: Auto-generate tests.
• ragebot_apply_edit: Apply natural language edits to files.

Start the Server:

# Stdio transport (default)
ragebot-mcp-server --project /path/to/your/project

# SSE transport
ragebot-mcp-server --transport sse --port 8765 --project /path/to/your/project

---

📁 Project Structure

ragebot/
├── agents/             # Context builders for specific AI roles
├── auth/               # Secure provider authentication (Keyring)
├── core/               # Orchestrator, Scanner, and Config management
├── llm/                # LLM implementations (Gemini, Groq, Ollama)
├── mcp/                # MCP Server implementation (Stdio & SSE)
├── parsers/            # AST-based code and document parsers
├── search/             # Vector embeddings and FAISS retrieval
├── storage/            # SQLite database and Snapshot logic
└── utils/              # UI helpers, logging, and token utilities

---

🛠️ Development

If you want to contribute or run from source:

1. Clone the repository:

   git clone https://github.com/atharvrahate296/Ragebot-MCP
   cd Ragebot-MCP
   

2. Install in editable mode:

   pip install -r requirements.txt
   

3. Run tests:

    python ./tests/tests.py
   

---

Contributions from all collaborators are welcome for the continued development of RageBot MCP.

---

📄 License

MIT © Atharv Rahate