axon-rag 0.2.0

General-purpose open-source RAG engine with multi-LLM, hybrid retrieval, GraphRAG, and MCP support

axon-rag

Your documents, answerable. On your hardware.

Drop in PDFs, code, spreadsheets, or URLs — ask anything, get cited answers from a local LLM. Nothing leaves your machine.

---

---

Why Axon?

Most RAG tools make you choose between cloud power and data privacy. Axon can run entirely on your hardware — full capability, zero egress in local-only mode.

• Private by default — by default, inference runs locally via Ollama. No API key required, no upload, no telemetry. Cloud providers (OpenAI, Gemini, Grok, vLLM, GitHub Copilot) are also supported via API keys.
• Ingest anything — 54 file formats (PDF, DOCX, PPTX, Jupyter, code, images, URLs) with SHA-256 dedup.
• Works in your tools — @axon in Copilot Chat, MCP server for Claude Code / Codex / Gemini CLI / Cursor, Graph panel in VS Code.
• Built for teams — share your knowledge base with signed, revocable read-only keys.
• Production-grade retrieval — hybrid search, reranking, HyDE, multi-query expansion, RAPTOR, GraphRAG.

---

Install

pip install axon-rag

Requires Python 3.10+. For local inference with the default config you'll also need Ollama running locally; cloud providers (OpenAI, Gemini, Grok, vLLM, GitHub Copilot) work via API keys and do not require Ollama.

Optional extras

pip install axon-rag[ui]        # Streamlit web UI (axon-ui)
pip install axon-rag[chroma]    # ChromaDB vector store
pip install axon-rag[qdrant]    # Qdrant vector store
pip install axon-rag[graphrag]  # GraphRAG (networkx + leidenalg)
pip install axon-rag[loaders]   # Extra file loaders (EPUB, RTF, email)
pip install axon-rag[all]       # Everything above

---

Quick Start

# Pull a model (first time only)
ollama pull llama3.2

# Launch the interactive REPL
axon

# Ingest a file or folder
axon ingest ~/docs/myreport.pdf
axon ingest ~/projects/myrepo/

# Ask a question
axon query "What are the key findings?"

---

Entry Points

Command	What it does
axon	Interactive REPL — day-to-day exploration
axon-api	FastAPI REST server on port 8000
axon-mcp	MCP stdio server — 30 tools for Claude Code, Codex, Gemini CLI, Cursor
axon-ui	Streamlit web UI on port 8501 (requires [ui] extra)
axon-ext	Install the bundled VS Code extension

---

VS Code Extension

Install the bundled extension to get the @axon chat participant, Knowledge Graph panel, Code Graph panel, and Governance dashboard directly inside VS Code alongside GitHub Copilot:

axon-ext

---

MCP Server

Connect any MCP-compatible agent to your local knowledge base:

{
  "servers": {
    "axon": {
      "command": "axon-mcp"
    }
  }
}

30 tools available: ingest, query, search, project management, graph operations, sharing, and more.

---

Configuration

On first run, Axon creates ~/.config/axon/config.yaml. Key settings:

llm:
  provider: ollama          # ollama | openai | gemini | grok | vllm | ollama_cloud | github_copilot
  model: llama3.2
  base_url: http://localhost:11434

embedding:
  provider: sentence_transformers
  model: BAAI/bge-small-en-v1.5

rag:
  top_k: 5
  hybrid: true
  rerank: false
  hyde: false

---

Documentation

Full documentation is available on GitHub:

• Getting Started — first-time walkthrough
• Setup Guide — install, models, VS Code, MCP
• Admin Reference — every endpoint, command, and config option
• Advanced RAG — HyDE, RAPTOR, GraphRAG, CRAG-Lite
• Troubleshooting — common errors and fixes

---

License

MIT — see LICENSE