Documentation

Get started in 10 seconds.

Two commands and your AI assistant understands your entire architecture. No config files. No cloud accounts. No setup meetings.

Quick start

Install Graqle

$ pip install graqle

Install via pip. Python 3.9+ required.

Build your knowledge graph

$ cd your-project && graq init

Scans your codebase, builds the graph. Takes ~10 seconds.

Ask your first question

$ graq reason "what breaks if I change auth?"

Queries the graph with multi-agent reasoning.

Full documentation lives on GitHub. Choose a section below:

graq init, graq reason, graq learn, graq compile, graq serve, graq mcp serve, and more.

7 MCP tools: graq_reason, graq_impact, graq_context, graq_preflight, graq_lessons, graq_learn, graq_inspect.

Embed the reasoning engine in CI pipelines, internal tools, or custom dashboards with the Python API.

14 backends: Anthropic, OpenAI, Bedrock, Ollama, Gemini, Groq, DeepSeek, Together, Mistral, OpenRouter, Fireworks, Cohere, and more.

DRACE scoring on 5 axes, hash-chained audit trails, evidence chains, and confidence scores for every answer.

The graq doctor command checks your installation, validates your knowledge graph, and reports any issues.

$ graq doctor