Getting Started
Canopy is a self-hosted code intelligence engine that makes AI coding agents (Claude Code, Cursor, Codex CLI, Windsurf, Zed) dramatically smarter. It runs locally, indexes your codebase, and exposes 21 specialized tools over MCP (Model Context Protocol).
This section gets you from zero to querying your codebase through an agent in under 10 minutes.
What you’ll need
Section titled “What you’ll need”- A repository to index (any size; 10K–100K files tested)
- Claude Code, Cursor, or another MCP-compatible AI coding agent
- macOS, Linux, or Windows
Time estimate
Section titled “Time estimate”| Step | Approximate time |
|---|---|
| Install Canopy | ~1 min |
| Index your first repo | ~2 min (5K files) — ~10 min (50K files) |
| Connect to Claude Code | ~2 min |
| Your first workflow | ~2 min |
| Total | ~7–15 min |
Start here
Section titled “Start here”- Install Canopy — download the binary for your platform.
- Index your first repo — build the Canopy index.
- Connect to Claude Code — wire Canopy into your agent.
- Your first workflow — run
canopy_prepareon a real task.
What Canopy is (and isn’t)
Section titled “What Canopy is (and isn’t)”Canopy does not include an AI model. It is a context engine — it gives existing AI tools precise, structured information about your codebase so they can write better code, catch real bugs, and plan refactors that account for all dependencies.
For the full story, see Concepts: What Canopy Is and Isn’t.