Multi-repo worktree manager with MCP server for AI agents

---

Canopy coordinates Git worktrees across multiple repositories. It creates isolated working directories for each feature, opens them in your IDE, runs pre-commit checks, and exposes every operation as both a CLI command and an MCP tool — so AI agents can operate your workspace through the same interface you use.

Why

Working on a feature that spans multiple repos means coordinating branches, switching, and managing worktrees across all of them. Git worktrees solve the context-switching problem, but the UX for managing them across multiple repos doesn't exist. Canopy provides it: one command to create worktrees in every repo, one command to open them in your IDE, one command to check everything before you commit.

How It Looks

More CLI screenshots

Install

Requires Python 3.10+.

pipx install git+https://github.com/ashmitb95/canopy.git

If you don't have pipx: brew install pipx && pipx ensurepath.

Or from source, for contributors:

git clone https://github.com/ashmitb95/canopy.git ~/projects/canopy
cd ~/projects/canopy && pip install -e .

Quick Start

cd ~/my-product/
canopy init                                        # scan for repos, generate canopy.toml

canopy worktree ENG-412-add-oauth2-login ENG-412   # create worktrees + link Linear issue
canopy code ENG-412                                # open in VS Code (alias resolves)
canopy preflight                                   # stage + run pre-commit hooks
canopy done ENG-412                                # clean up when merged

VSCode Extension

Prefer a sidebar over the CLI? Install Canopy from the VSCode marketplace — features, worktrees, per-repo changes, and review readiness in one native panel, backed by the same MCP tools the CLI uses.

MCP

Every CLI operation is available as an MCP tool:

{
  "mcpServers": {
    "canopy": {
      "command": "canopy-mcp",
      "env": { "CANOPY_ROOT": "/path/to/workspace" }
    }
  }
}

Canopy is also an MCP client — it spawns external MCP servers (Linear, GitHub) to fetch issue and PR data, so there are zero direct API integrations in the codebase.

Docs

• Commands — full CLI reference
• MCP Integration — server tool list + client configuration
• Workspace — layout, canopy.toml, alias resolution, context detection
• Architecture — module boundaries and design rules

Development

cd ~/projects/canopy
pip install -e ".[dev]"
pytest tests/ -v             # 192 tests, ~2s, all use real temporary Git repos

License

MIT