[ integrations / github-copilot ]

Team intelligence for GitHub Copilot

Copilot picks up agent skills and AGENTS.md. Codehabits generates both from your team's real review habits — so Copilot suggestions match how you merge, not generic training data.

How it works

  1. 01

    Extract team habits

    Enable analyzes code and PRs into conventions, anti-patterns, domain knowledge, and reviewer expertise.

  2. 02

    Ship as skills Copilot discovers

    Default skill output includes .agents/skills/codehabits-team-intel/ plus AGENTS.md updates.

  3. 03

    Optional MCP

    Where Copilot supports MCP, you can add @codehabits/mcp for active convention checks.

Setup

Step 1

Enable Codehabits in the repo

Writes intelligence files and generates .agents/skills + AGENTS.md.

npx @codehabits/cli enable

Step 2

Open the repo in Copilot Chat / coding agent

Copilot discovers skills under .agents/skills and repo AGENTS.md — no separate Copilot-only stub file required.

Step 3

Keep intelligence fresh

Run codehabits sync (or CI sync) after merges so skills stay current.

npx @codehabits/cli sync

What you get

  • .agents/skills/codehabits-team-intel/
  • AGENTS.md Team Intelligence section
  • .codehabits/ JSON intelligence pack
  • Optional MCP for environments that support it

Docs & guides

FAQ

Does Codehabits write copilot-instructions.md?
No. Copilot discovers team intelligence via .agents/skills and AGENTS.md. That keeps one shared surface for Copilot and other agents.
How do teams stay in sync?
Commit generated skills/AGENTS.md updates (or sync in CI). Everyone pulling the repo gets the same conventions without editing personal Copilot settings.
What about Codex, Windsurf, or Gemini CLI?
Same .agents/skills + AGENTS.md outputs. See the Agent Skills integration for the catch-all path and MCP config notes.