Getting Started

Add AIDDbot to your repo in three simple steps

Skills are plain markdown files — no package to install, no binary to run. You need a git repository and an AI coding agent that reads project instructions (Cursor, VS Code Copilot, Claude Code, Codex, and others).

Step 1 — Clone into your project

From inside your destination repository root, clone the AIDDbot repo:

git clone https://github.com/AIDDbot/AIDDbot AIDDbot-tmp

Move the .agents folder to your project root and delete the temporary clone:

# Bash (macOS / Linux / Git Bash)
cp -r AIDDbot-tmp/.agents ./.agents
rm -rf AIDDbot-tmp

# PowerShell (Windows)
Copy-Item -Path AIDDbot-tmp/.agents -Destination ./.agents -Recurse -Force
Remove-Item -Path AIDDbot-tmp -Recurse -Force

Commit the new .agents folder so skills travel with your repo.

Step 2 — Initialize your environment

In your agent chat, run the initialize skill:

/initialize this project

This creates or updates AGENTS.md with your project paths, stack, and conventions — the entry point for any agent joining the project. Then choose the path that matches your codebase.

Greenfield (starting fresh)

If you are starting fresh, /initialize is enough. Your agent now has project context and the full skill library under .agents/skills/.

Brownfield (legacy codebase)

For existing codebases, run the architect pipeline after initialize:

/initialize → /explore → /extract

/explore reverse-engineers architecture into .product/arch/

/extract captures real coding rules into .product/rules/

What gets created

Path	Purpose
`AGENTS.md`	Main project instructions — stack, paths, conventions
`.agents/skills/`	Skill library (markdown instructions your agent invokes)
`.product/arch/`	Architecture documentation (reverse-engineered from legacy project)
`.product/rules/`	Coding rules and conventions (extracted from your codebase)

Step 3 — Build your first feature

Once initialized, follow the builder pipeline for every new feature:

/specify → /planify → /codify

Verify functionality loop

End-to-end tests confirm specs are actually met — not just that code compiles.

/verify -> /repair? (optional) -> /verify

Review code quality loop

A stack of checks to ensure the code quality, and report any issues.

/review -> /repair? (optional) -> /review

Release

Release the code to the production environment. Bump the version and update the changelog.

/release

Review each step before moving on — you stay in control at every checkpoint.

Next: Skills catalog