Frequently Asked Questions

CodyMaster is a collection of 34+ AI skills that turn your coding AI (Gemini, Claude, Cursor, etc.) into a disciplined software engineering team. Instead of AI guessing randomly, skills enforce proven workflows: plan before code, test before deploy, verify before claim.

Copy theskills/folder into your project root. That's it — no npm install, no build step, no dependencies. Your AI agent automatically detects and uses the skills. See theGet Started page for platform-specific setup.

CodyMaster works with any AI that reads markdown files from your project: Google Gemini (Antigravity), Claude (OpenClaw/Projects), Cursor Agent, and GitHub Copilot Workspace. Skills are plain markdown — no proprietary format.

Say /cm-start and describe your objective. The AI will read the skill index and automatically chain the right skills together. For new projects, it starts with cm-identity-guard → cm-project-bootstrap → cm-planning. Check the Playbook for all workflows.

Pull the latest from the GitHub repo and replace your skills/ folder. Since skills are just markdown files, there are no dependency conflicts or breaking changes. Your project settings and CONTINUITY.md are preserved.

Skills don't directly communicate. Instead, cm-skill-chain composes them into pipelines, and cm-continuity maintains context between sessions via a shared CONTINUITY.md file. The AI reads this file to know what happened before and what to do next.

Yes! Use cm-skill-mastery to learn how to create custom skills. Just follow the SKILL.md format from any existing skill — it's plain markdown. Your AI agent will guide you through the process when you need a new skill.

Mention the skill directly: @cm-tdd or /cm-tdd. Skills use trigger keywords — so even saying "write tests first" activates cm-tdd. If the AI still skips, break the task into smaller steps and explicitly reference each skill.

The typical flow is: Identity → Analysis → Planning → Implementation → Quality → Deploy. But not all tasks need all phases. Bug fixes skip analysis. Translations only need cm-safe-i18n. See the Playbook for exact chains per scenario.

/cm-start is the autonomous mode — AI auto-detects which skills to use based on your objective. Manual calling (e.g., /cm-debugging) gives you explicit control. Both work. New users should start with /cm-start; power users prefer manual control.

You don't need to learn them all. Start with just 3: cm-start (auto-mode), cm-planning (think before code), and cm-continuity (memory). The AI loads others automatically via the progressive disclosure system — it only reads a full skill when it's needed.

Yes. Drop the skills/ folder into any existing project. CodyMaster is language-agnostic — it works with JavaScript, Python, Go, Rust, or any stack. Skills guide the AI's workflow, not the language or framework.

Minimal. The 3-layer progressive disclosure system means most skills load only ~100 tokens (the skill index). Full skill content (~300-2000 tokens) is loaded only when triggered. You'll actually save tokens because the AI makes fewer mistakes and retry loops.

The skills themselves are local markdown files — always available offline. But you still need internet for the AI model (Gemini, Claude, etc.) since the models run in the cloud. Some features like cm-ads-tracker also need network access for API integrations.

cm-identity-guard verifies your Git identity before every push. cm-git-worktrees creates isolated branches for feature work. cm-secret-shield adds pre-commit hooks to block secret leaks. cm-code-review handles the review and merge lifecycle.

No. CodyMaster doesn't install packages, modify configs, or add runtime dependencies. It's a folder of markdown files that guide your AI's behavior. It works alongside ESLint, Prettier, Husky, or any other tools you already use.

Yes, CodyMaster is free and open source. All 34+ skills, the full documentation, and future updates are free forever. We believe in making AI-assisted development accessible to everyone.

Absolutely. Use it for personal, commercial, or enterprise projects. There are no restrictions on commercial use. Many teams already use CodyMaster in production.

Yes! Fork it, modify skills, create your own, or contribute back. The best improvements come from real-world usage. We actively merge community contributions. PRs are welcome at GitHub.