FAQ

What is Another Agent Skills?

A framework of 58 composable skills that turn AI coding agents into disciplined senior engineers. It adds mechanical enforcement, not just prompts. Skills follow a six-phase lifecycle: Define, Plan, Build, Verify, Review, Ship.

Do I install it per project or once?

Once per machine. The installer places the skills globally, and any project can use them without duplicating the files. You run init-agents in each project to activate the framework there.

Does it work on Windows, macOS, and Linux?

Yes. The installer is POSIX-first and works on Linux and macOS. On Windows, use Git for Windows (Git Bash); install.ps1 provides Claude Code parity. The npm wrapper is the portable path for Node users.

A teammate clones my project and does not have the framework. Does it break?

No. The project still runs. Your teammate can install the framework when they want it, and the remote gates check enforces the rules for everyone on the repository regardless of their local setup.

I inherited or migrated a project that used the framework. How do I repair it?

Run aas doctor, then init-agents --dry-run, then init-agents --repair. The repair merges without losing data.

I changed machines. What do I do?

Install the same version with aas install, then run aas doctor to confirm the environment.

There is a new release. Do I have to update the project?

No. A non-blocking drift advisory appears in pre-commit and doctor when the installed version differs from the project’s pinned version. When you are ready, run aas upgrade, then init-agents --dry-run and init-agents --repair.

What are L1, L2, and L3, and what is actually enforced?

L1 is local git hooks for fast, advisory feedback. L2 is a required remote gates check that the committer cannot skip. L3 is CODEOWNERS review for sensitive paths. L2 and L3 require GitHub. A code change with no matching test is blocked, locally and remotely.

Which git and GitHub setups are supported?

Four flows, and the framework works in all of them. No git: convention only (rules, skills, AGENTS.md); run git init and re-run init-agents to add the local hooks. Local git: L1 hooks only, no remote gate. Git and GitHub: full L1, L2, and L3. Git later: the layers grow as they appear; re-run init-agents after git init and after adding the remote, because it installs each layer conditionally. Step-by-step: Start without git, add it later and Wire the remote enforcement.

Where are the step-by-step guides?

The Tutorials section walks through the common jobs with copy-paste commands and the output you should see: your first gated commit, wiring the remote enforcement, starting without git, migrating a legacy project, and moving to another machine. See Your first gated commit.

Which agents does it support?

Designed for OpenCode. Portable to Claude Code, Cursor, Codex, Gemini CLI, GitHub Copilot, Windsurf, Aider, Kiro, Zed, and any agent that reads AGENTS.md. The installer detects the agent and wires the matching skills and hooks.

Is it free?

Yes. MIT License, open source, no subscriptions and no paid tiers. The installer, the skills, and the enforcement are all free.

Can an agent still break the rules?

Yes. The hooks make it harder, not impossible. An agent could misread an ambiguous approval or act outside protocol. These gates create friction, not guarantees. The human stays in the loop, and the human stays alert.

What is the Harness?

The Harness is everything around the model that turns raw intelligence into reliable output: instructions, tools, sandboxes, orchestration, guardrails, and observability. Agent equals Model plus Harness. Most agent failures are configuration failures.

How do I install it?

Clone and run install.sh, or use the pinned curl bootstrap from the latest release. Git, curl and npm are available today. Then run init-agents in any project. See Getting started.