ratchet-init

Initializes a GitHub repository with labels, verification gates, memory files, and branch protection.

2|Updated Jun 13, 2026
One-click install
npx skills add https://github.com/praveenvijayan/Ratchet --skill ratchet-init-praveenvijayan
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ratchet-init
Source: https://github.com/praveenvijayan/Ratchet/tree/main/plugin/skills/ratchet-init
Command: npx skills add https://github.com/praveenvijayan/Ratchet --skill ratchet-init-praveenvijayan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Setting up a repo for agent-driven delivery requires manually creating state labels, configuring verification gates, scaffolding memory files, and wiring up tokens and branch protection — a repetitive, error-prone process that this Skill automates in one idempotent pass. ## Core Features & Use Cases - Label and gate setup: Creates the nine state:* and priority:* labels via gh, then detects the project's stack from real evidence (lockfiles, manifests, Makefile targets) and fills GATES.md with format, typecheck, lint, test, build, audit, and secret-scan commands — writing TODO instead of guessing. - Memory scaffolding and codebase mapping: Creates memory/USER.md, memory/MEMORY.md, and generates a coarse, language-agnostic memory/ARCHITECTURE.md describing directories and their purposes while ignoring vendor and build output. - Token verification and branch protection: Checks that the FACTORY_PAT secret and .env GITHUB_PAT exist (never reading or writing token values), and offers to protect main with required PRs, gates/size status checks, and admin enforcement — only after explicit user confirmation. - Use Case: After installing the Ratchet delivery loop into an existing Node.js repo, run this Skill once to create all labels, auto-detect pnpm-based gate commands, map the codebase, and confirm the PAT and branch protection are in place before agents start picking up issues. ## Quick Start Run the ratchet-init skill to set up this repository's labels, gates, memory files, and branch protection for the agent delivery loop.

Frequently Asked Questions about ratchet-init

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I set up a GitHub repo for agent-driven issue delivery?

Run the ratchet-init skill once per repo. It creates the state and priority labels, fills GATES.md with detected verification commands, scaffolds memory files, verifies the PAT configuration, and offers to protect the main branch.

How does the skill detect test and lint commands for my project?

It inspects real evidence only: lockfiles determine the Node package manager, and package.json scripts, Makefile targets, or pyproject.toml config supply the actual commands. If no evidence exists for a gate, it writes a TODO placeholder instead of guessing.

Is it safe to re-run ratchet-init on an already configured repo?

Yes, the skill is idempotent. Label creation uses --force to update in place, existing memory files are never overwritten, and already-applied branch protection is detected and skipped.

Does the skill handle my Personal Access Token directly?

No. It only checks presence via gh secret list and a non-empty GITHUB_PAT line in .env. Creating the token and setting the secret are credential actions the user must perform; the skill never reads, writes, or logs token values.

Why does branch protection setup fail with a 403 error?

Setting branch protection requires Administration: Read/Write permission, which the default GITHUB_TOKEN and PATs without that scope lack. The skill reports the missing permission and provides manual Settings-based steps or suggests re-running with an appropriately scoped PAT.

What happens if my repo has no code yet when I run the setup?

The skill treats it as greenfield: it leaves the default GATES.md and architecture placeholder, notes that you should re-run it once code lands, and continues with labels, memory scaffolding, and token checks.