prepare_skill

Overview

This skill guides a codebase through a full, disciplined preparation process for its first public or open-source release. It walks a maintainer through discovery, secrets and privacy checks, licensing, documentation, hardening, metadata, and a final review. Each phase reports findings, proposes changes, asks for approval, and implements only after consent.

How this skill works

The skill runs nine sequential phases, each following a simple loop: mark phase in_progress, scan current state, present findings, propose concrete changes, ask the user to skip/approve/modify, implement approved changes, commit with a descriptive message, and mark the phase complete. It detects ecosystem from common files (package.json, pyproject.toml, Cargo.toml, go.mod, etc.) and tailors checks and recommendations accordingly. The tool focuses strictly on public-readiness tasks and avoids code-style or architectural refactors unless explicitly requested.

When to use it

• Preparing a private repository for its first public/open-source release.
• Auditing a project for leakage of secrets or personal information before publishing.
• Standardizing repository metadata and docs for package registry submission.
• Adding minimal CI, dependency management, and contributor guidance before opening contributions.
• Cleaning .gitignore, tracked sensitive files, and licensing mismatches prior to release.

Best practices

• Work one phase at a time and approve proposed changes before implementation.
• Treat secrets and personal info separately: rotate credentials and scrub history when needed.
• Ask the user to choose licenses, CI providers, and hardening level; never impose defaults.
• Check for license consistency between LICENSE file and package metadata.
• Keep hardening minimal unless the user requests more (avoid over-engineering).

Example use cases

• Node.js library: detect package.json, add LICENSE, refine .gitignore, and propose GitHub Actions CI.
• Python CLI tool: scan for .env, ensure README includes usage, and propose SECURITY.md for disclosures.
• Rust crate: verify Cargo.toml metadata, suggest keywords and repository fields for crates.io.
• Repo with tracked secrets: surface files, recommend git history rewrite tools, and propose rotation steps.
• Project with no CONTRIBUTING.md: propose a simple PR workflow and testing expectations for contributors.