Repo to Claude Skill Converter

What does Repo to Claude Skill Converter do?

Repo to Claude Skill Converter takes any GitHub repository — a CLI tool, a library, an API client — and produces a draft Claude Code Skill package (SKILL.md + install.sh + README.md) bundled as a single downloadable .zip. Anthropic released the Claude Code Skills ecosystem in late 2025; tool authors increasingly want their project wrapped as a Skill so Claude Code users can invoke it as a one-line capability. Writing a Skill by hand takes 30+ minutes per repo — this Actor automates ~70% of that work so you only need to review, polish, and publish.

Try it on Apify Console by entering a repo like astral-sh/ruff, vercel/next.js, or cli/cli and downloading the generated skill.zip from the key-value store.

Why use Repo to Claude Skill Converter?

• Bootstrap your Skill library fast — wrap 10 favorite repos as Claude Code Skills in minutes.
• Tool authors can offer a "Claude Code Skill" download as a marketing artifact next to their npm/pip page.
• Agencies and consultants building bespoke Claude Code setups for clients can ship a tailored skill bundle per project.
• Reproducible — the same repo produces the same Skill draft, perfect for CI pipelines.
• Apify platform advantages: schedule re-generation on every upstream release, API access for batch processing, monitoring, integrations.

How to use Repo to Claude Skill Converter

1. Open the Actor on Apify Console.
2. Enter githubRepo in the form owner/name (e.g. astral-sh/ruff).
3. Optionally set skillName to override the default (derived from the repo name).
4. Click Start.
5. When the run completes, download skill.zip from the Key-Value Store tab.
6. Unzip into ~/.claude/skills/<skill-name>/, run sh install.sh, restart Claude Code.

Input

Field	Type	Required	Description
githubRepo	string	yes	owner/name, e.g. vercel/next.js
skillName	string	no	Override auto-derived name; lowercase letters, digits, hyphens

Example input JSON:

{
  "githubRepo": "astral-sh/ruff",
  "skillName": "ruff"
}

Output

A row in the dataset summarizing what was generated, plus the actual files in the key-value store:

• skill.zip — the bundle (<skill-name>/SKILL.md, <skill-name>/install.sh, <skill-name>/README.md).
• <skill-name>-SKILL.md — convenience copy for direct viewing.
• <skill-name>-install.sh — convenience copy.
• <skill-name>-README.md — convenience copy.

Example dataset item:

{
  "repo": "astral-sh/ruff",
  "skillName": "ruff",
  "detectedTool": { "type": "python-cli", "packageName": "ruff", "bin": "ruff", "installCmd": "pip install ruff" },
  "skillMdSize": 1842,
  "installSize": 412,
  "zipBytes": 1721,
  "zipUrl": "see key-value store → skill.zip",
  "stars": 31000,
  "language": "Rust",
  "license": "MIT"
}

You can download the dataset in JSON, HTML, CSV, or Excel.

Data table

Field	Description
repo	Source GitHub repo (owner/name)
skillName	Final skill name used in the zip and SKILL.md
detectedTool	Object with type, packageName, bin, installCmd
skillMdSize	Bytes of generated SKILL.md
installSize	Bytes of generated install.sh
zipBytes	Bytes of the final zip
stars	GitHub stars at generation time
language	Primary language reported by GitHub
license	SPDX license id

Pricing / Cost estimation

This Actor is lightweight — a single repo conversion does 2–8 GitHub API calls and bundles ~3 KB of generated text. On a typical Apify free tier, one run costs well under $0.01 of compute. Batch 100 repos and you are still under $1. There is no proxy cost (uses public GitHub APIs).

Set GITHUB_TOKEN as an environment variable in the Actor settings to raise the unauthenticated rate limit (60/hr → 5000/hr) — strongly recommended for batch usage.

Tips and advanced options

• Batch mode: orchestrate this Actor from another script — pass a list of repos and call the Actor once per repo via the Apify API.
• Pin versions: edit the generated install.sh to pin to a specific release tag instead of latest.
• Library vs CLI: the Actor detects bin fields in package.json, [project.scripts] in pyproject.toml, [[bin]] in Cargo.toml, and main.go for Go. If neither is found, it falls back to library-mode usage.
• Manual review: the output is a draft — the generated SKILL.md gets ~70% there; review usage examples and constraints against the upstream README before publishing.

FAQ, disclaimers, and support

• Is the generated SKILL.md production-ready? It is a draft. Anthropic's Skill format is simple but the content (description, tips, constraints) benefits from human polish.
• What if my repo isn't a CLI? The Actor handles libraries too — generated SKILL.md will show import-style usage instead of CLI commands.
• Does it run code? No. It only reads package.json, pyproject.toml, Cargo.toml, go.mod, README, and repo metadata via the GitHub API.
• Custom solutions: need a tailored Claude Code Skill build pipeline for your company? Open an issue or reach out via the Issues tab.

Built by Ian Mu as Actor #10 in a 100-actor portfolio. Pairs well with claude-verify-before-stop — the verification hook that makes sure Claude Code never claims "done" without proof.