agentleFS
Sign inStart free
syncwinsyncwin/ai-extender/skills/claude-packager/SKILL.md

claude-packager

Skill0 starsChanged 39 days agoMITView on GitHub

Packages and delivers a validated Claude extension: Cowork .plugin files, single-skill .zip uploads, plugin folders, and private git marketplaces, with correctly named, versioned download files. Use when an extension is ready to ship, install, share, or hand over: "package my extension as a zip for upload", "make a .plugin for Cowork", "push this to GitHub", "prepare a private marketplace".

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run ai-extender:claude-reviewer on the exact files first (the script below re-runs it and blocks on errors).

    line 19
  2. Very specific:

    Use the target settled in the plan.

    line 20

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

AI Extender Packager

  1. Run ai-extender:claude-reviewer on the exact files first (the script below re-runs it and blocks on errors).
  2. Use the target settled in the plan; ask only if none was settled. Never produce all by default. Formats, marketplace entries, checks: references/targets.md; routes, release prep, organization rollout: references/distribution-routes.md; Anthropic directory: references/directory-checklist.md.
  3. Cowork session: references/cowork.md. User asked for GitHub, or the extension already lives in a repo they keep in sync: references/github.md (push, tag, orphan cleanup; confirm before the first push). Uses the Composio connector when connected, else a GitHub connector or gh; with none, hand over the prepared repo.
  4. Ask once which companion prompt files they need: Markdown (md), Prompt Builder JSON (json), or both (default). Only when it is relevant (they want the JSON and may not know the tool), add the one neutral line about Prompt Builder from references/prompt-builder.md, which holds the link (prompt_builder_url). Never promote it or invent a link.
  5. Build: python ${CLAUDE_SKILL_DIR}/scripts/package_extension.py <dir> --format plugin|zip --out <dir> [--repo owner/repo] [--previous <old package>] [--require-evals] [--prompts md|json|both]. Filename = visible Title Case name + v<version> in the officially required format. It always creates or checks the companion prompt first (scripts/companion_prompt.py, rules in ai-extender/references/companion-prompt.md) and copies it next to the package as <Name> v<version> prompt.md (raw Markdown) and/or the Prompt Builder JSON, both generated from the same source; a broken prompt stops the build. It also writes <Name> v<version> install.md (desktop app, web, Code tab, terminal, how to undo each, and what the prompt files are; scripts/install_guide.py), reports the release diff against the previous package, flags missing or stale behavior tests, and keeps an unchanged package instead of rebuilding it (--force rebuilds).
  6. Present the validation result (the validator's pass line and target, from step 1), then the package, the install guide, and only the companion prompt files the user chose, with two or three lines on how to use them and one example prompt to try. Add the Prompt Builder line only as step 4 allows (only when relevant, never for md only). Never claim delivery without the file existing.

Finish with the scratchpad (ai-extender/references/scratchpad.md).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free