agentleFS
odere-proodere-pro/claude-software-3-0-marketplace/.claude/skills/add-plugin/SKILL.md

add-plugin

Skill2 starsChanged 5 months agoMITView on GitHub

Author-only maintenance skill for the odere-pro marketplace repo. Adds an odere-pro plugin to the registry end to end: vets the candidate repo, inserts a well-formed entry into .claude-plugin/marketplace.json, regenerates the README table, updates CHANGELOG, runs the gate suite, then opens a PR. Invoke explicitly as /add-plugin <repo> [--name <name>]. Edits files.

  • Commits and pushes

7 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 0 are distinctive, and 9 are very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 9

    Very specific

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

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

    Add an odere-pro plugin to this marketplace, all the way to a PR.

    line 16
  2. Very specific:

    If --description or --keywords were supplied by the caller, they override the subagent's curated values when passed to add-entry.sh in step …

    line 29
  3. Very specific:

    Stop on blockers: If the proposal is not ok, print each blocker and the fix and stop — make no edits.

    line 31
  4. Very specific:

    Insert the entry: Run bash ${CLAUDE_SKILL_DIR}/scripts/add-entry.sh <repo> [--name …] [--description "…"] [--keywords a,b], passing the …

    line 35
  5. Very specific:

    Sync docs: Run bash ${CLAUDE_SKILL_DIR}/scripts/sync-readme.sh to regenerate the README Plugins table and bash …

    line 39
  6. Very specific:

    Verify: Run bash tests/gates/run-all.sh and claude plugin validate .

    line 43
  7. Very specific:

    PR.: Create a branch feat/add-<name>, git add the manifest + README + CHANGELOG + site/, commit feat: list <name> in the odere-pro …

    line 45
  8. Very specific:

    Never sets version/sha/commit on an entry.

    line 53
  9. Very specific:

    If any step fails (a blocker surfaces, a gate goes red, claude plugin validate errors, or the PR push fails), stop and roll back so the …

    line 57

“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “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

/add-plugin

Add an odere-pro plugin to this marketplace, all the way to a PR. The deterministic file surgery is done by the bundled scripts under ${CLAUDE_SKILL_DIR}/scripts/; this skill orchestrates and handles judgment (curated description / keywords). The full process and contract are in docs/adding-plugins.md.

$ARGUMENTS is the candidate repo (<repo> defaults the owner to odere-pro, or pass owner/repo), plus optional --name <name> to override the entry name.

Steps

  1. Vet + curate. Delegate to the plugin-onboarder subagent with the repo argument. It runs vet-candidate.sh, and if clean, returns a structured proposal: repo, name, a marketplace-quality description, homepage, license, keywords, and any blockers. If --description or --keywords were supplied by the caller, they override the subagent's curated values when passed to add-entry.sh in step 3.
  2. Stop on blockers. If the proposal is not ok, print each blocker and the fix and stop — make no edits. The most common blocker: the candidate still ships its own .claude-plugin/marketplace.json (the odere-pro name collision). Tell the user to remove it in that repo first (the same change made for claude-oop-excellence / claude-calibration), then re-run. See docs/adding-plugins.md.
  3. Insert the entry. Run bash ${CLAUDE_SKILL_DIR}/scripts/add-entry.sh <repo> [--name …] [--description "…"] [--keywords a,b], passing the curated description/keywords. The script re-vets, refuses duplicates, and writes the entry with jq --indent 2.
  4. Sync docs. Run bash ${CLAUDE_SKILL_DIR}/scripts/sync-readme.sh to regenerate the README Plugins table and bash ${CLAUDE_SKILL_DIR}/scripts/sync-site.sh to regenerate the Pages site/ (gate 18-pages-in-sync fails if site/ drifts from the manifest), then add a CHANGELOG.md bullet under ## [Unreleased] (e.g. - List \<name>`.`).
  5. Verify. Run bash tests/gates/run-all.sh and claude plugin validate . --strict. Both must be green before continuing.
  6. PR. Create a branch feat/add-<name>, git add the manifest + README + CHANGELOG + site/, commit feat: list <name> in the odere-pro marketplace (human-authored — no AI attribution), push -u, and gh pr create with a short body. Report the PR URL.

Boundaries

  • odere-pro repos only; never edits a candidate's own repository (block-with-instructions).
  • Touches only .claude-plugin/marketplace.json, README.md, CHANGELOG.md, and the generated site/.
  • Never sets version/sha/commit on an entry; the scripts and the marketplace-guard hook enforce it.

Failure handling

If any step fails (a blocker surfaces, a gate goes red, claude plugin validate errors, or the PR push fails), stop and roll back so the working tree is left exactly as it started — never half applied. Undo both the staged and the working-tree state of the paths this skill touches:

git restore --staged .claude-plugin/marketplace.json README.md CHANGELOG.md site/
git restore .claude-plugin/marketplace.json README.md CHANGELOG.md site/

git restore --staged … first unstages anything git added in step 6; the second git restore … discards the working-tree edits from steps 3–4. Then report what failed and what you reverted. Do not commit or push a partial change.

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.