Imported from WillEastbury/skillcli (plugins/skillcli-skill-one/skills/skill-one/SKILL.md). Install upstream with npx skills add WillEastbury/skillcli --skill skill-one. Copyright stays with the author.

Skill One: GitHub Intake

Purpose

Submit a proposed Markdown skill and any supporting files for review. The submission is a complete plugin package placed in the catalogue's own plugins/ tree on an isolated review branch, and a pull request proposes it for merge.

The package is submitted with review state pending. Submission does not approve, publish, merge, or install the skill. Maintainer approval and merge publish it.

Procedure

1. Run skillcli search --role skill-author --query "<proposed purpose>", or use native copilot plugin marketplace browse <marketplace> --json, to check for an existing approved plugin.

2. If the user selects an existing result, run skillcli install --skill <owner>/<repo>/<skill-id> and stop the submission flow.

3. Otherwise ask the user to confirm the target OWNER/REPO selected during Skill Two. If no repository was selected, stop and ask; never infer one.

4. Confirm the proposed lowercase kebab-case skill ID and plugin name.

5. Review the proposed plugin package and require plugin.json, skillcli.json, and skills/<skill-id>/SKILL.md. Accept EVIDENCE.md when Skill Two produced it.

6. Reject secrets, credentials, personal data, customer data, hidden files, symbolic links, binaries, and unrelated files.

7. Summarise the purpose, intended roles, categories, runtime, dependencies, capabilities, MCP servers, authentication, source, and licence.

8. Show the selected repository, branch, and pull request plan. The package is submitted to whichever repository the user selected, public or private, and never to both.

9. An explicit request to submit is approval to create the branch and pull request.

10. Create an isolated skill-intake/<skill-id>-<timestamp> branch in the selected repository.

11. Add the package under plugins/<plugin-name>/ and register it in .github/plugin/marketplace.json.

12. Record the pending review state and regenerate the catalogue views so the branch is mergeable and continuous integration passes:

    python tools/sync_plugin_metadata.py --plugin <plugin-name> --state pending
    python tools/render_readme.py
    python tools/validate_manifest.py

    Commit the regenerated skills.json, SKILLS.md, and README.md alongside the package. The catalogue table row appears in the pull request diff for review.

13. Commit and push the branch.

14. Open a pull request linking the immutable commit and compare view, with a file inventory and a maintainer checklist that includes flipping the review state:

    python tools/sync_plugin_metadata.py --plugin <plugin-name> --state approved \
      --reviewer "<maintainer>"
    python tools/render_readme.py

    Merging the approved pull request publishes the skill immediately.

Use the AI host's authenticated GitHub tools. Do not request a pasted token.

Boundaries

• Do not hand-edit skills.json, SKILLS.md, or the generated catalogue block in any README.md. Regenerate them with the repository tools so the row is derived from plugin metadata.
• Do not merge or approve the pull request.
• Do not set the review state to approved; submit as pending and leave approval to a maintainer.
• Do not install or execute the proposed skill.
• If the pull request cannot be created after the branch is pushed, return the branch and commit so maintainers can recover the submission.