---

name: contribution-and-project-maintenance description: "Add, fix, review, and triage mini-project folders while keeping README, requirements, assets, and repository hygiene aligned with the repo's contribution style." disable-model-invocation: true metadata: disco-role: operating license: MIT

Contribution and Project Maintenance

Use this sub-skill when the task is about adding a new mini-project folder, fixing an existing one, reviewing a PR for repo hygiene, or triaging README, requirements, asset, and cache issues.

Use when

• A new mini-project folder needs a safe starter scaffold.
• An existing project README needs alignment with the repo template.
• Project-local dependencies, run steps, or asset paths need cleanup.
• A PR should be checked for naming, cache noise, or stale links.

Do not use when

• The task is to run or debug a GUI, web, network, data, or ML demo.
• The request is mainly about runtime behavior rather than static maintenance.
• The change belongs in a sibling sub-skill that owns execution or backend-specific work.

Workflow

1. Read references/contribution-workflow.md for the add/fix/review checklist and PR expectations.
2. Read references/project-template.md to shape README.md, file names, and optional requirements.txt.
3. Read references/troubleshooting.md when the folder has missing files, broken requirements, stale badges, or cache noise.
4. Use scripts/create_mini_project_skeleton.py --help before creating a new starter folder.
5. Keep changes inside the target project folder unless the task explicitly calls for repo-root or .github/ edits.

Boundaries

• Create or repair README.md, main.py, and optional requirements.txt inside the project folder.
• Use README.md exact case for new work; keep legacy case variants only if the task is explicitly preserving them.
• Keep assets local and referenced with relative paths.
• Remove generated caches and notebook checkpoints from review changes.
• Avoid touching the root requirements.txt as a shared dependency list.
• Escalate GUI, web, data, ML, or destructive runtime work to the matching sibling sub-skill.

Expected result

A good contribution leaves the project folder easy to review: clear name, readable README, project-specific dependencies, no generated cache noise, and no stale or unsafe links.