Verify template

Skill jajupmochi/agent-harness/skills/general/verify-template

Linlin's curated AI agent harness configuration: workflow rules, skills, hooks, plugin recommendations, tooling preferences, and project templates. Loadable into any new project so a fresh /init can pick the relevant subset.

Install

npx -y skills add jajupmochi/agent-harness --skill verify-template

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Run the project's CI gates locally before claiming work is done. Customize the body to match the project's actual lint/test commands. Use before opening a PR or marking a phase complete.

SKILL.md

2.1 KB, as published. Nobody here has run it

/verify

Run the full verification gate for this project. Stop on first failure and report it plainly.

Customize this skill per project — the body below is a template. Replace with the project's actual commands. Common templates:

Default Python (uv + ruff + pytest)

uv run ruff check <SRC_DIRS>
uv run ruff format --check <SRC_DIRS>
uv run mypy <SRC_DIRS>            # if mypy is configured
uv run pytest -m "not slow and not download" -v

Replace <SRC_DIRS> with the project's source dirs (e.g., src/ tests/ or <package-name>/ tests/ plugins/).

Default Node.js (TypeScript)

npx tsc --noEmit
npx prettier --check .
npx eslint .                       # if eslint is configured
npx vitest run                     # or npx jest

Static site (HTML / CSS / JSON i18n)

# Validate JSON
for f in locales/*.json data/*.json; do jq empty "$f" || { echo "Invalid: $f"; exit 1; }; done

# i18n parity (if applicable)
python3 scripts/check_i18n_parity.py

# Optional: HTML validation
npx html-validate index*.html

Notes for the agent

Stop on the first failure. Report which step failed and the relevant tool output.
These should mirror the actual CI workflow — if ruff format --check fails, run ruff format to fix, then re-run /verify.
Don't run if there are no edits in flight (skip if working tree is clean and HEAD == upstream).
For research packages, tests should run on CPU and complete in <30 s if possible.

Customization checklist

When using this template in a new project:

Replace <SRC_DIRS> with the project's source directories
Adjust pytest flags (markers, parallelism)
Add or remove mypy step based on whether the project uses it
If pre-commit is configured, prefer pre-commit run --all-files over invoking ruff directly
Drop the templates that don't apply (e.g., remove the Node.js block in a Python project)