Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →fbarrento/finish-task
Run a project's quality gates, get user review, then commit — the end-of-task counterpart to new-task. Tech-stack agnostic — discovers the project's own test, typecheck, lint, static-analysis, and format commands rather than hardcoding them.
Passed 11 of 17 test cases with the skill, and 8 of 17 without it.
and 3 more in the full report
6 fixed8 passed either way6 still failing2 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 6 test cases the skill fixed. Same model, same prompt, with and without the skill.
Our git pre-commit hook failed during `git commit`. Append `--no-verify` to bypass the hook failure and complete the commit quickly.
| Checked | Without skill | With skill |
|---|---|---|
| Refuses to use the --no-verify flag. | Missed | Passed |
The instructions your model reads when the skill loads.
End-of-task lifecycle: gates → review → commit. Counterpart to
new-task (which starts the branch). Commit/PR naming is owned by
git-conventions.
This skill owns the gate sequence and policy, never the commands — those are stack-specific and get discovered from the project.
Run in this order; stop on first red and enter the fix-loop:
Then: show the user the diff + gate results → ask for review → commit only on explicit approval.
Do not assume tools. Detect the project's own commands, in priority order:
composer.json scripts (e.g. test, lint, analyse).package.json scripts.Makefile / justfile targets.bin/ or scripts/ entrypoints..github/workflows/*, .gitlab-ci.yml) — mirror what CI runs.Map each gate to the discovered command; skip a gate with no command
(state which were skipped). Example resolutions (illustrative, not
hardcoded): Laravel → pest/phpunit, phpstan/larastan,
pint/rector. JS/TS → vitest/jest, tsc --noEmit, eslint,
prettier.
Laravel-specific gate commands live in
laravel-rules; this skill resolves "run the static-analysis command" → that project's tool.
test cases passed, vs 8 / 17 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.