About implement
A coding skill for implementing work from user specs or tickets. It guides agents through testing, review, cleanup, and committing changes.
Key features
- 01Use TDD at agreed seams
- 02Run typechecks and tests regularly
- 03Review work at the assigned depth
- 04Stop processes started during the task
- 05Commit changes to the current branch
Use cases
- 01Implement a ticket with a test-first approach
- 02Verify changes with project tests and typechecking
- 03Review and commit completed work
Install
Copy the command and run it in a terminal.
npx skills add https:/After installing, type/implementin Claude Code to confirm it loaded.
Bundled filesagents/openai.yamlLICENSE.mattpocock
Details
From the project READMEImplement the work described by the user in the spec or tickets.
Use /tdd where possible, at pre-agreed seams. With no one to agree them with, choose the seams yourself and name them in your report.
Run typechecking regularly (when the project has a typechecker), single test files regularly, and the full test suite once at the end.
Once done, review the work at the depth the delegating agent sets: none, one light pass of your own, or /code-review. With no delegating agent, use /code-review.
Before you report, stop everything you started: dev servers, preview and browser tabs, background commands and watchers. Name anything you leave running on purpose in your report, with the reason. In zsh, split a list with ${=VAR} or an array, and run a formatter only on a file list you've checked isn't empty: with none, it waits on stdin forever.
Never run rm -rf: move what's no longer needed (a temporary folder, a throwaway script) into the gitignored .scratch/ with mv.
Commit your work to the current branch, unless the agent that delegated the work to you commits it.