Menu
Implement a tightly scoped piece of work from the current conversation, an explicit request, or a spec file.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Review the changes since a fixed point (commit, branch, tag, or merge-base) along two axes — Standards (does the code follow this repo's documented coding standards?) and Spec (does the code match the user's request, PRD, or explicit spec?). Runs both reviews in parallel sub-agents and reports them side by side. Use when the user wants to review a branch, a PR, work-in-progress changes, or asks to "review since X".
Disciplined diagnosis loop for hard bugs and performance regressions. Reproduce → minimise → hypothesise → instrument → fix → regression-test. Use when user says "diagnose this" / "debug this", reports a bug, says something is broken/throwing/failing, or describes a performance regression.
Use when the user asks for a rich explanation of a code change, diff, branch, or PR. Produces HTML output.
Walk the user through the changes since a fixed point (commit, branch, tag, or merge-base) as a guided, interactive code tour — the live Neovim counterpart to the `explain-html` skill. Drives the user's Neovim (scroll, highlight, virtual-text annotations) when connected, and degrades to prose with file:line references otherwise. Advances one stop at a time and waits for the user to say "next". Use when the user wants to be walked through a diff/branch/PR, asks for a guided tour or walkthrough of changes, or says "tour since X" / "lead me through the changes".
Build a throwaway prototype to answer a design question. Use when the user wants to sanity-check whether a state model or logic feels right, or explore what a UI should look like.
Use when you need to resolve an in-progress git merge/rebase conflict.
| name | implement |
| description | Implement a tightly scoped piece of work from the current conversation, an explicit request, or a spec file. |
| disable-model-invocation | true |
Implement the work described by the user in the current conversation, a directly supplied request, or an explicit spec file.
This repo does not use an agent-managed issue tracker as the default work-packaging layer. The human owns the roadmap and long-term work selection; the agent owns one tightly scoped task at a time.
CONTEXT.md, CONTEXT-MAP.md, ADRs) when they exist, so names and decisions match the repo./tdd where possible, at pre-agreed seams./code-review to review the work when the user wants a review pass..scratch/ only for temporary notes/specs when the user asks for a durable artifact or the task genuinely needs one./to-tasks; do not turn it into a ticket system.Do not commit. Leave changes staged or unstaged for the user to inspect.
If multiple commits would be advantageous, stop and explain the proposed split. Before doing any commit-oriented work, create a new branch from a clean slate: stash or otherwise clear unrelated work first, then branch. Never commit on main.