Engineer production-ready code with AI agent skills

A five-step process, sourced from mattpocock/skills (MIT), for breaking a plan, spec, or PRD into independently-grabbable issues on the project's issue tracker using tracer-bullet vertical slices rather than horizontal layer slices. It gathers context from the conversation or a referenced issue, fetching its full body and comments, optionally explores the codebase to align issue vocabulary with the project's domain glossary and relevant ADRs and to look for prefactoring opportunities under the principle "make the change easy, then make the easy change," then drafts vertical-slice issues where each slice cuts through every integration layer end-to-end - schema, API, UI, tests - and is independently demoable or verifiable, with any needed prefactoring done first. It then presents the proposed breakdown as a numbered list showing title, blocking dependencies, and covered user stories per slice, and explicitly asks the user whether the granularity feels right, whether dependencies are correct, and whether slices should merge or split, iterating until approved. Finally it publishes each approved slice as an issue in dependency order, blockers first, so real issue identifiers can be referenced in "Blocked by," using a fixed issue-body template covering the parent reference, a description of end-to-end behavior rather than layer-by-layer implementation, a checklist of acceptance criteria, and blocking references - explicitly warning against inlining specific file paths or code snippets since they go stale fast, with a narrow exception for a prototype-derived snippet that encodes a decision such as a state machine, reducer, schema, or type shape more precisely than prose can. Published issues get the correct triage label since they are considered ready for AFK agents, and the skill never closes or modifies a parent issue.

Use it when a user request matches breaking a plan, spec, or PRD into independently-grabbable tracer-bullet issues on the project's issue tracker. It requires the issue tracker and triage-label vocabulary to already be configured - if not, the instruction is to run a separate /setup-matt-pocock-skills command first rather than guessing conventions.

Input is a plan, spec, PRD, or an existing issue reference, a number, URL, or path, to fetch and read in full. Output is a user-approved numbered breakdown followed by published issues on the tracker, each following the fixed template:

### What to build

A concise description of this vertical slice. Describe the end-to-end behavior, not layer-by-layer implementation.

### Acceptance criteria

- [ ] Criterion 1
- [ ] Criterion 2
- [ ] Criterion 3

### Blocked by

- A reference to the blocking ticket (if any)

Publishes to the project's configured issue tracker using its triage-label vocabulary, set up via a separate /setup-matt-pocock-skills command; grounds issue titles and descriptions in the project's domain glossary and existing ADRs.

Engineers and coding agents turning an approved plan or spec into a dependency-ordered, independently-workable issue backlog, especially one meant to be picked up by unattended (AFK) agents.