> ## Documentation Index
> Fetch the complete documentation index at: https://docs.plannotator.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Matt Pocock's AI Engineering Skills Profile

> A source-backed profile of Matt Pocock's AI engineering skills for specs, tickets, wayfinding, domain modeling, ADRs, and issue-centered planning.

Matt Pocock's public skills repository provides focused procedures that can fit into an existing engineering process. Rather than requiring one universal lifecycle, different skills produce specifications, issue tickets, uncertainty maps, domain language, and small decision records.

**Source stamp:** [`mattpocock/skills` at `9603c1cc`](https://github.com/mattpocock/skills/tree/9603c1cc8118d08bc1b3bf34cf714f62178dea3b), reviewed July 18, 2026. Repository license: MIT.

## Representative artifacts

| Skill           | Main output                                                                                                  |
| --------------- | ------------------------------------------------------------------------------------------------------------ |
| `to-spec`       | Problem statement, solution, user stories, implementation and testing decisions, out of scope, further notes |
| `to-tickets`    | Tracer-bullet tickets with outcome, acceptance criteria, and blockers                                        |
| `wayfinder`     | One map issue plus question issues that expose the known frontier and remaining fog                          |
| Domain modeling | Context glossary, avoided synonyms, context map, and small ADRs when a durable tradeoff exists               |

Issue-tracker output is the default model, with a local `.scratch/<feature>/issues/` fallback for ticket files. `to-tickets` favors vertical slices and uses expand, migrate, and contract for broad refactors that do not fit that grain.

## Workflow and task grain

`to-spec` focuses on decisions and testing seams. It generally avoids exact file paths and large code snippets unless a prototype excerpt carries an important decision. `to-tickets` turns the accepted specification into independently useful work with explicit dependencies. `wayfinder` handles earlier uncertainty by separating decisions, questions, agent research, human input, and work that is not precise enough to schedule.

The domain-modeling skill recommends a very small ADR only for a surprising, hard-to-reverse tradeoff. A separate `CONTEXT.md` can preserve preferred terms and terms to avoid.

## Inspect the same-case output

Inspect the [generated specification and 11 local tickets for reliable webhook delivery](https://github.com/plannotator/spec-planning-frameworks/tree/main/examples/matt-pocock-skills/raw). The run completed the specification and dependency-aware local-ticket flow without creating remote issues.

## Choose these skills when

* the team wants composable procedures rather than a required end-to-end method;
* an issue tracker is the shared planning memory;
* uncertain work should be mapped before it becomes implementation tickets; and
* outcome-focused tickets are more useful than code-level instructions.

The official repository describes these as skills for engineering work. A real `CONTEXT.md` example exists in the linked Course Video Manager project, but that repository has no clear reuse license; this site links and paraphrases it rather than copying it.

See the [AI coding plan template](/templates/ai-coding-plan), [ADR formats](/frameworks/adr-formats), and [coding-agent handoff template](/templates/agent-handoff).

*Reviewed July 19, 2026. Researched and maintained by the Plannotator documentation team. Matt Pocock's name is used for identification; Plannotator is not affiliated with or endorsed by him.*