mattpocock-skills
claudeMatt Pocock's agent skills for real engineering (grilling, spec/ticket flows, TDD, code review, domain modelling and more).
39 skills
Shared task records and roles for agent work: Markdown records of work, candidates, evidence, and assessments, with four role presets, reusable skills, and checks against each task's declared requirements.
Plugin castwork from marketplace castwork-dev in bartoszarendt/castwork. Commands read from the repo's manifest — 2 days ago.
assessmentUse when reviewing or auditing an implementation against its task record and recording a verdict, and when responding to a verdict you received. Covers the three lenses, independence, and what accept, reject, and needs_revision mean.
blocked-stateUse when work hits a wall it cannot clear on its own — a missing credential, an outage, a contradictory requirement, a decision only the user can make — and the pause needs to be durable enough for someone else to resume.
decision-captureUse when a choice is made that future work should not silently reverse — an architecture, a dependency, a convention, a deliberate tradeoff. Covers what is worth recording and where it goes.
task-record-contractUse when writing or revising a task record — deciding its intent, scope, acceptance criteria, and which requirements to declare. Covers the frontmatter fields, the status vocabulary, and what each requirement kind actually demands.
verification-evidenceUse when recording what a check actually did — after running tests, a linter, a build, or a manual inspection. Covers candidates, evidence entries, and the difference between what was observed and what was claimed.
/castwork:startRead the project's working policy, the documents it points to, and open task records, report where things stand, and continue with the requested work or propose the next step.
coordinatorDecides which roles act next and keeps the user informed. Produces no result and assesses nothing.
thinkerExamines the problem, its assumptions, and the approach, before, during, or after the work: investigates and researches, shapes task records, plans and breaks down work, critiques partial results, diagnoses stalls and disputes, reviews how completed work went, and recommends a way forward. Produces
verifierAssesses a specific result, the exact recorded candidate against its task record or what a brief names against what it asks, for responsiveness, completeness, and correctness, and gives a verdict with findings. Read-only on the result: changes nothing and fixes nothing.
workerProduces the result of scoped work, such as code, documents or a research write-up, and runs its checks; on a Castwork task record, also records the candidate and evidence. Its own acceptance is never independent.
Matt Pocock's agent skills for real engineering (grilling, spec/ticket flows, TDD, code review, domain modelling and more).
39 skills
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
15 skills · hooks
Persistent memory for Claude Code: the next session starts where the last one left off.
29 skills · 1 MCP server · hooks
AI-powered codebase understanding — analyze, visualize, and explain any project
9 skills · 10 agents · hooks
Open-source AI job search that runs locally in your AI coding CLI: scan job portals, evaluate listings into a structured A-H report with a 1-5 score, tailor your CV, track applications
5 skills · hooks
Cross-session memory and token savings for coding agents.
6 skills · 1 agent · 1 MCP server · hooks
This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.