mattpocock-skills
claudeMatt Pocock's agent skills for real engineering (grilling, spec/ticket flows, TDD, code review, domain modelling and more).
39 skills
Automated development workflow orchestration with org-mode tracking
Plugin workflow from marketplace zb-ss-workflow in zb-ss/claude-plugin-workflow. Commands read from the repo's manifest — 4 days ago.
This plugin runs code on your machine — its hooks fire automatically on events like tool use. The exact commands are listed below; read them before installing.
autoAutonomous driver — outer loop that wakes on a schedule, resumes or advances the in-flight task for a repo, then pulls and launches the next queued task. Reconstructs all state from disk and the GitHub issue on every wake; never relies on session memory.
doTake a task in plain words from plan to reviewed change. Runs the plan workflow, asks you its questions, then runs implement, which reviews, fixes and signs off the work.
migrate-legacyAssign or archive unscoped legacy workflow state files (created before repo-scoping) so they stop leaking into every repo's session.
commonCommon instructions shared by all workflow phase agents - tool usage, state updates, context referencing
completionCompletion guard procedures - final verification checklist, post-approval cleanup, learnings extraction
epic-orchestrationEpic workflow orchestration - multi-component projects with worktree isolation and dependency ordering
implementationImplementation phase procedures - fix protocol, code writing rules, review issue handling
planningPlanning phase procedures - plan output format, architecture analysis, risk identification
post-merge-reviewN-reviewer fan-out review-gate engine. Scales a roster of independent review lenses, adversarially verifies every finding, routes confirmed issues to fixes, and loops until dry under zero tolerance. Used by the code_review, security_review, quality, and post_merge_review gates.
quality-gateQuality gate procedures - build/type/lint/test pipeline, auto-fix loop, changes signaling
reviewReview-lens contract - finding format, confidence+severity, re-review protocol. Reviewers report everything; the fan-out engine verifies and decides the gate verdict.
queue/workflow:queue — manage the autonomous task queue backed by a private GitHub-issues repo
resumeResume an interrupted or paused workflow from its persisted repository-scoped state.
setup-statuslineEnable or disable the workflow usage status line
setupConfigure Claude Code permissions, initialize workflow directories and fix the startup setup reminders.
startStart a long-running autonomous workflow in swarm or epic mode for work that warrants a full execution pipeline.
test-e2eRun E2E Playwright testing workflow for web applications
test-liveInteractive live testing of a running web application via Playwright MCP browser automation
verifyRun verification loop on current changes
architectJudge — turns a task plus exploration findings into a complete implementation plan, or a list of open questions when the plan cannot be completed honestly
codebase-analyzerWorker — extracts a project's stack, conventions and patterns into a reusable context file that later agents read
completion-guardJudge — final sign-off before work is called complete; verifies every requirement is delivered, nothing is partial, and build and tests pass
e2e-explorerWorker — explores a running web app with browser automation and writes an app map (pages, forms, buttons, navigation, auth) for test generation
e2e-generatorWorker — generates @playwright/test specs, auth setup and page objects from an app map
e2e-reviewerWorker — reviews and runs Playwright E2E tests for flakiness, selector quality and isolation; returns a PASS/FAIL verdict with [ISSUE-N] findings
epic-integratorWorker — merges component branches in dependency order, resolves conflicts by intent, and runs the full build and test suite
executorWorker — implements a planned work item or a list of review fixes, matching the surrounding code
explorerWorker — fast read-only reconnaissance of a codebase area; answers specific questions with file references
finding-mergerMechanical — groups review findings that describe the same defect, so each defect is verified and fixed once
quality-gateWorker — runs the build/type/lint/test/audit pipeline, auto-fixes deterministic failures, and reports what still fails as [ISSUE-N] findings; never the gate verdict
reviewer-deepJudge — code-quality review lens; reports every candidate finding in the changed code for independent verification, never a gate verdict
scopeMechanical — resolves which files a review covers and classifies the change's risk; runs one read-only command and reports its result
security-deepJudge — security review lens; OWASP-complete pass over the changed code, reports every candidate finding for independent verification, never a gate verdict
spec-conformanceJudge — acceptance-criteria check; maps each criterion to a CRITERION block with evidence; PASS only if zero FAIL
supervisorOrchestrates workflows - NEVER implements directly, only delegates
test-writerWorker — writes and runs tests for changed code, following the project's test conventions
verifierWorker — read-only skeptic; judges whether each reported finding is a real, reachable defect in the code, quoting the line each verdict rests on, and defaults to refuted when unsure
web-testerWorker — tests a running web application live in a browser, guided by instructions or exploratory, and reports pass/fail with evidence
node "${CLAUDE_PLUGIN_ROOT}/hooks/scrub-guard.js"
node "${CLAUDE_PLUGIN_ROOT}/hooks/stop-guard.js"
node "${CLAUDE_PLUGIN_ROOT}/hooks/session-start.js"
node "${CLAUDE_PLUGIN_ROOT}/hooks/tier-check.js"
node "${CLAUDE_PLUGIN_ROOT}/hooks/subagent-stop-track.js"
node "${CLAUDE_PLUGIN_ROOT}/hooks/task-completed-gate.js"
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
Cross-session memory and token savings for coding agents.
6 skills · 1 agent · 1 MCP server · hooks
Up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your coding agent's context.
1 MCP server
Reliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer
8 skills · 1 MCP server
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.