mattpocock-skills
claudeMatt Pocock's agent skills for real engineering (grilling, spec/ticket flows, TDD, code review, domain modelling and more).
39 skills
E2E issue resolution, Codex-assisted design, risk-scaled review, context-isolated investigation — language-agnostic harness engineering
Plugin ai-dev from marketplace ai-dev-templates in rioX432/ai-dev-templates. Commands read from the repo's manifest — 21 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.
auditAudit a codebase for evidenced tech debt, code quality, architecture, performance, security, dependency, and broken-UI risks, then optionally hand right-sized findings to the issue skill. Use for a repository or module health check; use ux-audit instead for task-flow, usability, or accessibility eva
clean-slopRemove redundant AI narration, change-history prose, and verbose comments from the current change's diff while preserving useful rationale and documentation. Use for an explicitly requested comment cleanup or as the cleanup pass of /dev or a PR workflow; not for logic refactoring or repository-wide
competitive-auditCore Value-filtered competitive analysis: research user pain points, identify gaps within Core Value scope, file at most 3 high-impact GitHub Issues and record the rejected ideas as Won't Do. Requires the gh CLI.
decomposeTurn a confirmed change into a short, ordered plan of verifiable subtasks: behavior slices with the test in the same step, prerequisites and spikes first, each with a Verify command and explicit dependencies. Use after investigation and ambiguity resolution, before the implementation loop. Not for s
dev-allProcess a batch of issues sequentially: /dev per issue in an isolated sub-agent → review validation → CI wait → conditional merge → next. GitHub-only; requires the gh CLI.
dev-investigateInvestigate the codebase for one issue in a forked context, writing a self-contained investigation-report.md. Use when /dev needs codebase investigation without spending the caller's context on file reads.
devEnd-to-end development for one issue: investigate → design → resolve ambiguities → decompose → implement → test → review → PR. Reads GitHub or Linear issues; writes branches and pull requests to GitHub via the gh CLI.
digResolve design ambiguities before implementation: keep only the decisions that would change the result, auto-decide what codebase patterns already settle, ask the user at most 5 questions with a recommendation each, and output a decision matrix with acceptance criteria. Use after investigation, befo
implement-guidanceImplement one confirmed, bounded change in the current working tree: read before editing, follow existing patterns, work subtasks in dependency order, run each subtask's Verify step, and report changed files, verification output, assumptions, and blockers. Use when the change and its subtasks are al
init-projectInitialize a new project with AI-driven development templates and layer-specific configuration
investigateInvestigate the codebase and report: trace data flows, map dependencies, assess impact. Use for standalone questions like \"how does X work\" or \"what breaks if we change Y\". Reports only — no implementation, no proposals. For investigation inside a /dev run, use dev-investigate instead.
issueWrite and size GitHub Issues so each one is a single agent-executable unit — one outcome, one PR, one proof. Use when filing an issue, when an audit/scan/research skill converts findings into issues, or when an existing issue looks too large to implement in one pass.
monitorKPI monitoring and issue proposal: analyze crash rates, store reviews, and metrics to suggest next priorities, then file them as GitHub Issues. Assumes a mobile app with store presence; requires the gh CLI.
orchestrateCoordinate several workers toward one goal that does not fit a single agent run: gate the fan-out decision, cut non-overlapping lanes, brief each worker with an objective, output format, tool guidance and boundaries, accept a return only on printed evidence, and checkpoint into resumable state files
prOpen a GitHub pull request for the current branch using the project PR template, with issue linking and a generated changelog. Use when the user asks to open a PR, or as the PR step of /dev once the branch is committed and reviewed. Requires the gh CLI.
reviewReview the current branch against its base with reviewer count and specialists chosen from semantic risk (fast: coordinator only; highRisk: independent reviewers), producing verified, deduplicated, severity-ranked findings. Use before committing or opening a PR.
syncSync common skills, agents, hooks, rules, and layer-specific files to target projects
thinkZero-base deep research: structured investigation → synthesis → proposal with counter-arguments
update-docsAudit docs for active-vs-stale authority, reconcile outdated instructions, and update architecture, changelog, README, or OSS docs
ux-auditAudit a user-facing flow or bounded set of screens using current-run visual, interaction, code, and accessibility evidence. Use for UX, accessibility, responsive, or platform-guideline reviews; report strengths, risks, verification gaps, and incremental fixes, then optionally hand selected findings
case-analyzerDeep dive analysis of a specific case: outcomes, success factors, and reception
counter-argumentStress-test proposals: find counter-arguments, logical leaps, and hidden risks
deep-researcherSupplement Deep Search results with web/SNS research. Collector only — no analysis
repo-analyzerAnalyze GitHub repository: code, features, Issues/PRs, external reviews
security-reviewerSecurity vulnerability scanner based on OWASP. Use during code review to check for security issues.
social-scannerScan X/Reddit/HackerNews/communities for sentiment and reception on a topic
source-verifierValidate all URLs exist and check claim-source consistency
test-writerGenerate unit tests for changed or new code. Use during development to ensure test coverage.
${CLAUDE_PLUGIN_ROOT}/scripts/block-dangerous-commands.sh
${CLAUDE_PLUGIN_ROOT}/scripts/block-secret-access.sh
${CLAUDE_PLUGIN_ROOT}/scripts/auto-lint.sh
${CLAUDE_PLUGIN_ROOT}/scripts/log-failure.sh
${CLAUDE_PLUGIN_ROOT}/scripts/save-context.sh
${CLAUDE_PLUGIN_ROOT}/scripts/restore-context.sh
${CLAUDE_PLUGIN_ROOT}/scripts/restore-context.sh
${CLAUDE_PLUGIN_ROOT}/scripts/log-subagent.sh
${CLAUDE_PLUGIN_ROOT}/scripts/log-subagent.sh
${CLAUDE_PLUGIN_ROOT}/scripts/log-subagent.sh
${CLAUDE_PLUGIN_ROOT}/scripts/log-subagent.sh
Matt Pocock's agent skills for real engineering (grilling, spec/ticket flows, TDD, code review, domain modelling and more).
39 skills
Harness-native ECC operator layer - 71 agents, 314 skills, 98 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
200 skills · 98 commands · 71 agents · 1 MCP server · hooks
Automated code review for pull requests using multiple specialized agents with confidence-based scoring to filter false positives
1 command
Use Alibaba Open Code Review from local Codex.
2 skills
Investigate incidents, review code and execute UAT for release readiness, scan code for vulnerabilities, and run penetration tests with AWS DevOps Agent and AWS Security Agent.
13 skills · 9 commands · 1 MCP server
Your AI development teammate. Turns a story file into a shipped, reviewed, and tested feature via orchestrator → architect → developer → PR reviewer → QA → Playwright.
1 skill · 2 commands · 6 agents
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.