mattpocock-skills
claudeMatt Pocock's agent skills for real engineering (grilling, spec/ticket flows, TDD, code review, domain modelling and more).
39 skills
AI-driven development workflows — planning, implementation, testing, review, and release
Plugin nexus from marketplace claude-skills in nexus-a1/claude-skills. Commands read from the repo's manifest — 2 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.
add-product-knowledgeAdd a new entry to the product knowledge base. Wizard-guided — prompts for category, title, and content, then writes a structured markdown file and rebuilds the manifest.
archive-requirementsManually archive completed requirements to team knowledge base
brainstormExplore implementation strategies for business requirements. Interactive brainstorming that presents multiple approaches, trade-offs, and creates high-level implementation picture before committing to detailed specs.
bug-stubTurn a free-text bug description into a paste-ready ticket stub — a short title and a one-paragraph summary — with no code discovery, no fix, and no commit. Use right after finding a bug, before any tracker ticket exists, to get text you can file yourself and then feed into no-ticket requirements ga
commitStage and commit changes with a conventional commit message. Runs pre-commit checks automatically. Prefixes the ticket number when there is one, asks once when there is not, and commits without it if none is given.
configuration-initInitialize project configuration file with interactive wizard. Also supports `validate` and `migrate` modes for existing installs.
create-proposalCreate a formal design document for a feature or component. Guides through requirements, approach brainstorming, and iterative proposal drafts with approval gate before optional implementation. Use when you need a written design approved before committing to code.
create-release-branchCreate a release/vX.Y.Z branch from the repo's default branch (default), any branch, or a specific tag using [email protected] syntax. Takes a version and an optional source. Step 1 of the release workflow.
create-releasePush a release branch and open a PR to the target branch. Takes an optional target branch and version. Step 2 of the release workflow — runs after branching, before merging.
create-requirementsRun a multi-agent pipeline to produce detailed technical requirements and a ticket-ready summary. Creates a feature branch, persists session state, and supports resume. Optionally seeds from a prior brainstorm or meeting session, takes the ticket from --ticket <KEY> and auto-fetches its Jira descrip
epicDecompose large initiatives into dependency-mapped, wave-sequenced tickets with per-ticket requirements. Use when a feature is too large for a single /create-requirements run — typically 5+ tickets with complex interdependencies.
explainExplain one thing in plain English — a file, a function, an error message, a config key, a ticket, or a concept. Says what it is, how it works and what to watch out for, in a few short sections with no filler. It never edits, commits or fixes anything; the one command it runs is the shared Jira read
feedbackGenerate a retrospective report analyzing agent pipeline execution, duplication, scope adherence, and output quality from a completed work session.
implementImplement a feature from saved requirements. Chunk-based commits, parallel QA (tests + review + security), and PR creation. Resumes interrupted sessions from saved state. Runs in the current working tree by default — set `worktree.enabled: true` in `.claude/configuration.yml` to isolate work in a gi
jiraRead a Jira work item and its recent comments from the terminal via the Atlassian CLI (acli). The key is optional — with no arguments, proposes one from the session and asks you to confirm before reading. Can also add a comment, transition a status, assign/unassign, or create a new work item — each
load-contextLoad all available context for a ticket or topic — work state, brainstorms, proposals, requirements KB, product knowledge, and git history — into a single unified summary.
load-requirementsLoad a specific archived requirement from the requirements KB for detailed review. For in-flight tickets that have not been archived yet, use /load-context instead — it also covers work sessions, brainstorms, and proposals.
meetingLive meeting companion — capture notes as the meeting happens while background probes ground each topic against the Product Knowledge Base and the live codebase, surfacing relevant findings inline without stalling capture. On wrap it emits two distinct professional documents (a shareable summary and
merge-releaseMerge an approved release PR into its target branch. Takes the release branch or the version it is for. Validates approval/checks/conflicts via gh, then merges. Step 3 of the release workflow.
monitor-prMonitor an open pull request — watch CI, investigate and fix failing workflows, address new review comments, and loop until the PR is approved, merged, or the iteration limit is reached.
pr-reviewReview a pull request (or local branch with --local) with thorough analysis, measured severity, a read-only second opinion on every serious finding (Opus by default, Fable with --ultra), and actionable feedback; --auto-fix loops review, fix, re-review until nothing verified is left to address
rebuild-indexRebuild manifest.json for any artifact storage type. Scans directories and regenerates from scratch.
rebuild-requirements-indexRebuild search index for requirements repository
refactorAnalyze code and suggest refactoring improvements with agent-driven analysis. Runs in the current working tree by default — set `worktree.enabled: true` in `.claude/configuration.yml` to isolate work in a git worktree.
releaseCreate a GitHub release with a version tag and LLM-authored release notes. Takes an optional version and source branch; --pre-release marks it a pre-release, --fasttrack (-y / --yes) skips the confirmation prompt. Final step of the release workflow.
report-issueDraft and submit a bug report or feature request to the nexus plugin repository, using current conversation context to auto-populate details.
resume-workResume any interrupted work session — brainstorm, requirements, proposal, epic, or implementation. Scans for incomplete sessions and continues from the last saved checkpoint. Accepts a ticketed identifier (JIRA-123-slug), a provisional draft (DRAFT-slug, offers to reconcile with a real ticket), or a
review-planValidate an ad-hoc implementation plan through architect and quality-guard (and optionally security-auditor), then output a revised plan with adjustments applied.
search-requirementsSearch team requirements repository for similar past work
second-opinionPut a conclusion the session has already reached in front of a different model and report back what it says. Packages the claim, the rejected alternatives, and pointers to the primary sources into a neutral brief, dispatches the read-only second-reader agent (Fable by default), and reports the verdi
standupAnswer "where do we stand right now" in a few lines — what landed, what is in flight, what is waiting on you, what is next. Checks git, open PRs and work sessions, and keeps what it verified separate from what it could not.
todo-workPick a pending or in-progress task from the task store, mark it in progress, and hand it off to /review-plan, /create-requirements or /implement. A task promoted to /create-requirements is linked to the session it becomes.
todoAdd a task to the project's task store from just a title; priority, category and scope get defaults and can be set later with /todo edit <number|id>. Also lists open tasks, edits one (edit), closes one (done), and imports an existing TODO.md once (migrate).
troubleshootSystematically troubleshoot a failing feature or error. Discovers code, investigates root cause, applies fix, verifies with tests, and commits. Use when something isn't working as expected. Runs in the current working tree by default — set `worktree.enabled: true` in `.claude/configuration.yml` to i
update-contextAnnotate an active work session with a note, scope change, or new finding. Auto-detects the active session, synthesizes the salient points of the current conversation, and appends a timestamped entry to state.json after a single target confirmation. Use mid-session when you learn something that shou
update-documentationReview and update project documentation using an agent team. Inventories docs, identifies gaps and drift, updates technical and API docs in parallel.
work-statusShow all active work sessions across brainstorms, requirements, proposals, and epics. Supports --brief for a narrative digest with latest updates and suggested next actions, --update to advance lifecycle on one session, and --sync to sweep them all.
archaeologistAnalyze code patterns, data flow, and modification risks for a feature. Complements context-builder by going deep, not wide.
architectValidate implementation plans, and the diff of a pull request, against the project's architecture rules and existing patterns; also contributes architecture constraints during requirements research. Read-only — reports each violation with a specific fix and never rewrites the plan or the code.
archivistStore, search, and retrieve team requirements from the requirements knowledge base — searches past tickets during requirements research, archives completed work after implementation (commits it; pushes only a git-backed knowledge base), and rebuilds the index. Refuses when no requirements repository
aws-architectAWS architecture and infrastructure guidance. Use for infrastructure decisions, CloudFormation review, cost optimization, and multi-service design.
business-analystAnalyze requirements and consolidate findings from other agents. Decision maker in requirements phase.
code-reviewerSingle-pass code review focused on real issues. No subjective nitpicks. Includes performance analysis and investigation.
context-builderBuild structured context inventory for requirements gathering. First agent in the requirements pipeline.
data-modelerAnalyze database schema, entities, and relationships for requirements gathering.
database-analystExecute database queries and analyze data patterns. Use when the user needs to query databases, explore schemas, analyze data distributions, or understand database relationships. Returns concise summaries and key findings.
doc-writerWrite technical and API documentation. Use for new features, undocumented code, or API endpoints.
git-operatorResolve merge conflicts, drive complex rebases, and author PR bodies from large commit ranges. Not used for routine commit/push/branch-create — those run inline via Bash under the `git-mutation-guard.sh` hook.
integration-analystAnalyze external API integrations and implement API clients. Use for requirements gathering and API implementation.
playwright-engineerWrite and maintain end-to-end tests using Playwright. Expert in locator strategies, test isolation, POM patterns, and cross-browser testing.
product-expertProvide context and guidance from a project's product knowledge base. Use when working on features that have product-specific documentation. Needs `storage.artifacts.product-knowledge` configured in `.claude/configuration.yml`; without it the agent reports the knowledge base as unconfigured and retu
quality-guardContrarian challenger agent that independently verifies claims, refuses unsubstantiated assumptions, and forces other agents to produce evidence-backed, high-quality output.
refactorerApply behavior-preserving refactorings — analyzes impact, edits the files, runs the test suite, rolls back on failure and commits on success. Tracks multi-step work in progressive sessions. Changes structure only; does not add features or fix bugs.
second-readerRead-only reviewer for a second opinion on conclusions or findings, meant to run on a different model than the session that produced them. Verifies against the sources, answers the inverse question, and reports a verdict without adopting or defending anything.
security-auditorSecurity audit for code changes including PII/sensitive data scanning, plus a config-audit mode that scans a project's .claude/ configuration for permission, hook-injection, CLAUDE.md-injection, MCP supply-chain, and unscoped-tool risks. Use for payment, auth, or sensitive data code. Use before comm
security-requirementsIdentify security and compliance requirements early in the requirements phase.
test-fixerDiagnose failing tests with root cause analysis and fix the test, fixture or mock when the test is what is wrong. Detects flaky tests. When the production code is wrong it reports the bug instead of changing that code.
test-writerWrite comprehensive unit and integration tests. Coverage-aware with framework-specific patterns.
${CLAUDE_PLUGIN_ROOT}/hooks/git-mutation-guard.sh
${CLAUDE_PLUGIN_ROOT}/hooks/validate-commit.sh
${CLAUDE_PLUGIN_ROOT}/hooks/redact-output.sh
${CLAUDE_PLUGIN_ROOT}/hooks/reverse-substitute.sh
${CLAUDE_PLUGIN_ROOT}/hooks/read-guard.sh
${CLAUDE_PLUGIN_ROOT}/hooks/message-guard.sh
${CLAUDE_PLUGIN_ROOT}/hooks/audit.sh
python3 ${CLAUDE_PLUGIN_ROOT}/hooks/output-guard.py
${CLAUDE_PLUGIN_ROOT}/hooks/auto-context.sh
${CLAUDE_PLUGIN_ROOT}/hooks/notify.sh
${CLAUDE_PLUGIN_ROOT}/hooks/messaging-context.sh
${CLAUDE_PLUGIN_ROOT}/hooks/messaging-context.sh
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
Official Next.js skills: adopt and optimize Cache Components, adopt Partial Prefetching, and verify runtime behavior against a running dev server.
6 skills
Persistent memory for Claude Code: the next session starts where the last one left off.
29 skills · 1 MCP server · hooks
tasteskill 2.5.1: professional company websites at the figma.com and x.ai level, with one section per screen, real photographs, real product UI, effects kept for tech and gaming pages, and a real-browser checker and finish gate.
1 skill
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.