mattpocock-skills
claudeMatt Pocock's agent skills for real engineering (grilling, spec/ticket flows, TDD, code review, domain modelling and more).
39 skills
5-phase feature lifecycle toolkit — discover, design, sprint, review, release. From idea to production with Stitch MCP design, iteration loops, and quality gates.
Plugin cks from marketplace cks-marketplace in cardinalconseils/claude-starter. Commands read from the repo's manifest — 5 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.
ad-creativeanalytics-trackingbrand-marketingcold-emailcompetitor-alternativescontent-strategycopywritingcustomer-researchemail-sequencelaunch-strategylead-magnetsmarket-mappingmarketing-psychologypage-cropaid-adspositioningseo-auditsignup-flow-crosocial-contentab-test-setupaccessibilityaeo-geoagent-build-sequenceagent-personaagent-safetyagentic-os-builderagentmemoryagile-eagleahe-evolutionai-agent-projectsai-marketingai-seoanti-patternsapi-designapi-docsarchitectureattractorauthenticationauto-modeautoresearchcachingcampaigncanarycavemanchannel-setupchief-of-staffchurn-preventioncicd-startercisocks-stackcli-generationcode-excellencecode-simplificationcomplianceconcept-evaluationcontext-researchcontractscontrol-planecoordinationhardeningimprovementsmemoryobservabilityconversation-statecopy-editingcore-behaviorsdatabase-designdatabase-recoverydebugdecision-memodeep-researchdesign-fluencydesign-studiodesign-systemecosystem-watchenvironment-managementevalsexecutive-assistantexpertsexpert-builderexpert-debuggerexpert-productfailure-taxonomyfinopsgithub-issuesgithub-project-setupgrill-meguardrailshandoffharness-evalsheadroom-integrationhoncho-memoryidea-validationideationincremental-implementationjev-routingkarpathy-guidelineskickstartkpi-architectlanguage-ruleslearningslegibilityloopluv-model-routingmarketing-ideasmarketingmcp-server-authoringmigrationsmonetizemonitoringno-codeonline-marketingopenrouteroperating-modelorchestrationpaymentspeersperformancephoto-directionprdpricing-strategyproduct-marketing-contextproduct-marketingproduct-maturityprogrammatic-seoprompt-cachingremotionretrospectiverevopsroutinesrpisaas-dashboard-sequencesales-enablementscale-ladderscheduled-agentsschema-markupsecurity-hardeningsentry-sdk-setupseo-localshipping-checklistsite-architectureskill-creatorslacksleep-cyclestrategic-frameworksstrategic-optionstesting-disciplineuatultrareviewuser-memoryuser-profilevideo-ai-directionvoice/cks:README/cks:adoptAdopt CKS into an existing codebase mid-development — scans git history, generates CLAUDE.md, creates feature entry at sprint phase, detects secrets
/cks:agentic-osScaffold and manage an Agentic OS — architecture + memory + observability — inside any project
/cks:agentsShow active agent sessions, claimed resources, and conflicts — multi-session coordination
/cks:analyticsAnalytics tracker — GA4 event taxonomy, GTM setup, ad pixel checklist, tracking gap audit by maturity stage
/cks:architectureGenerate or refresh ARCHITECTURE.md — project topology diagram, component table, ADR index
/cks:assessAssess any existing codebase — health, code review, security audit, and debug triage via the Attractor pipeline
/cks:assistantExecutive assistant — inbox triage, calendar review, reply drafts, meeting prep. Drafts only; nothing is sent. Dispatches the assistant role.
/cks:autonomousRun all 5 phases autonomously — discover → design → sprint → review → release. No interruption.
/cks:autoresearchAutonomous keep/discard optimization loop — runs overnight, ratchets a single metric
/cks:bgLaunch any CKS command as a background session — appears in Agent View
/cks:board-exportExport board state (features, sessions, notes, activity) as a committed markdown summary
/cks:board[legacy] Launch the CKS Kanban board — visual dashboard for your project lifecycle
/cks:bootstrapScaffold project — .claude/, CLAUDE.md, .prd/, rules, deploy config
/cks:brainstormOpen-ended brainstorming — ideas, features, pivots, business models — seeded from current project context. Enters plan mode after ideation so output can be reviewed before routing to kickstart or feature lifecycle.
/cks:browseBrowser automation — open websites, fill forms, click buttons, screenshot, extract data. Auto-escalates to investigator when issues are found.
/cks:campaignCampaign orchestrator — intake, specialist dispatch, and artifact generation for outbound email, product launch, ABM, and content/paid campaigns
/cks:canaryPost-deploy browser verification — checks for console errors and page failures
/cks:cavemanCaveman mode — default CKS voice. Compress agent output. Brain still big. Mouth small.
/cks:cccs-intelCCCS threat intelligence monitor — setup, run, status, and stop the scheduled CCCS alert feed to Telegram
/cks:changelogAuto-generate a CHANGELOG.md entry from git history
/cks:chiefChief of staff — triage inbound work, cap it at three priorities, dispatch specialists, and report one brief
/cks:cisoPersonal CISO — audit repos and infra for supply chain attacks, secrets exposure, RLS gaps, webhook exposure, and GitHub Actions hardening
/cks:cks-wikiRead and write wiki pages in the project memory layer (memory/wiki/)
/cks:codegraphGraft codebase context graph — install, init, build, check, blast, viz, or uninstall the opt-in graph that backs codebase exploration
/cks:complianceCompliance surface scan — detect GDPR, PCI, HIPAA, SOC 2 obligations from feature context
/cks:conceptEvaluate a concept for CKS plugin (or project) feasibility — enters plan mode first, then reads external resources (URLs, GitHub repos, articles, transcripts), brainstorms, scores across business value, technology fit, and data impact. Produces FEASIBILITY.md.
/cks:contextResearch a topic, library, or API and save a persistent context brief to .context/
/cks:control-planeManage the CKS v6 control plane — init, health status, backup, restore, drain sync queue, reset
/cks:copyCopywriter — hero copy, email sequences, ad copy, landing pages using PAS/AIDA/BAB/FAB frameworks adapted to brand voice and ICP
/cks:costShow session cost breakdown, dev time, and tool-call metrics from control plane observability
/cks:creativeLuv creative suite — dispatch copywriters, brand strategist, photo creator, and video creator directly without going through the full agency hierarchy
/cks:dbDatabase operations — investigate schema/RLS, fix issues, debug errors, generate ERD, or run the audit+fix pipeline
/cks:debugDebug anything — trace app runtime errors, GitHub issues, or diagnose CKS skill/agent/command issues
/cks:decideStop asking — diagnose and act
/cks:deployDeploy project to configured Railway environment with health checks
/cks:design-systemGenerate a DESIGN.html — interactive HTML design system with rendered swatches, live type specimens, and mini-site nav
/cks:designPhase 2: Design — UX, API contract, screens, component specs; or studio websites and decks rendered and checked with Playwright, REA and Figma
/cks:discoverPhase 1: Discovery — structured requirements gathering with the 11 Elements
/cks:docsGenerate or update project documentation — API docs, architecture, component docs, onboarding guide
/cks:doctorRun a full project health diagnostic — env vars, TODOs, tests, PRD state, git hygiene
/cks:eodEnd of day — summarize today's work into a dated DEVLOG entry with state and next steps
/cks:evalsRun LLM output quality evals — memory, API, tool-use, regression, safety, structured output, or role evals for an agents/<role>.md body
/cks:evaluateBuild the Process Evaluator feature — complete process card generation from text input
/cks:evolveAHE Evolution Agent — reads telemetry, governance, and harness-eval signals to propose golden cases for hook validation
/cks:expertInvoke an expert persona for architecture, product, debugging, or specialist deep-dives
/cks:factoryAFK software factory — pull labeled GitHub Issues and run the full lifecycle pipeline for each one autonomously
/cks:finopsMoney on track — cost audit, margin per client, invoice draft (gated send), budget burn, SR&ED evidence. Dispatches the finops role.
/cks:fixBuild error resolver — diagnose and fix build/compile/runtime errors fast
/cks:freezeRestrict file edits to a specific directory for this session
/cks:gateReview candidate skills in quarantine — human-gated: checks format/conflict/scope, then approve (promote to validated/) or reject (archive)
/cks:go[legacy] Quick action command — commit, PR, dev, build, start. PRD-aware: guides you through the lifecycle.
/cks:grillGrill-me plan interrogator — relentless one-question-at-a-time plan stress-test with recommended answers
/cks:handoffSave session context to .prd/handoffs/HANDOFF-{date}-{time}.md (unique per session) so parallel sessions don't clobber each other
/cks:harness-evalRun hook fixture evals against CKS harness handlers — validate hook behavior for G2 AHE Evolution Agent
/cks:headroomHeadroom — input token compression via MCP. Compress large tool payloads before they flood the context window.
/cks:heartbeatManage CKS v6 heartbeat agents — register a new agent or view all agents' heartbeat status
/cks:helpShow all available CKS commands and usage guide
/cks:hermesHermes Mode readiness — init, status, and smoke checks for always-on channel brain
/cks:honchoSelf-hosted Honcho memory layer — add theory-of-mind user representations on top of CKS file memory (augment, local-only)
/cks:hqWorkforce HQ repo — scaffold cross-venture state (North Star, finops, routines, memory, users) or show what exists at $CKS_HQ
/cks:ideate/cks:improve/cks:investigate/cks:kickstart/cks:launch-check/cks:learn/cks:logs/cks:loop-migrate/cks:loop/cks:luv-profile/cks:luv/cks:map-codebase/cks:market/cks:marketing-analytics/cks:marketing-build/cks:marketing-dev/cks:marketing/cks:me/cks:memory/cks:migrate/cks:model/cks:monetize-cost-analysis/cks:monetize-discover/cks:monetize-evaluate/cks:monetize-report/cks:monetize-research/cks:monetize-roadmap/cks:monetize/cks:new/cks:next/cks:observe/cks:optimize/cks:payments/cks:peers/cks:persona/cks:personas/cks:pivot/cks:preflight/cks:print-cli/cks:progress/cks:project-init/cks:refactor/cks:remind/cks:remotion/cks:research/cks:resume/cks:retro/cks:review-rules/cks:review/cks:routine/cks:rpi/cks:sandbox/cks:save-context/cks:scale/cks:schedule/cks:security/cks:seo-audit/cks:setup-webhooks/cks:ship/cks:simplify/cks:slack/cks:sleep/cks:sprint-auto/cks:sprint-run/cks:sprint-start/cks:sprint/cks:standup/cks:status/cks:tdd/cks:telegram/cks:test/cks:triage/cks:uat/cks:unfreeze/cks:version/cks:virginize/cks:voice/cks:workREADMEarchitectTurns discovery into buildable design — PRD and execution plan, UX flows, API contracts, screens and component specs, ARCHITECTURE.md and ADRs, pre-flight dependency mapping (PREFLIGHT.md), Supabase/pgvector data design and ERDs, DESIGN.html, scaling and payment-integration advice, and agent-system
assistantExecutive assistant — triages the inbox, keeps the calendar honest, drafts replies and follow-ups in the owner's EN/FR voice, preps meetings, tracks reminders, runs the daily brief. Prefers Brain 1 tools when connected, falls back to Gmail and Google Calendar MCP. Never sends; every outbound comes b
builderBuilder — writes application code from a PLAN.md task group with a TDD loop, refactors with behavior preserved, generates and rollback-tests schema migrations, scaffolds no-code workflows and API CLIs, and writes SUMMARY.md before returning. Use for "sprint", "build", "implement", "execute", "code i
chief-of-staffChief of staff — triages inbound work, decides what deserves attention, dispatches specialist agents, and enforces the three-priority limit. Decides and delegates; never executes. Use at session start, when work is piling up, or when it is unclear what to do next.
debuggerDebugger — root-cause diagnosis and the minimal fix: classifies the failure, traces the causal chain to where bad state was introduced, applies a scoped edit, verifies, and closes the issue. Triage mode scans, files every finding to GitHub, and returns a queue; db-fix mode traces and repairs Supabas
finopsKeeps money on track — API/token/infra costs, margin per client and venture, budget burn vs ceiling, Stripe invoicing (gated), SR&ED evidence from git history, Stripe integration advice. Reports and drafts; never moves money without approval.
historianHistorian — the workforce's memory: persists REMEMBER blocks, maintains the wiki with OKF frontmatter, curates validated learnings by PR, runs retrospectives, drafts improvement proposals from dispatch traces, reviews sleep-cycle proposals, and writes the session journal and user profile. Writes onl
marketerOne marketing role — loads the persona the brief needs (copy, brand, paid, SEO/GEO/AEO, creative, analytics, outbound prospecting, claims compliance) and runs campaigns end to end. Writes only under .campaign/ and .marketing/; every send, load, post, or spend comes back as a draft for approval.
observerReads runtime signals and reports — Sentry errors, Vercel and Cloudflare logs, Supabase advisors, LangSmith traces, control-plane session metrics, coordination locks, and post-deploy canaries. Reports only; never fixes, never files issues.
operatorSetup and integrations — bootstrap and adopt scaffolding, Agentic OS, channels (Telegram, Slack), voice via Telnyx, schedules and heartbeats, sandbox policy, control-plane writes, migrations, plugin dependencies. Writes project config and scaffolds only; never application code, never a Routine witho
project-managerKeeps the board true — every piece of work is a GitHub Issue with an owner, an outcome and an honest state; sole writer of the work hierarchy, session handoffs and the DEVLOG; wires the Kanban board; turns follow-ups into issues. Never invents work and never decides priority — the chief of staff doe
researcherExternal research — social and market signal first via last30days, then multi-hop web research across Perplexity, Context7, Firecrawl and the web; competitor and tech evaluations, market and pricing research, codebase questions for planning, ecosystem bulletins, and CCCS threat intel. Writes reports
reviewerReviewer — static review with no write path: code review against the checklist, OWASP security audit with secrets and dependency scans, design-fluency lint, Supabase RLS audit, contract review, and Canadian compliance surface. Returns findings as a severity table with file:line and names every block
shipperShipper — takes verified work to users: build, commit, push, PR, CI watch, changelog and version bump, environment promotion dev → staging → RC → production with quality gates and health checks. Production deploys are drafted and returned as GATED, never executed alone. Use for "deploy", "release",
strategistDiscovery and strategy — client intake and scoping, the 11-element feature discovery, kickstart intake and gates, ideation, feature scope, concept feasibility scoring, monetization evaluation and roadmap, pivots, personas and profiles, plan interrogation, compliance surface and Canadian legal risk c
testerTester — runs the verification that proves work is done: test suites against acceptance criteria, browser UAT with human sign-off, LLM evals (smoke/standard/comprehensive/red-team), and hook harness evals. Writes VERIFICATION.md with the Evidence Bundle front-matter and CONFIDENCE.md, and files GitH
watchdogFinds friction nobody reported — rules nothing enforces, automation that stopped, assets never used, work that silently stalled, spend without output, and agency KPIs drifting; also runs the health check, rules audit, launch-readiness gate, loop health and cost estimates, and the skill-quarantine re
writerWriter — generates project documentation from the codebase (API, architecture, components, onboarding, coding-agent llms.txt) and drafts contracts (MSA, SOW, NDA) from templates. Every contract is a draft for the reviewer; every doc cites real paths. Use for "docs", "document this", "write the guide
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/session-start.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/peer-announce.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/tool-counter.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/destructive-op-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/user-memory-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/secrets-scan-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/version-bump-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/pre-commit-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/merge-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/integrity-check.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/worktree-isolation-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/dispatch-first-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/freeze-boundary-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/budget-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/jev-model-router.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/test-fail-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/large-read-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/post-tool-trace.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/governance-log.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/post-edit-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/design-md-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/test-fail-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/kickstart-phase-complete.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/ideation-complete.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/design-system-complete.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/subagent-stop-trace.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/peer-announce.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/slack-notify.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/context-guard.sh
bash ${CLAUDE_PLUGIN_ROOT}/hooks/handlers/pre-compact.sh
Matt Pocock's agent skills for real engineering (grilling, spec/ticket flows, TDD, code review, domain modelling and more).
39 skills
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
Automates browser interactions for web testing, form filling, screenshots, and data extraction
12 skills
Orchestrate multi-agent teams for parallel code review, hypothesis-driven debugging, and coordinated feature development using Claude Code's Agent Teams
6 skills · 7 commands · 4 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.