Skills Overview

MeowKit ships 43 skills using the meow: namespace prefix.

Architecture

Each skill folder follows a two-layer design: a compact decision router (SKILL.md, ~100 lines) that handles routing logic inline, and a references/ directory for detailed procedures loaded on demand. This keeps context overhead low — agents load only what each phase requires.

meow:skill-name/
├── SKILL.md          # Decision router (~100 lines)
├── references/       # Detailed procedures (loaded on-demand)
│   ├── workflow-*.md
│   └── patterns.md
├── scripts/          # Validation scripts
├── bin/              # Shell hooks
└── assets/           # Templates, configs

On-demand loading

Skills only load their references/ files when the relevant phase is active. This prevents flooding the context window with procedures that don't apply to the current step.

Skill Types

Skills fall into distinct types based on purpose and phase alignment.

Type	Purpose	Examples
Pipeline	End-to-end workflows orchestrating multiple phases	cook, fix, ship, workflow-orchestrator
Quality Gate	Enforce standards at specific phase boundaries	review, cso, vulnerability-scanner, lint-and-validate
Planning	Plan creation, review, and ideation	plan-creator, plan-ceo-review, plan-eng-review, office-hours
Exploration	Codebase search, debugging, doc retrieval	scout, investigate, docs-finder
Domain	Stack-specific patterns and conventions	typescript, vue, frontend-design
Safety	Prevent destructive actions, scope edits	careful, freeze, skill-template-secure
Infrastructure	Agent routing, loading, skill creation	agent-detector, lazy-agent-loader, skill-creator
Reference Toolkit	Guides loaded by agents during specific phases	development, testing, planning, shipping, documentation, memory
Utility	Cross-cutting tools (browser, media, docs gen)	browse, agent-browser, multimodal, llms, qa, qa-manual, playwright-cli

Plan-First Gate

Most pipeline and quality skills check for an approved plan before proceeding. The table below documents each skill's gate behavior and the conditions under which planning is skipped.

Skill	Gate behavior	Skip condition
meow:cook	Create plan if missing	Plan path arg, --fast mode
meow:fix	Plan if > 2 files	--quick mode
meow:ship	Require approved plan	Hotfix with human approval
meow:cso	Scope audit via plan	--daily mode
meow:qa	Create QA scope doc	Quick tier
meow:review	Read plan for context	PR diff reviews
meow:workflow-orchestrator	Route to plan-creator	Fasttrack mode
meow:investigate	Produces input FOR plans	Always skips
meow:office-hours	Pre-planning skill	Always skips
meow:retro	Data-driven, no plan	Always skips
meow:document-release	Scope from diff	Post-ship sync

Why some skills always skip planning

Skills that skip planning have documented reasons — they either produce planning input (investigate, office-hours) or are data-driven and operate after the ship phase (retro, document-release).

Pipeline Skills

Full development pipelines that orchestrate multiple phases.

Skill	What it does
meow:cook	End-to-end feature pipeline: plan → test → build → review → ship
meow:fix	Structured bug investigation with auto-complexity detection
meow:ship	Ship pipeline: merge, test, review, version, commit, PR
meow:workflow-orchestrator	5-phase TDD workflow with token budgets and fast-track
meow:session-continuation	Save/resume workflow state across sessions

Quality & Review

Code quality enforcement, review, and security scanning.

Skill	What it does
meow:review	Multi-pass code review with adversarial analysis
meow:cso	Chief Security Officer — infrastructure-first audit
meow:vulnerability-scanner	OWASP 2025 code-level scanning
meow:clean-code	Pragmatic coding standards (SRP, DRY, KISS)
meow:lint-and-validate	Auto-run linters + type checks after changes

Planning & Design

Plan creation, review, and ideation.

Skill	What it does
meow:plan-creator	Auto-select plan template by task scope
meow:plan-ceo-review	CEO-mode plan review with scope expansion
meow:plan-eng-review	Engineering plan review — architecture, edges, tests
meow:office-hours	YC-style brainstorming (startup + builder modes)
meow:brainstorming	Structured ideation with scoring + plan-creator handoff

Exploration & Research

Codebase exploration, debugging, and documentation retrieval.

Skill	What it does
meow:scout	Parallel codebase exploration via Explore subagents
meow:investigate	Systematic 5-phase root cause debugging
meow:docs-finder	Library/project docs via Context7 + Context Hub
meow:sequential-thinking	Hypothesis-driven reasoning for root cause analysis

Browser & QA Testing

Browser automation, QA testing, and E2E code generation.

Skill	What it does
meow:qa-manual	Spec-driven manual QA + Playwright E2E code gen
meow:qa	Systematic QA with bug fixing and health scores
meow:browse	Fast headless browser (~100ms/command)
meow:agent-browser	Chrome/CDP with session persistence and auth
meow:playwright-cli	Playwright MCP with code generation

Frontend

TypeScript, Vue, React, and UI/UX design patterns.

Skill	What it does
meow:typescript	Strict TS: null handling, type guards, utility types
meow:vue	Vue 3 Composition API, Pinia, reactivity
meow:angular	Angular v20+ patterns — 10 topics consolidated (components, signals, DI, forms, routing, HTTP, SSR, testing, tooling)
meow:react-patterns	React/Next.js performance — 45+ rules from Vercel Engineering
meow:frontend-design	UI/UX with anti-AI-slop enforcement
meow:ui-design-system	Design intelligence: styles, WCAG 2.1 AA, palettes, quality checklists

Analysis & Media

Multimodal analysis and documentation generation.

Skill	What it does
meow:multimodal	Image/video/audio/PDF via Gemini API
meow:llms	Generate llms.txt per llmstxt.org spec

Safety & Scoping

Destructive command prevention and edit restrictions.

Skill	What it does
meow:careful	Warn before rm -rf, DROP TABLE, force-push
meow:freeze	Restrict edits to one directory
meow:skill-template-secure	Secure template for skills handling untrusted input

Infrastructure

Agent detection, loading, skill creation, and project organization.

Skill	What it does
meow:agent-detector	Auto-detect agent + complexity + model per message
meow:lazy-agent-loader	On-demand agent loading for token savings
meow:skill-creator	Scaffold + validate + register new skills
meow:project-organization	File naming, directory structure standards
meow:bootstrap	Application scaffold for any stack with progressive generation

Documentation & Release

Post-ship docs, retrospectives, and release management.

Skill	What it does
meow:docs-init	Generate initial docs from codebase analysis
meow:document-release	Post-ship doc sync — cross-reference diff against docs
meow:retro	Sprint retrospective with commit analysis + trends

Reference Toolkits

Collections of reference guides loaded by agents during specific phases.

Skill	What it does
meow:development	Code patterns, TDD, skill loading (Phase 3)
meow:documentation	Living docs, API sync, changelog (Phase 6)
meow:memory	Session capture, patterns, cost tracking (Phase 0, 6)
meow:shipping	Ship pipeline, canary deploy, rollback (Phase 5)
meow:testing	Red-green-refactor, validation, visual QA (Phase 2-3)

See Also

• Agent-Skill Architecture — how agents and skills interact across workflow phases
• Workflow Phases — the 6-phase pipeline from Orient to Reflect
• Agents Overview — the full agent roster and ownership model