# Agent Skills

> Agent Skills is an autonomous product-building operating system for AI coding agents: 23 skill categories, 28 specialized agents, and 149 reference docs that turn one-line natural-language prompts into deployed, tested, production-ready products on Cloudflare. It is agent-neutral — the same skill content compiles to 37 platform variants (Claude Code, Codex, Cursor, Copilot, OpenCode, Windsurf, Gemini, and more), with no privileged tool.

Built by a Principal Software Engineer (14 years, front-end to OS). The skill router loads the smallest useful subset of skills for each prompt to minimize context-window usage.

## Core

- [GitHub repository](https://github.com/heymegabyte/agent-skills): the skills monorepo (skills, agents, rules, templates) — install with `gh skill install heymegabyte/agent-skills`
- [README](https://raw.githubusercontent.com/heymegabyte/agent-skills/master/README.md): overview, install paths, and the full capability map
- [llms-full.txt](https://agent.skillsl.ink/llms-full.txt): the complete reference — every skill, agent, platform variant, and capability in one fetch
- [Changelog](https://raw.githubusercontent.com/heymegabyte/agent-skills/master/CHANGELOG.md): what ships and when

## Main capabilities

- [What It Does](https://agent.skillsl.ink/#capabilities): the full feature list, grouped
- Autonomous engineering: one-line prompt to a deployed, tested product; architect to build to test to deploy to verify with zero hand-holding; 4-tier approval (autonomous / review / approval / blocked); 28 specialized agents orchestrated in parallel; self-improving
- Cloudflare-native full stack: Workers + Hono edge APIs with Zod at every boundary; D1 / Neon + Drizzle ORM, R2, KV, Durable Objects, Queues; Workflows, AI Search + Vectorize RAG, Workers AI + AI Gateway; Clerk auth, Stripe + Link payments, Amazon SES email; React 19 + Vite + shadcn/ui or Angular 22 + Spartan UI
- Quality & verification: TDD-first (a failing Playwright E2E before any code); real-browser journeys across 6 breakpoints with zero console errors; AI visual QA, axe WCAG 2.2 AA, Lighthouse >= 95; deploy + production E2E on every change; feature flags, contract-first AI, drift detection
- Cinematic websites: one line to a gorgeous, cinematic, functional site; competitor research to outscore every peer by >= 15%; maximalist enrichment + AI-native spiral; SEO / GEO, JSON-LD, PWA, Core Web Vitals budgets; generative media (Ideogram 4.0, GPT Image 2, Veo 3.1)
- Runs everywhere: 37 platform variants from one skill set; per-platform adapters preserve precision; agent-neutral core with no privileged tool; MCP-native, browser-agentic, edge-first
- Intelligence & growth: remembers + evolves your preferences (confidence-scored Voice-of-Customer model); autonomous idea engine proposing evidence-backed improvements; observability from day one (PostHog analytics + feature flags + experiments); conversion optimization + A/B testing; cost-aware routing
- Trust & compliance: legal pages + GDPR / CCPA / ADA Title II (WCAG 2.1 AA); security hardening (CSP Level 3 + Trusted Types + OWASP Top 10); email deliverability (SPF / DKIM / DMARC); every feature dark-launched behind a flag; contract-first AI (Zod-validated typed outputs)

## Skills (23 categories)

- [Browse all skill categories](https://agent.skillsl.ink/#skills): each category holds specialized submodules + reference docs
- Lifecycle: Operating System, Goal & Brief, Preference & Memory, Architecture & Stack, Build & Slice Loop, Quality & Verification, Deploy & Runtime
- Craft: Brand & Content, Experience & Design, Motion & Interaction, Media Orchestration, Observability & Growth, Idea Engine, Site Generation, Cinematic Website Prime Directive
- Delivery requirements: App Foundation, Visual Experience, Content & SEO, Functional, Quality & Performance, Platform & Delivery, Trust & Compliance, CF Integrations Reference

## Agents (28 specialized)

- [Browse all agents](https://agent.skillsl.ink/#agents): each declares a model tier (deep reasoning / balanced / fast) that maps to whatever models your tool provides
- Reasoning tier: architect, completeness-checker, computer-use-operator, incident-responder, meta-orchestrator, performance-profiler, security-reviewer, visual-qa
- Balanced tier: accessibility-auditor, browser-operator, content-writer, deploy-verifier, media-orchestrator, migration-agent, resource-broker, seo-auditor, test-writer, motion-choreographer
- Fast tier: changelog-drafter, changelog-generator, code-simplifier, dead-code-remover, dependency-auditor, formatter, model-router, renamer, transcriber, cost-estimator

## Install

- [All install methods](https://agent.skillsl.ink/#install)
- GitHub Skills: `gh skill install heymegabyte/agent-skills`
- Claude Code plugin: `claude plugin install heymegabyte/agent-skills`
- Any tool (manual): `git clone https://github.com/heymegabyte/agent-skills` then use the convention file for your platform (AGENTS.md, CODEX.md, GEMINI.md, .cursor rules, .windsurfrules, and 33 other variants)

## Default stack

- CF Workers + Hono (edge APIs) · React 19 + Vite + shadcn/ui or Angular 22 + Spartan UI (frontend) · D1 / Neon + Drizzle (data, Zod at boundaries) · Clerk (auth) · Stripe + Link default / Square on request (payments) · Inngest / CF Workflows (jobs) · Amazon SES (email) · Bun (runtime) · Playwright + Vitest (testing) · PostHog + Sentry + Workers Tracing (observability)

## Optional

- [Supported platforms](https://agent.skillsl.ink/#platforms): the full list of 37 variant surfaces
- [FAQ](https://agent.skillsl.ink/#faq)
- [Author](https://megabyte.space): Brian Zalewski / HeyMegabyte