One-line prompts.
Shipped products.
23 skill categories. 28 agents. 149 reference docs.
The autonomous product-building OS for every AI coding tool.
gh skill install heymegabyte/agent-skills
How It Works
From idea to deployed product in three steps.
Agent Skills turns a single natural language prompt into a deployed, tested product. The system handles architecture, code generation, testing across 6 breakpoints, deployment to Cloudflare Workers, and visual verification — all without human intervention.
Install
gh skill install heymegabyte/agent-skills
Drop into Cursor, Copilot, Claude Code, Codex — or any AI coding tool.
Prompt
"Build a SaaS at projectsites.dev"
One line. The system infers product type, stack, and architecture automatically.
Ship
✓ Deployed to Cloudflare Workers
Architect → build → test → deploy → verify. Fully autonomous. Zero hand-holding.
What It Does
Every main capability, grouped. One skill set that architects, builds, tests, deploys, and verifies complete products.
Agent Skills gives an AI coding agent the full capability set of a senior product engineer: autonomous end-to-end delivery (one prompt → a deployed, tested product), a Cloudflare-native full stack, TDD-first verification across six breakpoints, cinematic website generation that outscores competitors, portability across 37 AI-tool platform variants, preference memory with an autonomous idea engine, observability and growth instrumentation, and built-in trust & compliance (legal, security, accessibility, deliverability) — all from a single agent-neutral skill set.
Autonomous Engineering
- One-line prompt → deployed, tested product
- Architect → build → test → deploy → verify, zero hand-holding
- 4-tier approval: autonomous · review · approval · blocked
- 28 specialized agents orchestrated in parallel
- Self-improving — folds lessons back into the skills
- Self-healing — Sentry alert → agent traces the error → auto-merging fix PR
Cloudflare-Native Full Stack
- Workers + Hono edge APIs, 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 × 6 breakpoints, zero console errors
- AI visual QA · axe WCAG 2.2 AA · Lighthouse ≥ 95
- Deploy + production E2E on every single change
- Feature flags · contract-first AI · drift detection
Cinematic Websites
- One line → gorgeous, cinematic, functional site
- Competitor research — 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
- Claude Code · Codex · Cursor · Copilot · OpenCode · Windsurf · Gemini
- Per-platform adapters preserve precision
- Agent-neutral core — no privileged tool
- MCP-native · browser-agentic · edge-first
Intelligence & Growth
- Remembers & evolves your preferences — confidence-scored Voice-of-Customer model
- Autonomous idea engine — evidence-backed improvement proposals, like an internal co-founder
- Observability from day one — PostHog analytics + Sentry error tracking + feature flags
- Conversion optimization + A/B testing + growth instrumentation
- Cost-aware routing — spends promo credits before metered API
Trust & Compliance
- Legal pages + GDPR / CCPA / ADA Title II (WCAG 2.1 AA) compliance
- Security hardening — CSP Level 3 + Trusted Types + OWASP Top 10
- Email deliverability — SPF / DKIM / DMARC, bulk-sender compliant
- Every feature dark-launched behind a flag (0% rollout → promote)
- Contract-first AI — Zod-validated typed outputs at every boundary
See the Skill Router
One prompt loads only the skills it needs — never the whole mesh. Pick an example, or describe your own build.
6 of 23 categories loaded · 74% of the mesh skipped
23 Skill Categories
Each category contains specialized submodules and reference docs. The router loads the smallest useful subset for every prompt.
The 23 skill categories cover the full product lifecycle: goal definition, architecture, build, test, deploy, brand, design, motion, media, and growth — plus delivery-requirement categories for app foundation, visual experience, content and SEO, functional completeness, quality and performance, platform delivery, trust and compliance, and a consolidated Cloudflare integrations reference. Each category contains specialized submodules with 149 reference docs total, loaded on-demand to minimize context window usage.
23 of 23
Operating System
Supreme policy layer. Autonomy, speed standards, cross-skill coordination, conflict resolution.
6 reference docsGoal & Brief
Project thesis before code. Product type inference, user identification, business model.
Core skillPreference & Memory
User preferences with confidence levels. Voice of the Customer model, auto memory system.
3 reference docsArchitecture & Stack
Cloudflare-first platform selection. Workers, D1, R2, KV, Durable Objects, Queues, Vectorize, AI Search.
25 reference docsBuild & Slice Loop
Vertical slices, homepage-first. Anti-placeholder rules — real content, real interactions.
26 reference docsQuality & Verification
5-level verification pyramid. Playwright E2E, AI visual QA, agentic security testing.
27 reference docsDeploy & Runtime
Deploy after every change. Typecheck → deploy → purge → E2E → visual verify → fix-forward.
11 reference docsBrand & Content
Copy system, headline rules, trust surfaces, SEO, structured data, anti-AI-slop filters.
8 reference docsExperience & Design
Anti-AI-slop design system. Dark-first, bold typography, cascade layers, container queries.
3 reference docsMotion & Interaction
Meaning-first animation. Scroll-driven, View Transitions API, prefers-reduced-motion mandatory.
1 reference docMedia Orchestration
Image generation, logo/icon sets, video, social previews, OG images, compression pipelines.
12 reference docsObservability & Growth
PostHog analytics, GA4 via GTM, Sentry, Stripe billing, feature flags, conversion optimization.
11 reference docsIdea Engine
Autonomous internal co-founder. Evidence-backed improvements, self-critique filter, auto-implement.
Core skillSite Generation
Research-saturated, maximalist site builds. Competitor-beating, AI-native, deploy-verified.
16 reference docsCinematic Prime Directive
One-line prompt to 100 build-breaking rules across 10 quality categories before DONE.
Core skillApp Foundation
Non-inferable business requirements plus the exact stack and brand choices every build must satisfy.
Core skillVisual Experience
The cinematic UI bar: anti-slop premium, brand tokens, motion, logo and image quality, WCAG 2.2 visuals.
Core skillContent & SEO
Real-brand copy, per-page SEO and JSON-LD, pSEO, GEO/AI search, citations, i18n, trust surfaces.
Core skillFunctional
What every app must DO: complete features, working forms, notifications, flags, embarrassingly easy UX.
Core skillQuality & Performance
Core Web Vitals, WCAG 2.2 AA, TDD-first real-browser E2E, and the build-fail quality gates.
Core skillPlatform & Delivery
Cloudflare-first hosting, prod-only discipline, deploy → prod-verify loop, atomic deploys and rollback.
Core skillTrust & Compliance
Legal pages, PII and deletion rights, CSP, Trusted Types, Zod boundaries, AI-agent security, RFC 7807 errors.
Core skillCF Integrations Reference
Consolidated Cloudflare and integrations reference: platform, data, AI edge, media, payments, site generation.
Core skillNo skills match — .
28 Specialized Agents
Each agent has a defined model tier, permission mode, and skill set. The orchestrator spawns them in parallel.
Agent Skills uses 28 specialized agents — each assigned to an optimal model tier (deep reasoning for architecture/security/incident response, balanced for implementation/testing/content, fast for changelogs/formatting/renames), mapped to whatever models your AI tool provides. The meta-orchestrator spawns agents in parallel, coordinating architect, test-writer, and deploy-verifier simultaneously to ship faster.
28 of 28
Architect
ReasoningAnalyzes project structure, generates repo maps, designs task graphs, identifies architectural seams.
Code Simplifier
FastSimplifies and refines code for clarity, consistency, and maintainability while preserving all functionality.
Completeness Checker
ReasoningVerifies nothing was missed. Feature Completeness Engine, Zero Recommendations Gate, visual verification.
Deploy Verifier
BalancedPost-deploy smoke tests. Checks console errors, screenshots 6 breakpoints, runs axe-core, validates SEO.
Security Reviewer
ReasoningOWASP Top 10, secrets exposure, injection flaws, auth bypasses, CSP issues. Read-only — never modifies code.
Test Writer
BalancedTDD-first. Writes failing Playwright tests that emulate real users before implementation. Vitest for units.
SEO Auditor
BalancedTitle, meta, H1, JSON-LD, OG tags, internal links, sitemap, robots.txt. Playwright-powered live audits.
Visual QA
ReasoningScreenshots at all breakpoints. AI vision detects layout breaks, misalignment, text overflow, broken images.
Computer Use
ReasoningDesktop automation via Computer Use MCP. Finder, System Settings, native apps, cross-app workflows.
Dependency Auditor
FastScans for outdated deps, security advisories, license violations, unused imports. Prioritized upgrade reports.
Meta Orchestrator
ReasoningMaster agent. Knows every tool, MCP, and skill. Plans multi-system workflows, chains MCPs, spawns agents.
Migration Agent
BalancedDrizzle schema migrations. Generates from diffs, validates against D1, tests rollbacks, zero-downtime.
Content Writer
BalancedMarketing copy, blog posts, SEO content. Brand voice, Flesch≥60, active voice, zero AI slop.
Performance Profiler
ReasoningLighthouse audits, Core Web Vitals analysis. Targets LCP≤2.0s, CLS≤0.05, INP≤100ms with specific fixes.
Incident Responder
ReasoningSentry-triggered. Reads error events, traces to source, proposes fixes, creates branches, opens PRs.
Accessibility Auditor
Balancedaxe-core + Playwright. WCAG 2.2 AA audits, reports violations with fix suggestions, verifies remediation.
Cost Estimator
FastEstimates Cloudflare Workers costs. Reads wrangler.toml, counts D1 tables, calculates monthly cost.
Changelog Generator
FastAuto-generates from conventional commits. Parses git log, groups by type, writes user-outcome-focused entries.
Media Orchestrator
BalancedCreates and optimizes media — images, video clips, TTS audio, podcasts, 3D — coordinating generation pipelines.
Motion Choreographer
BalancedView Transitions API, scroll-driven animations, FLIP, staggered reveals. prefers-reduced-motion, always.
Changelog Drafter
FastReads git log since the last tag, drafts user-outcome entries grouped by conventional-commit type.
Dead Code Remover
FastFinds unreachable exports, unused imports, orphaned functions. Confirms zero references, then removes safely.
Formatter
FastRuns Prettier plus lint autofix on modified files. Consistent formatting with zero human intervention.
Model Router
FastRecommends the right model tier and effort level for each task before work is dispatched.
Renamer
FastSemantic rename across the codebase. Greps every reference, updates imports and usages, verifies the result.
Transcriber
FastTurns a plan step into code, one file at a time. Clean production implementation straight from the spec.
Browser Operator
BalancedDrives a real browser after every deploy — golden paths, visuals, console, and network verification.
Resource Broker
BalancedNormalized registry of owned account resources — credits, quotas, free tiers — with secret-redacted tracking.
No agents match — .
Built for the Modern Stack
Opinionated defaults, escape hatches when you need them.
37 Platform Variants
One skill system. Every AI coding tool. Auto-generated convention files for each platform's format.
Modern Formats
Legacy Single-File
Named Formats
Directory Formats
Get Started
Pick your tool. Drop in the skills. Start shipping.
GitHub Skills
gh skill install heymegabyte/agent-skillsClaude Code Plugin
claude plugin install heymegabyte/agent-skillsOpenAI Codex
git clone https://github.com/heymegabyte/agent-skills \
~/.codex/skills/agent-skillsManual (any tool)
git clone https://github.com/heymegabyte/agent-skills
# Copy the convention file for your toolBefore & After
What changes when you add 14 years of engineering judgment to your AI.
Without skills, AI coding tools produce generic boilerplate with no tests, no deploy pipeline, and placeholder content. With Agent Skills, the same prompt produces a Cloudflare-deployed product with Playwright E2E tests across 6 breakpoints, Lighthouse 95+ accessibility, real content, and zero manual fixups.
Without Skills
With Agent Skills
How Skills Flow
The router loads the smallest useful subset. No wasted context.
FAQ
Does this work with Cursor, Copilot, and Windsurf?
Yes. Agent Skills generates 37 platform-specific convention files. When you install it, your tool automatically picks up the right format — MDC rules for Cursor, copilot-instructions.md for Copilot, .windsurfrules for Windsurf, and 34 other variants.
Is this different from .cursorrules files?
Fundamentally. A .cursorrules file is a flat text dump. Agent Skills is a 23-category system with 149 reference docs, intelligent routing (load only what you need), and 28 specialized agents. It generates .cursorrules as one of 37 outputs.
What if I don't use the default stack?
The system has escape hatches for every choice. Don't use Angular? The frontend skill adapts. Don't use Cloudflare? The deploy skill handles other platforms. The architecture decisions are opinionated defaults, not locked dependencies.
How do the 28 agents work?
Each agent has a defined model tier — deep reasoning, balanced, or fast, mapped to whatever models your AI tool provides — plus a permission mode and skill set. The meta-orchestrator spawns them in parallel — architect designs the structure, test-writer creates failing tests, deploy-verifier checks production. They coordinate automatically.
Is this free?
Yes. Open source, free forever. The skills are published on GitHub and install with a single command. You bring your own AI tool subscription — Agent Skills just makes it dramatically more effective.
How do I create custom skills?
Follow the SKILL.md format with frontmatter (name, description, submodules). Place it in the numbered directory structure. The router will pick it up automatically. See the CONVENTIONS.md for the full spec.
How does the skill router minimize context usage?
The router analyzes your prompt and loads only the relevant skill categories and their submodules. A billing feature loads skills 05 (Architecture) and 13 (Observability) but skips 11 (Motion) and 12 (Media). This keeps context usage under 20% of the available window for most prompts.
Does Agent Skills work with OpenAI Codex CLI?
Yes. Agent Skills generates a CODEX.md file and supports the ~/.codex/skills/ directory format. Install with: git clone https://github.com/heymegabyte/agent-skills ~/.codex/skills/agent-skills. The same skill content works across every supported tool.
What verification does the system run after deployment?
Five layers: TypeScript compilation check, Playwright E2E tests across 6 breakpoints (375px to 1920px), axe-core accessibility audit (zero violations required), AI visual QA via vision-model screenshot analysis, and Lighthouse scoring (accessibility ≥95, performance ≥75). Deploy failures trigger automatic fix-forward cycles.
Can I use this for existing projects, not just new ones?
Absolutely. Drop the skills into any repo and the system adapts to your existing stack, patterns, and conventions. It reads your codebase, respects your architecture, and applies its quality standards to new features while preserving what's already there.
How is this different from a SaaS template or boilerplate?
Templates give you starting code. Agent Skills gives you an engineering brain. It makes real-time decisions about architecture, generates tests before code (TDD), deploys and verifies automatically, writes production copy (not "Lorem ipsum"), and iterates until zero recommendations remain. It's a system, not a scaffold.
What security checks are built in?
The Security Reviewer agent scans for OWASP Top 10 vulnerabilities, hardcoded secrets, injection flaws, auth bypasses, and CSP issues. Every form gets Turnstile protection. Every database query uses parameterized Drizzle ORM. detect-secrets scan runs on every commit via pre-commit hooks.