AI agent skills: architect (PRD planner) + ralph (autonomous dev subagent)
Agent
Staff-level system architect. Creates parallelizable PRDs with junior-proof technical specs. Use when planning features, designing implementations, or when the user says 'plan', 'architect', 'design', or 'PRD'. Receives research reports and synthesizes them into a structured PRD with Priority groups for parallel ralph…
Agent
Overnight skill optimizer. Runs N iterations of the Karpathy autoresearch loop: generate variant → eval against binary test cases → commit if better → discard if not. Use to improve ralph, tester, architect, or any skill. Makes 12-15 experiments/hour. Run with --allow-all --max-autopilot-continues 100 for overnight…
Agent
Board member: Business Analyst. Evaluates business viability, revenue model, unit economics, market size, and competitive moat. Used by the visionary skill during the Board of Directors phase.
Agent
Build cinematic websites with AI-generated assets (Marcus/KIE API), scroll-driven canvas animations, and premium interactive modules. Transforms any brand into a scroll-driven cinematic experience using React/Next.js. 4 phases: brand analysis → asset generation with cost check → React build → polish+deploy. Triggers…
Agent
Board member: Compliance Advisor. Evaluates legal, regulatory, and compliance requirements — payments, data privacy, accessibility, licensing. Used by the visionary skill during the Board of Directors phase.
Agent
Marketing copywriter for landing pages, product copy, email campaigns, ads, and brand messaging. Use for writing or rewriting any customer-facing text. Triggered by: 'write copy', 'landing page copy', 'email copy', 'ad copy', 'rewrite this', 'marketing text', 'headline'.
Agent
Data engineer for building pipelines, ETL processes, analytics infrastructure, and data modeling. Use for database optimization, data transformations, reporting queries, and data integrations. Triggered by: 'data pipeline', 'ETL', 'analytics', 'data model', 'query optimization', 'reporting'.
Agent
Board member: Devil's Advocate. Stress-tests the vision — finds failure modes, over-engineering risks, missing assumptions, and fatal flaws. Used by the visionary skill during the Board of Directors phase.
Agent
DevOps engineer for CI/CD pipelines, Docker, GitHub Actions, infra automation, environment setup, and deployment configuration. Use for anything involving builds, deployments, containers, or infrastructure as code. Triggered by: 'CI/CD', 'GitHub Actions', 'Docker', 'deploy', 'pipeline', 'infra'.
Agent
Documentation and commit specialist. Runs after ralph subagents complete a Priority group. Updates PRD checkboxes, appends to progress.md, and makes one atomic git commit per completed user story. Also writes the implementation summary when the full PRD is done. Use after ralph finishes — never during active…
Agent
AI/ML engineer for integrating LLMs, building RAG systems, embeddings, vector stores, and AI-powered features. Use for anything involving AI SDK, OpenAI, Anthropic, Vercel AI SDK, LangChain, or ML pipelines. Triggered by: 'AI feature', 'RAG', 'embeddings', 'LLM integration', 'AI pipeline', 'vector search'.
Agent
Expert backend developer for APIs, databases, authentication, and server logic. Use for implementing endpoints, database schemas, business logic, and integrations. Autonomously implements tasks and signals ENGBACKENDDONE on completion. Triggered by: 'implement API', 'build endpoint', 'database schema', 'backend…
Agent
Expert frontend developer for React, React Native, TypeScript, and Expo. Use for implementing UI components, screens, navigation, state management, and frontend logic. Autonomously implements tasks and signals ENGFRONTENDDONE on completion. Triggered by: 'implement frontend', 'build screen', 'create component', 'fix…
Agent
Senior software engineer for complex implementation, refactoring, architecture decisions, and cross-cutting concerns. Use when tasks require deep judgment, cross-file understanding, or technical leadership. Triggered by: 'refactor', 'architect this', 'complex feature', 'technical decision', 'senior review'.
Agent
Adversarial QA evaluator. Validates that a sprint is ACTUALLY done before committing. Default position is REJECTION — approves only with screenshot proof. Part of the Generator→Evaluator loop with ralph. Uses playwright-cli to navigate the live app and find what ralph missed. Triggered by: 'evaluate sprint', 'validate…
Agent
Growth hacker for designing and implementing growth experiments, acquisition loops, referral programs, and viral mechanics. Use for growth strategy, experiment design, and implementing growth features. Triggered by: 'growth experiment', 'acquisition', 'viral loop', 'referral', 'growth feature', 'activation'.
Agent
Board member: Growth Marketer. Evaluates GTM strategy, positioning, ICP, channels, and competitive differentiation. Used by the visionary skill during the Board of Directors phase.
Agent
React Native and Expo specialist for building mobile screens, native features, device APIs, and app store preparation. Use for mobile-specific implementation tasks. Triggered by: 'mobile screen', 'React Native', 'Expo', 'native feature', 'iOS', 'Android', 'mobile'.
Agent
Autonomous mobile dev subagent for Expo/React Native. Implements a single user story from a PRD — screens, native components, data fetching, navigation. Run in parallel with other ralph-mobile instances. Uses building-native-ui, vercel-react-native-skills, native-data-fetching, and expo-dev-client skills. Returns…
Agent
Autonomous developer subagent. Implements exactly ONE user story from a PRD. Use for parallel, independent coding tasks. Each ralph instance owns specific files and does not touch anything else. Returns a RALPHDONE signal when complete. Does NOT commit or modify the PRD.
Agent
Rapid prototyper that ships a working version fast for validation. Prioritizes speed and learning over perfection. Use when you need a quick proof-of-concept, MVP feature, or validation prototype. Triggered by: 'prototype', 'MVP', 'quick version', 'validate fast', 'proof of concept', 'ship fast'.
Agent
Investigates a feature request from multiple angles: technical feasibility, UX/product considerations, risks, edge cases, and existing codebase patterns. Use at the start of a feature to enrich the vision before the architect creates the PRD. Runs fast in parallel with other researcher instances.
Agent
Security engineer for auditing code, identifying vulnerabilities, threat modeling, and hardening applications. Use for security reviews, penetration testing analysis, dependency audits, and implementing security controls. Triggered by: 'security audit', 'vulnerability', 'pentest', 'harden', 'OWASP', 'auth security'.
Agent
SEO specialist for keyword research, on-page optimization, technical SEO audits, and content strategy. Use for improving search visibility, analyzing rankings, and implementing SEO recommendations. Triggered by: 'SEO', 'keyword research', 'rank higher', 'meta tags', 'search visibility', 'SEO audit'.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: