kitfunso

Suite of tools for creating elaborate, multi-component claude.ai HTML artifacts using modern frontend web technologies (React, Tailwind CSS, shadcn/ui). Use for complex artifacts requiring state management, routing, or shadcn/ui components - not for simple single-file HTML/JSX artifacts.

Patterns for autonomous agent loops — sequential pipelines, parallel generation, continuous PR loops, and de-sloppify passes. Use when setting up automated workflows, CI/CD agent loops, batch processing with Claude, or multi-agent orchestration. Covers loop architecture from simple sequential to RFC-driven DAG orchestration.

You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.

Fast headless browser for QA testing and site dogfooding. Navigate any URL, interact with elements, verify page state, diff before/after actions, take annotated screenshots, check responsive layouts, test forms and uploads, handle dialogs, and assert element states. ~100ms per command. Use when you need to test a feature, verify a deployment, dogfood a user flow, or file a bug with evidence. Use when asked to "open in browser", "test the site", "take a screenshot", or "dogfood this".

Bump iOS/Android build numbers, build the Android .aab bundle, commit and push. Use this skill whenever the user says 'build release', 'bump build', 'build aab', 'bump and push', 'bump ios build number', 'build the aab file', or any combination of bumping build numbers, building Android bundles, and pushing. Also trigger when user says 'release build' or 'ship the build'.

Safety guardrails for destructive commands. Warns before rm -rf, DROP TABLE, force-push, git reset --hard, kubectl delete, and similar destructive operations. User can override each warning. Use when touching prod, debugging live systems, or working in a shared environment. Use when asked to "be careful", "safety mode", "prod mode", or "careful mode".

OpenAI Codex CLI wrapper — three modes. Code review: independent diff review via codex review with pass/fail gate. Challenge: adversarial mode that tries to break your code. Consult: ask codex anything with session continuity for follow-ups. The "200 IQ autistic developer" second opinion. Use when asked to "codex review", "codex challenge", "ask codex", "second opinion", or "consult codex".

Pattern for caching expensive file operations using SHA-256 content hashes instead of file paths. Use when building data pipelines that process files (PDF parsing, OCR, image analysis, signal computation) and need caching. File rename = cache hit. Content change = auto-invalidation.

Project memory powered by Hippo. Use when starting a session, after learning something, after hitting an error, or when the user mentions memory/remember/recall. Auto-invoked at session start.

Make the agent use proportional effort and spend fewer tokens. Use whenever the user wants concise answers, quick mode, low-token mode, fast diagnosis, minimal prompting, compact planning, answer-first behavior, or cheaper agent behavior. Also use for low-stakes repo-local checks like "old version", "stale badge", "release tag", "current repo", or "ready to push", especially when the user says things like "just answer", "diagnose only", "one line", "short answer", "no tools", or "no browser". Use it whenever a task risks expanding beyond what was asked, when duplicate messages arrive, when browser/tool overuse is likely, or when the agent should solve the task via the cheapest adequate path first and escalate only if needed.

Scaffold a new project using a four-document engineering framework — PRD, Architecture, AI Rules, and Plan — that turns "vibe coding" into disciplined engineering. Use when the user starts a new project, app, or major feature, or asks to scaffold, plan, or set up a project with proper structure and guardrails against scope creep and spaghetti code.