Frontend design skills. Build: emil-design-eng (start here for components and screens), landing-page-design (single-CTA conversion landing page with a visual system), build-awwwards-quality-sites (immersive scroll-storytelling marketing sites), apple-design (Apple HIG principles). Motion: animate (add or refine animation on an already built site), animation-vocabulary (turn vague motion words into exact animation terms), motion (Framer Motion implementation + discipline), motion-system. Review and fix: screen-critique with the /critique-screen command (judge a rendered screen against seven critique lenses), interface-review (review UI changes in a PR, branch, or working tree with regression detection), better-interface (the holistic consolidation and verdict layer over the better-* family; uninstalled domains report as Not reviewed), better-layout (spacing, alignment, and grouping to the pixel), perception-laws (Gestalt laws, Von Restorff, visual hierarchy), accessibility-audit (WCAG 2.2 POUR). Systems: design tokens, typography scale, shadcn/ui composition, stitch-skill (create/update DESIGN.MD), taste-skill (targeted visual polish pass). Bridges (not vendored): impeccable-bridge for the full end-to-end UI/UX pass via /pro-dev install, taste-skills-bridge for the Leonxlnx anti-slop, visual-style, image-gen, and brand-kit library via /pro-dev install, plus web-design-engineer-bridge (reference-grounded HTML artifacts, style recipes, design scoring), tastemaker-bridge (extract tokens from a reference image's real pixels and lock the style across a project), and mengto-skills-bridge (scroll storytelling, WebGL and shader effects, video-to-superprompt) - all three installed with /pro-dev install. Install figma from claude-plugins-official via install-companions.sh. Bridges: react-patterns-bridge (Vercel composition and view-transition patterns). Lint: web-design-guidelines (file:line check of UI code against Vercel Web Interface Guidelines at a pinned commit). /pro-des
26 skills · 2 commands
Structured Prompt-Driven Development workflow: story decomposition, analysis, REASONS Canvas, generation, prompt update, sync, API test, code review, and reverse engineering. Includes openspdd-ts CLI (bin/openspdd.mjs) for installing templates into any AI editor. Depends on pro-execution for disciplined implementation workflows. Adapted from gszhangwei/open-spdd.
9 skills · 10 commands
Quality gate skills: requesting + receiving code review, verification-before-completion, user-validation. A Stop hook suggests /validate after repo changes; /validate writes a validation handoff (what changed, what was verified, what needs your eyes) to .pro-dev/validation/latest.md. Ships three documentation subagents (technical-writer, documentation-engineer, api-documenter) with /document and /api-docs, adapted from VoltAgent/awesome-codex-subagents. Nudges api-documenter when an OpenAPI/GraphQL/proto contract changes, and blocks plan saves outside the project directory. A floor guard warns on git push and gh pr create when a diff lowers the bar; /constraints writes a repo's CONSTRAINTS.md. A code-reviewer agent (Opus, read-only) reviews with fresh context from a four-field handoff.
5 skills · 6 commands · 6 agents · hooks
The AI-native SDLC as one stack-agnostic plugin. Stage commands: /intent (interview an idea or bug into intent/<id>-<slug>/intent.md, or intent/<date>-<slug>/ in a repo with no shortcode), /spec, /plan, /execute (build an approved plan in its own worktree and open the PR, never merge), /adr (record a decision with no change behind it), /review (REVIEW.md's bug, security and compliance passes), /pr-babysit (loop a PR to green, never merge); repo setup for the loop moved to /pro-dev install in pro-core. The templates the commands read ship as files under templates/ (intent, spec, plan, adr, REVIEW) so other tools can read them. Subagents: researcher (haiku), verifier (sonnet), simplifier (sonnet) and reviewer (opus). Four hooks ship in warn mode behind the mode files: protect-tests (no test edits in fix mode), plan-sync (a Stop hook that flags code changed with no plan.md update), production-gate (flags push to main, deploys, publishes and merges) and change-id (flags a commit on a change-id branch whose subject does not start with the id). /execute pauses before steps that touch auth, migrations, payments, deploys or secrets unless the plan's Risks names them; every plan cites references/definition-of-done.md. A doubt-driven-development skill sends a hard-to-undo build step to a fresh-context reviewer before the next step, with an optional Codex second opinion.
1 skill · 9 commands · 4 agents · hooks
Default Define + Plan skills: interview-me and idea-refine (Define-phase intent extraction and idea refinement, forked from agent-skills), brainstorming, written implementation plans, adhd — parallel divergent ideation under cognitive frames via /adhd (forked from UditAkhourii/adhd), and lavish — a bridge plus /lavish command for the kunchenguid/lavish-axi CLI that renders agent output (plans, tables, diagrams, diffs, reports) as reviewable HTML artifacts the user annotates in the browser (installed on demand via /pro-dev install). Depends on pro-execution for worktree and subagent execution handoff. /pro-pdd:install sets up its engines when pro-core is installed.
6 skills · 3 commands
Universal base skills and guardrails: find-skills, using-pro-dev (the cross-lifecycle router that maps a task to the right phase and skill, including bridge skills like impeccable-bridge, qa-suite, and lavish), and dream (memory consolidation over Claude Code auto-memory: scans session transcripts for corrections, decisions, preferences, and patterns, merges the findings into the memory files, and rebuilds a lean MEMORY.md index under 200 lines). /gh universal GitHub workflow command (briefing, new, commit, push, pr, ship, sync, resolve). /dream runs a consolidation on demand, with dry-run and status modes. /kaizen raises the bar one notch per run from a planning repo: it judges each tracked project against its own profile and current practice, red-teams it, applies one reversible improvement, briefs the rest and writes a dated ledger and HTML report; /kaizen init scaffolds the planning repo. PreToolUse hook blocks dangerous Bash (force-push to main, rm -rf /, .env redirects, sudo, chmod 777, curl|bash) and reads each git push for the forms that block misses (a delete of a protected branch, refs/heads/main, a bare push on main, git -C, --all, the default branch or a declared one), which it flags under its own protected-push mode: warn by default, deny only when mode.protected-push says enforce. A second PreToolUse hook on Read, Grep and Bash flags a read of a dotenv file (not .env.example), a private SSH key, a .pem file, .npmrc, a credentials file or a service-account JSON under its own secret-read mode: warn by default, deny only when mode.secret-read says enforce. scripts/env-check.py says whether a key is set (present, empty or missing) and never prints its value. A second PreToolUse Bash hook flags rm, rmdir, unlink, find -delete, find -exec rm and git clean whose target resolves outside the project root, with the temp folders and the pro-dev state folder exempt, under its own delete-outside mode: warn by default, deny only when mode.delete-outside says enforce
3 skills · 4 commands · hooks
/pro-ship:pr-check runs a repo's own checks, the floor guard, a mutation check and fresh-context reviewers, then writes a stamp; a gate hook on gh pr create warns when the stamp is missing or stale. A risk classifier sorts each change into low, medium or high and a merge gate on gh pr merge asks for approval at medium and high, pinned to the head commit. In pro-dev-starter. The documentation-and-adrs skill, adapted from agent-skills, says which docs a change touches and how to write an ADR; docs are judgment, never a gate. Deploy rules for Vercel: the vercel-deploy-bridge skill keeps the token in the environment, uploads and commits no .env file, defaults to preview and waits for a yes before every stage, commit, push, link, global install, anonymous --temporary deploy and production deploy. The upstream deploy-to-vercel skill installs via /pro-dev install. /pro-ship:install sets up its engines when pro-core is installed.
6 skills · 1 command · hooks
Deep web research engine for any topic. /research <question> runs an adaptive intake interview, then decomposes into angles, fans out research subagents over a free-first retrieval ladder (Serper -> Jina/Firecrawl -> agent-browser), verifies citations, and synthesizes a cited report. /lead-research <company|domain|ICP> is the lead/account specialization: same engine plus ICP scoring and cited lead profiles. Domain memory + evidence persisted to .research/ to survive compaction. Mieruka integration: drives the Research tab via MCP tools (list_research_threads, get_research_thread, update_research_thread) when .mieruka/ is present; falls back to file-mirror.
1 skill · 2 commands · 4 agents
The Security phase. security-architect challenges a specific idea, plan, spec or diff before it ships: trust boundaries, unstated assumptions and the attack that breaks each, returned as a ## Security challenge section (/threat-model, plus a read-only security-architect subagent). security-audit is a read-only whole-repository audit covering secrets in git history, dependency supply chain, CI/CD, containers and IaC, webhooks, LLM/AI attack surface, agent-skill supply chain, OWASP Top 10 and STRIDE, with confidence-scored findings and the security-auditor subagent judging each one independently (/security-audit). One push gate hook holds a git push or gh pr create whose diff against its upstream touches security surface, once per distinct diff, so the architect reviews that diff and nothing else. A key-literal hook on Edit, Write and MultiEdit flags a live-looking key by kind and line number, never by value: warn by default, deny before the write only when mode.key-literal says enforce. Adapted from the cso skill in garrytan/gstack.
2 skills · 2 commands · 2 agents · hooks
Next.js skills: next-best-practices (App Router conventions written from the Next.js docs: file conventions, Server and Client Components, async request APIs, data, errors, metadata, images, fonts), next-upgrade (structured upgrade workflow with the official codemods), next-skills-bridge (routes to the MIT Next.js agent skills in vercel/next.js for Cache Components, Partial Prefetching, bundles and the dev loop; /pro-dev install installs them), shadcn (component installation, theming, composition patterns). Also installs vercel and figma from claude-plugins-official via install-companions.sh. vercel-skills-bridge routes React and Next.js performance and Vercel cost tuning to two pinned vercel-labs skills. /pro-nextjs:install sets up its engines when pro-core is installed.
5 skills · 1 command
Data & auth skills: Drizzle ORM (architecture + schema), Prisma schema patterns, NextAuth/Auth.js patterns
4 skills · hooks
The Verify phase: native Vitest (unit/component), agent-browser (interactive verification), and Storybook interactions, plus qa-suite — a bridge to the petrkindlmann/qa-skills library installed on demand via /pro-dev install (npx skills add): qa-do/qa-start routers, playwright-automation, visual-testing, api/contract-testing, test-reliability, and QA strategy/risk/planning. Playwright is provided on demand via the qa-suite bridge (/pro-dev install), not as a hard plugin dependency. /pro-testing:install sets up its engines when pro-core is installed.
4 skills · 1 command
Astro skills: astro-best-practices (auto-applied guardrails for routing, output modes, islands and server islands, the content layer, actions, sessions, astro:env, images, view transitions and agent-friendly dev servers), astro-upgrade (major-version upgrade workflow for Astro 5 to 6 to 7 and its adapters), astro-on-cloudflare (@astrojs/cloudflare on Workers: cloudflare:workers env, wrangler bindings, sessions on KV, Pages-to-Workers migration) and astro-on-vercel (@astrojs/vercel: on-demand routes, maxDuration, ISR, redirects, images). Points agents at the official Astro Docs MCP for lookups. Pairs with the optional cloudflare/skills and vercel@claude-plugins-official companions.
4 skills
Glue for Google's Elephant-Goldfish Model (egm@egm): /pro-egm:goldfish runs its review loop on a spec or plan with goldfish reviewers a repo keeps in .egm/, shows the resolved reviewers and reading list first and confines custom paths to the repo. The egm-bridge skill points to /pro-dev install, which installs egm@egm. Opt-in: pro-ai-sdlc offers the review at /spec and /plan once it is installed. /pro-egm:install sets up its engines when pro-core is installed.
1 skill · 2 commands · 1 agent
Electron skills: electron-skills-bridge (routes Electron work to two pinned upstream engines installed with /pro-dev install: jwynia electron-best-practices for architecture, security, IPC and packaging; laststance electron-release for signing, notarization and publishing), electron-upgrade (major-version upgrade workflow for Electron and its tooling) and electron-security-audit (read-only audit of an Electron app against the official security checklist). Driving the running app through Playwright MCP over CDP is a manual setup the skills describe, not something the plugin installs.
3 skills · 1 command
Execution discipline skills: TDD, systematic debugging, git worktrees and subagent-driven development. Depends on pro-core for universal guardrails and repo safety.
4 skills
The Legal phase: software-counsel, an in-house product counsel persona that spots legal issues in plans and releases. Plan mode reviews an idea, plan or spec before code exists and returns a ## Legal review section; release mode reviews what is about to ship. Covers open-source licensing and attribution, privacy and data protection, AI regulation and model terms, scraping and third-party API terms, consumer protection and auto-renewal, children's data, health and financial data, export controls, accessibility, trademarks in names and client IP for consulting work. Ships the software-counsel subagent and /legal-review. Hooks enforce it: an ExitPlanMode gate and a plan-document gate require an independent review on plans that touch legal surface, a UserPromptSubmit nudge flags legal surface while an idea is cheap to change, and a release gate holds git tag, gh release, npm publish, production deploys, PR creation and dependency-manifest commits once for review. A dated legal-landscape reference is kept current by a staleness check. Issue spotting, not legal advice.
1 skill · 1 command · 1 agent · hooks
Motion and video skills: remotion (React-based programmatic video rendering), hyperframes (advanced Framer Motion orchestration), and gsap (GSAP timeline + ScrollTrigger animations for marketing and hero surfaces).
3 skills
Glue for GitHub's spec-kit: a bridge skill for its constitution, specify, plan, tasks and implement flow plus a remove command that takes it out of a repo. spec-kit and its specify CLI are installed by /pro-dev install or /pro-speckit:install, which needs pro-core installed.
1 skill · 2 commands
Meta-plugin: installs the default pro dev stack (core, pdd, execution, quality, design, motion, testing, data, research, security and ship). Run /pro-dev install after it. pro-ai-sdlc, pro-nextjs, pro-astro and pro-electron are opt-in and installed by hand.
Deprecated shim for one release: renamed to pro-dev-starter. Run /pro-dev update to switch.