The strongest prompts in the library, scored and ranked. One click copies the full text — role, workflow, constraints, output contract — ready for any model.
Filter by discipline or search across every word of every prompt. Scores rank the best-performing picks first.
One button copies the entire prompt to your clipboard. No truncation, no "see full text at" links.
Replace the [HIGHLIGHTED] placeholders with your own context, then send it.
Fill a short brief, paste the whole file, and the AI designs and builds one complete award-grade website — no lorem ipsum, no stalling.
Use when: Any new website build where you want an authored piece, not a template instantiation.
Refines a website into a distinctive, senior-crafted product — with an explicit blacklist of the clichés that make a site read as AI-generated.
Use when: An existing site or draft that looks templated and needs a principal-level craft pass.
Answer an intake, paste, and Claude Code plans, builds, then screenshots and judges its own work across three self-review passes.
Use when: Immersive, experience-led sites where you want the builder to critique and iterate on itself.
The execution half of APEX: craft system, archetype paradigms, performance and accessibility budgets, build order and acceptance checklist.
Use when: You already have a brief written — paste this to set the quality bar and build order.
Principal-engineer repo audit that finds, prioritizes, fixes, and documents every verifiable bug with a proving test.
Use when: Run before a major release or after inheriting an unfamiliar codebase.
Ethical pentester audits your full source against OWASP Top 10 and ASVS, with evidence-cited findings and a prioritized fix plan.
Use when: Use before shipping or exposing a web app to real users.
Forensic 11-dimension TypeScript review — type safety to architecture — with evidence-backed findings and corrected code.
Use when: Deep review of TypeScript modules or whole codebases before merge or refactor.
A pragmatic co-founder persona that takes a raw idea through discovery, planning, building, polish, and handoff — you stay in control.
Use when: Non-technical or busy founders building a real product with AI as the engineer.
Six-step architect audit rating readability, maturity, threats, and edge cases, ending in a Must/Should/Nice refactor roadmap.
Use when: Assess unfamiliar or inherited code for production readiness.
Generates a sub-150-line, high-signal CLAUDE.md/AGENTS.md instruction file tuned to your project's real conventions and gotchas.
Use when: Setting up Claude Code, Cursor, or any coding agent on a new repo.
Applies Butterick and Bringhurst principles to produce complete, accessible CSS typography systems with loading strategy and type scale.
Use when: Building or overhauling a site's typography with production CSS or Tailwind.
Four-phase debugging discipline: reproduce with failing tests, rank hypotheses with evidence, try isolated fixes, merge the best.
Use when: Fixing bugs in a codebase where regressions and guess-fixes are unacceptable.
A rigorous build workflow: discover context, plan atomic todos, execute methodically, and report — with a deep-think protocol throughout.
Use when: Complex coding tasks where careful reasoning beats fast guessing.
Framework-fluent frontend specialist delivering fast, accessible, maintainable UI code with explained tradeoffs and Core Web Vitals targets.
Use when: Building production UI in React, Vue, Angular, Svelte, or vanilla TS.
Production-grade analysis of Edge Function cold starts and their mobile UX impact, with per-function verdicts and latency budgets.
Use when: Diagnosing slow launches and blocking calls in an Expo + Supabase app.
Peer-level architecture sparring: clarify intent, pressure-test tradeoffs, propose alternatives — diagrams only after alignment.
Use when: Pressure-testing a system design before committing to diagrams or code.
Eleven-dimension forensic PHP audit — security-first, OWASP/CWE-referenced — with secure fixes and a remediation roadmap.
Use when: Auditing legacy or production PHP for vulnerabilities and rot.
Use to have the model design a complete, reusable prompt template for any task you describe - it engineers the inputs, structure, guardrails, and output…