What should your AI get better at?
Eight quick questions about the way you work. You get the skills and prompts to install first, and a free account keeps them.
Get my skill picksor browse all 34 resources
Ask your assistant for plain-language explanations, clear steps, and helpful suggestions as you learn.
Custom instructionsStarter
Build a source-connected knowledge system with project briefs, useful retrieval, a working capture routine, and a clear way to maintain it.
SkillStarter
Find where AI could save meaningful time in your recurring work, then try the most promising opportunity with a ready-to-paste starter prompt.
PromptStarter
Find what your AI loads before you give it a task, even when you just say hi, and reduce the clutter while preserving necessary instructions.
SkillPracticedCheck that important instructions load when they apply and stay out when they do not. Find false positives, false negatives, and rules that arrive too late.
SkillAdvancedImprove context for your current model by replacing recurring causes. Get copy-ready updates and a clear HTML audit.
PromptStarterFind the decisions and workflow defaults behind repeated mistakes, then replace their causes while preserving your requirements.
SkillPracticedLearn which decisions made your work useful, then preserve their benefits without copying incidental wording, layouts, or tools.
SkillPracticedUse earlier attempts to repair one recurring problem and replace the workflow that keeps producing it.
SkillPracticedRepair obsolete advice and conflicting workflows for your current model, with exact changes and a clear HTML audit.
SkillStarter
Research guidance for any task, examine evidence and criticism, and apply practices that fit your goals and circumstances.
SkillPracticed
Draft and edit with AI while keeping your voice. Use your writing and instructions to plan the piece, and replace generic habits without flattening the prose.
SkillPracticed
Make writing, interfaces, and learning experiences easier to understand and use while preserving information, depth, functionality, and quality.
SkillStarterExplore three distinct design directions as runnable prototypes, with real content, responsive layouts, and a clear choice between them.
SkillPracticedBuild refined, responsive UI from a live route, screenshot, or accepted mockup, then verify its design and interactions.
SkillAdvancedGive AI-built interfaces a design that belongs to your product. Plan from its audience and existing design system, replace generic defaults, and verify the rendered result.
SkillPracticed
Carry the goal, decisions, progress, and next action into a new chat without inventing missing history.
PromptStarterFind avoidable token consumption in your recent Claude Code work. Get evidence-backed proposals that preserve quality, completeness, and reliability.
SkillPracticedTurn a mixed working tree into the smallest truthful set of cohesive commits without sweeping unrelated work into the history.
SkillPracticedResearch, Plan, Implement, Verify, Refactor. Build features with clear responsibilities, safe renames, and behavior checks, and keep project guidance accurate as requirements change.
SkillAdvancedExtract what an app or feature must do, assess its data, logic, and UI, then guide a rebuild that accounts for every capability.
SkillPracticedFind and fix unclear names, functions, components, modules, and folders in React and TypeScript without changing observable behavior.
SkillAdvancedInstall Codex as a Claude Code reviewer and verify a completed manual review. Add automatic reviews when you choose.
PromptPracticedPrepare a complete plan another coding agent can follow, with the decisions, files, and checks it needs.
PromptAdvanced
Find the responsible software and workflow causes, fix them, and verify the result while preserving valid requirements.
SkillPracticed
Link the right Vercel project, add both official tracking packages, and verify that production traffic and real-user speed data can be collected.
PromptPracticedFind the routes users rely on, compare their real-user Web Vitals, and turn the strongest evidence into a prioritized improvement plan.
PromptAdvancedInstall the supported Supabase CLI, link the correct project, and establish read-only access to database linting and advisor evidence.
PromptPracticedTranslate Supabase Advisor findings into verified risks, context-dependent warnings, safe migrations, and production-ready checks.
PromptAdvancedInstall Lighthouse with the project, create repeatable mobile and desktop commands, and verify JSON and HTML reports on a real page.
PromptPracticedAudit the pages real users depend on, separate repeatable findings from run noise, and prioritize fixes without flattening the site's experience.
PromptAdvancedScan a public production site, explain each HTTP security finding, and plan compatible header fixes that preserve working user journeys.
PromptAdvancedInstall Sentry and its command-line tool, keep development noise out, preserve intentional staging, and verify production alerts when the app is ready.
PromptPracticedInvestigate a real production issue, prove its root cause, make the smallest safe fix, and verify the new release without hiding the error.
PromptAdvanced
