Recipe List

Every KitRocket project includes 12 Prompt Recipes in the /prompts/ directory. Here's what each one does.

implement

File: prompts/implement.md

Build a complete feature from scratch. This is the most versatile recipe — it handles database schema, API routes, business logic, and UI.

When to use: You want to add a new feature to your app.

Example:

claude "Read prompts/implement.md. Add a team workspace system where users can create teams, invite members, and share resources."

What it produces: Database tables, API endpoints, service layer, UI components, and navigation updates.

review

File: prompts/review.md

Code review with actionable feedback. Checks for bugs, security issues, performance problems, and adherence to project conventions.

When to use: After implementing a feature, before committing.

Example:

claude "Read prompts/review.md. Review the changes I made to the team system in the last 5 commits."

What it produces: A structured review with issues categorized as critical, high, medium, or low priority, plus specific fix suggestions.

fix-bug

File: prompts/fix-bug.md

Systematic debugging. Analyzes symptoms, traces the issue through the codebase, identifies root cause, and applies a fix.

When to use: Something is broken and you're not sure why.

Example:

claude "Read prompts/fix-bug.md. Users are getting a 500 error when trying to update their profile. The error started after the last deployment."

What it produces: Root cause analysis, the fix, and a test to prevent regression.

add-api

File: prompts/add-api.md

Create a new API endpoint with input validation, authentication, error handling, and proper HTTP status codes.

When to use: You need a new backend endpoint.

Example:

claude "Read prompts/add-api.md. Add a POST /api/feedback endpoint where authenticated users can submit feedback with a message and rating (1-5)."

What it produces: Route handler, zod validation schema, service function, database operations, and TypeScript types.

add-page

File: prompts/add-page.md

Add a new page with layout, metadata, data fetching, and navigation. Follows the existing route group pattern.

When to use: You need a new page in the dashboard or marketing site.

Example:

claude "Read prompts/add-page.md. Add a /dashboard/analytics page that shows charts of user activity over the last 30 days."

What it produces: Page component, metadata, data fetching, layout integration, and sidebar navigation link.

add-component

File: prompts/add-component.md

Create a reusable UI component with proper TypeScript props, variants, and accessibility.

When to use: You need a new UI building block.

Example:

claude "Read prompts/add-component.md. Create a StatusBadge component that shows a colored dot and label for subscription statuses: active (green), cancelled (red), past_due (yellow), trialing (blue)."

What it produces: A typed component with variants, built on shadcn/ui primitives, with proper ARIA attributes.

refactor

File: prompts/refactor.md

Improve code quality without changing behavior. Extracts functions, reduces duplication, improves naming, and simplifies logic.

When to use: Code works but is messy, duplicated, or hard to follow.

Example:

claude "Read prompts/refactor.md. Refactor src/lib/payments/webhook.ts — it's over 300 lines and the switch statement is getting unwieldy."

What it produces: Cleaner code with the same behavior, split into smaller functions or files if needed.

test

File: prompts/test.md

Write unit and integration tests for existing code. Covers happy paths, edge cases, and error scenarios.

When to use: You need tests for code that doesn't have them.

Example:

claude "Read prompts/test.md. Write tests for the checkout flow in src/lib/payments/checkout.ts and src/app/api/checkout/route.ts."

What it produces: Test files with comprehensive coverage, mocking external services, and testing both success and failure paths.

deploy

File: prompts/deploy.md

Pre-deployment checklist. Verifies environment variables, database migrations, build success, and configuration.

When to use: Before deploying to production or a new environment.

Example:

claude "Read prompts/deploy.md. I'm deploying to Vercel for the first time. Check everything."

What it produces: A checklist of issues found and fixes applied, plus manual steps you need to complete.

security

File: prompts/security.md

Security audit of your codebase. Checks for hardcoded secrets, SQL injection, XSS, missing auth checks, and more.

When to use: Before launch, after major changes, or on a regular schedule.

Example:

claude "Read prompts/security.md. Audit the entire src/app/api/ directory for security issues."

What it produces: A prioritized list of vulnerabilities with severity ratings and fix instructions.

performance

File: prompts/performance.md

Find and fix performance bottlenecks. Checks for unnecessary re-renders, missing indexes, N+1 queries, large bundles, and slow API routes.

When to use: The app feels slow or you want to optimize before scaling.

Example:

claude "Read prompts/performance.md. The dashboard page takes 3 seconds to load. Find out why and fix it."

What it produces: Performance analysis with specific optimizations — database indexes, query improvements, caching, code splitting.

document

File: prompts/document.md

Generate documentation for your code. Creates JSDoc comments, README sections, API docs, or architecture docs.

When to use: Code needs documentation for your team or API consumers.

Example:

claude "Read prompts/document.md. Document all API routes in src/app/api/ with request/response examples."

What it produces: Documentation in the format you specify — inline comments, markdown files, or OpenAPI specs.

Using multiple recipes together

The most effective workflow combines recipes:

# 1. Plan and implement
claude "Read prompts/implement.md. Add a webhook retry system."

# 2. Review the implementation
claude "Read prompts/review.md. Review the webhook retry system."

# 3. Fix any issues found
claude "Read prompts/fix-bug.md. Fix the issues from the review."

# 4. Write tests
claude "Read prompts/test.md. Write tests for the webhook retry system."

# 5. Check security
claude "Read prompts/security.md. Audit the webhook retry system."