Claude Code Plugins

Community-maintained marketplace

Feedback

|

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name cursor-custom-prompts
description Create effective custom prompts for Cursor AI. Triggers on "cursor prompts", "prompt engineering cursor", "better cursor prompts", "cursor instructions".
allowed-tools Read, Write, Edit, Bash
version 1.0.0
license MIT
author Jeremy Longshore <jeremy@intentsolutions.io>

Custom Prompts for Cursor

Prompt Engineering Fundamentals

Effective Prompt Structure

[CONTEXT] What you're working on
[TASK] What you want to accomplish
[CONSTRAINTS] Requirements and limitations
[FORMAT] How you want the output
[EXAMPLES] Reference patterns if needed

Example Well-Structured Prompt

CONTEXT: Working on a Next.js 14 e-commerce app with TypeScript

TASK: Create a shopping cart hook that:
- Manages cart items in localStorage
- Syncs with server when user logs in
- Handles quantity updates
- Calculates totals with tax

CONSTRAINTS:
- Use Zustand for state management
- Follow existing patterns in @hooks/useAuth.ts
- Handle offline scenarios gracefully
- TypeScript strict mode

FORMAT:
- Single useCart.ts file
- Export named hook and types
- Include JSDoc comments

Prompt Templates

Code Generation

Template:
"Create [COMPONENT_TYPE] for [PURPOSE]:

Requirements:
- [Requirement 1]
- [Requirement 2]

Technology:
- [Framework/Library]
- [Styling approach]

Follow patterns from @[reference-file]

Include:
- TypeScript types
- Error handling
- Tests (optional)"

Code Review

Template:
"Review this code for:
- [ ] Security vulnerabilities
- [ ] Performance issues
- [ ] Best practice violations
- [ ] Missing error handling
- [ ] Type safety problems

Suggest specific improvements with code examples."

Debugging

Template:
"Debug this issue:

ERROR: [exact error message]

EXPECTED: [what should happen]

ACTUAL: [what actually happens]

RELEVANT CODE:
[code snippet]

What I've tried:
- [attempt 1]
- [attempt 2]"

Refactoring

Template:
"Refactor [CODE/FILE] to:
- [Goal 1]
- [Goal 2]

Preserve:
- [What must not change]
- [Behavior to maintain]

Apply patterns from:
- @[reference-file]"

Domain-Specific Prompts

API Development

"Create REST endpoint for [resource]:

Method: [GET/POST/PUT/DELETE]
Path: /api/[path]

Request:
- Body: [schema]
- Params: [params]
- Query: [query params]

Response:
- Success: [schema]
- Errors: [error codes]

Include:
- Input validation (Zod)
- Error handling
- TypeScript types
- Rate limiting (optional)"

React Components

"Create React component [Name]:

Purpose: [what it does]

Props:
- [prop1]: [type] - [description]
- [prop2]: [type] - [description]

State:
- [what state it manages]

Behavior:
- [interaction 1]
- [interaction 2]

Styling: Tailwind CSS
Accessibility: WCAG 2.1 AA"

Database Operations

"Create database [operation] for [entity]:

ORM: Prisma
Operation: [CRUD type]

Schema context:
@prisma/schema.prisma

Requirements:
- [Requirement 1]
- [Requirement 2]

Return:
- [What to return]

Handle:
- Not found
- Validation errors
- Database errors"

Advanced Techniques

Chain-of-Thought Prompting

"Let's approach this step by step:

1. First, analyze the current code structure
2. Identify what needs to change
3. Plan the refactoring steps
4. Implement each step
5. Verify the changes

Start by analyzing @[file]"

Few-Shot Learning

"Following this pattern:

Example 1:
Input: User model
Output: [show expected code]

Example 2:
Input: Product model
Output: [show expected code]

Now create for:
Input: Order model"

Role-Based Prompting

"Acting as a senior security engineer:

Review this authentication code for vulnerabilities:
@auth/login.ts

Focus on:
- Injection attacks
- Session handling
- Password security
- Token management"

Constraint Prompting

"Create a solution that:

MUST:
- Use existing types from @types/
- Follow error handling patterns
- Be fully typed

MUST NOT:
- Use any external libraries
- Modify existing interfaces
- Use 'any' type

SHOULD:
- Be performant for large datasets
- Include comprehensive comments"

Prompt Library

Quick Actions

Explain: "Explain what this code does in plain English"
Simplify: "Simplify this code while maintaining functionality"
Optimize: "Optimize this for performance"
Document: "Add comprehensive JSDoc comments"
Type: "Add TypeScript types to this JavaScript code"
Test: "Write unit tests for this function"

Transformations

"Convert this to TypeScript with strict types"
"Refactor to use async/await instead of callbacks"
"Extract this into a reusable hook"
"Convert class component to functional"
"Split this file into smaller modules"

Analysis

"What are potential bugs in this code?"
"How can this be more maintainable?"
"What edge cases are not handled?"
"Are there any security concerns?"
"What would break if [scenario]?"

.cursorrules Integration

Embedding Prompts

# .cursorrules

prompts:
  component:
    template: |
      Create React component following:
      - Functional component with TypeScript
      - Tailwind for styling
      - Handle loading/error states
      - Include prop types interface

  api:
    template: |
      Create API route following:
      - Zod validation
      - Proper error responses
      - TypeScript types
      - Logging included

Prompt Aliases

shortcuts:
  /component: "Create a new React component..."
  /api: "Create a new API endpoint..."
  /test: "Write tests for..."
  /review: "Review code for issues..."

Best Practices

Do's

DO:
- Be specific about requirements
- Provide context with @-mentions
- Include examples when helpful
- Specify constraints clearly
- Break complex tasks into steps
- Reference existing patterns

Don'ts

DON'T:
- Be vague ("make it better")
- Assume context is understood
- Ask for too much at once
- Ignore generated code review
- Use ambiguous terms
- Skip error case requirements

Iteration

Prompt refinement cycle:

1. Initial prompt
2. Review output
3. Identify gaps
4. Refine prompt
5. Regenerate
6. Repeat until satisfied

Track effective prompts for reuse