api-design

Good API design makes the right thing easy and the wrong thing hard. This skill helps you create APIs that are a pleasure to use and maintain. </objective>

<quick_start> Resource naming: Nouns, plural, lowercase, hyphenated (/users, /blog-posts)

HTTP methods: GET (read), POST (create), PUT (replace), PATCH (update), DELETE (remove)

Status codes: 200 OK, 201 Created, 204 No Content, 400 Bad Request, 401 Unauthorized, 403 Forbidden, 404 Not Found, 429 Rate Limited

Pagination: Prefer cursor-based for real-time data; use ?limit=20&cursor=xxx

Versioning: URL path (/api/v1/) is clearest approach </quick_start>

<success_criteria> API design is successful when:

• Resources use plural nouns with consistent naming
• HTTP methods match semantics (GET=read, POST=create, etc.)
• Error responses follow consistent format with code, message, details, requestId
• Pagination implemented (cursor or offset based)
• Rate limiting headers included (X-RateLimit-Limit, Remaining, Reset)
• Versioning strategy defined before breaking changes needed </success_criteria>

<core_principles>

API Design Principles

1. Consistency - Same patterns everywhere (naming, errors, pagination)
2. Predictability - Developers can guess how things work
3. Simplicity - Easy cases should be easy, complex cases possible
4. Backwards compatibility - Don't break existing clients
5. Self-documenting - Clear naming, helpful error messages </core_principles>

<rest_basics>

RESTful Conventions

Resource Naming

# GOOD: Nouns, plural, lowercase, hyphenated
GET    /users
GET    /users/{id}
GET    /users/{id}/posts
GET    /blog-posts
GET    /api/v1/user-preferences

# BAD: Verbs, singular, mixed case, underscores
GET    /getUser
GET    /user/{id}
GET    /User/{id}/getPosts
GET    /blog_posts

HTTP Methods

*PATCH is idempotent if you apply the same patch

CRUD Operations

# Collection operations
GET    /users           # List all users
POST   /users           # Create a user

# Single resource operations
GET    /users/{id}      # Get one user
PUT    /users/{id}      # Replace user
PATCH  /users/{id}      # Update user fields
DELETE /users/{id}      # Delete user

# Nested resources
GET    /users/{id}/posts       # User's posts
POST   /users/{id}/posts       # Create post for user
GET    /posts/{id}/comments    # Post's comments

Actions (Non-CRUD Operations)

# When you need actions, use verbs as sub-resources
POST   /users/{id}/activate
POST   /users/{id}/deactivate
POST   /orders/{id}/cancel
POST   /invoices/{id}/send
POST   /auth/login
POST   /auth/logout
POST   /auth/refresh

Status Codes

<error_handling>

Error Responses

Consistent Error Format

// Standard error response
interface ErrorResponse {
  error: {
    code: string;           // Machine-readable code
    message: string;        // Human-readable message
    details?: ErrorDetail[]; // Field-level errors
    requestId?: string;     // For support/debugging
  };
}

interface ErrorDetail {
  field: string;
  message: string;
  code: string;
}

Examples

// 400 Bad Request - Validation error
{
  "error": {
    "code": "validation_error",
    "message": "Invalid request parameters",
    "details": [
      { "field": "email", "message": "Invalid email format", "code": "invalid_format" },
      { "field": "age", "message": "Must be at least 13", "code": "min_value" }
    ],
    "requestId": "req_abc123"
  }
}

// 401 Unauthorized
{
  "error": {
    "code": "unauthorized",
    "message": "Invalid or expired authentication token",
    "requestId": "req_abc123"
  }
}

// 403 Forbidden
{
  "error": {
    "code": "forbidden",
    "message": "You don't have permission to access this resource",
    "requestId": "req_abc123"
  }
}

// 404 Not Found
{
  "error": {
    "code": "not_found",
    "message": "User not found",
    "requestId": "req_abc123"
  }
}

// 429 Rate Limited
{
  "error": {
    "code": "rate_limited",
    "message": "Too many requests. Please retry after 60 seconds",
    "requestId": "req_abc123"
  }
}

Implementation

// Error class
class ApiError extends Error {
  constructor(
    public statusCode: number,
    public code: string,
    message: string,
    public details?: ErrorDetail[]
  ) {
    super(message);
  }
}

// Error handler middleware
function errorHandler(error: Error, req: Request, res: Response) {
  const requestId = req.headers['x-request-id'] || crypto.randomUUID();

  if (error instanceof ApiError) {
    return res.status(error.statusCode).json({
      error: {
        code: error.code,
        message: error.message,
        details: error.details,
        requestId,
      },
    });
  }

  // Log unexpected errors
  logger.error('Unexpected error', { error, requestId });

  // Don't expose internal details
  return res.status(500).json({
    error: {
      code: 'internal_error',
      message: 'An unexpected error occurred',
      requestId,
    },
  });
}

</error_handling>

Cursor-Based (Recommended)

// Request
GET /posts?limit=20&cursor=eyJpZCI6MTAwfQ

// Response
{
  "data": [...],
  "pagination": {
    "hasMore": true,
    "nextCursor": "eyJpZCI6MTIwfQ",
    "prevCursor": "eyJpZCI6MTAwfQ"
  }
}

Pros: Consistent results, handles real-time data Cons: Can't jump to page N

Offset-Based

// Request
GET /posts?page=2&limit=20
// or
GET /posts?offset=20&limit=20

// Response
{
  "data": [...],
  "pagination": {
    "page": 2,
    "limit": 20,
    "total": 150,
    "totalPages": 8
  }
}

Pros: Can jump to any page Cons: Inconsistent with real-time data, slow on large tables

Implementation (Cursor)

// Encode/decode cursor
function encodeCursor(data: object): string {
  return Buffer.from(JSON.stringify(data)).toString('base64url');
}

function decodeCursor(cursor: string): object {
  return JSON.parse(Buffer.from(cursor, 'base64url').toString());
}

// Query with cursor
async function getPosts(limit: number, cursor?: string) {
  const where: any = {};

  if (cursor) {
    const { id } = decodeCursor(cursor);
    where.id = { lt: id };
  }

  const posts = await db.post.findMany({
    where,
    orderBy: { id: 'desc' },
    take: limit + 1, // Fetch one extra to check hasMore
  });

  const hasMore = posts.length > limit;
  const data = hasMore ? posts.slice(0, -1) : posts;

  return {
    data,
    pagination: {
      hasMore,
      nextCursor: hasMore ? encodeCursor({ id: data[data.length - 1].id }) : null,
    },
  };
}

Use query params: ?status=active, ?sort=-created_at (prefix - for desc), ?fields=id,name,email. Validate with Zod schema, build Prisma where clause from parsed filters.

See reference/filtering-sorting.md for full query parameter patterns and implementation. </filtering>

URL Path Versioning (Recommended)

GET /api/v1/users
GET /api/v2/users

Pros: Clear, easy to understand Cons: More maintenance

Header Versioning

GET /api/users
Accept: application/vnd.api+json; version=2

Pros: Clean URLs Cons: Hidden, harder to test

When to Version

Create a new version when:

• Removing fields from responses
• Changing field types or formats
• Removing endpoints
• Changing authentication

Don't create a new version for:

• Adding new optional fields
• Adding new endpoints
• Adding new optional parameters
• Bug fixes </versioning>

<rate_limiting>

Rate Limiting

Include headers: X-RateLimit-Limit, X-RateLimit-Remaining, X-RateLimit-Reset. Return 429 with Retry-After when exceeded. Use sliding window algorithm (e.g., @upstash/ratelimit).

See reference/rate-limiting.md for tier definitions and implementation. </rate_limiting>

To load: Ask for the specific topic or check if context suggests it. </references>

Emit Outcome Sidecar

As the final step, write to ~/.claude/skill-analytics/last-outcome-api-design.json:

{"ts":"[UTC ISO8601]","skill":"api-design","version":"1.0.0","variant":"default",
 "status":"[success|partial|error]","runtime_ms":[estimated ms from start],
 "metrics":{"endpoints_designed":[n],"patterns_applied":[n]},
 "error":null,"session_id":"[YYYY-MM-DD]"}

Use status "partial" if some stages failed but results were produced. Use "error" only if no output was generated.