You get:
- raw requests calls everywhere (duplication)
- no authentication handling (hardcoded tokens)
- poor error handling (unhandled HTTP errors)
- no rate limiting (get blocked)
- no retry logic (transient failures crash)
But an API wrapper is not a collection of requests.
It is a reusable client with consistent patterns.
- Authentication: API key, OAuth, or token
- Session management: reuse connection, handle headers
- Endpoint methods: named methods for each API call
- Error handling: raise custom exceptions for HTTP errors
- Rate limiting: respect API limits with backoff
Without a wrapper, API integration is fragile.
This framework forces AI to build robust API clients.
Assume the role of a Python developer who builds robust API wrappers. Your task is to create a Python class that wraps an external API. Generate: 1. CLASS DEFINITION - __init__ with authentication - Session management 2. AUTHENTICATION METHOD - API key, OAuth, or token handling 3. REQUEST METHOD (_request) - Handles HTTP methods - Adds headers, params - Error handling (timeout, status codes) 4. ENDPOINT METHODS (3-5) - Named methods for each API endpoint - Type hints for parameters and returns - Docstrings 5. ERROR HANDLING - Custom exception classes - HTTP status code handling 6. USAGE EXAMPLE - How to instantiate and use the wrapper INPUTS: API Name: [INSERT] Base URL: [INSERT] Authentication Type: [API KEY / BEARER TOKEN / OAUTH / BASIC AUTH] Endpoints to Wrap (3-5 with descriptions): [LIST] Rate Limit (if known): [E.G., "100 requests per minute"] RULES: - Use requests.Session for connection reuse - Raise custom exceptions for HTTP errors (not just print) - Include timeout parameter for all requests - Add retry logic for transient failures (5xx errors) - Type hints for all public methods - Docstrings for all public methods - Keep API keys out of code (use environment variables)
- Get API documentation before using this prompt (base URL, endpoints, auth).
- Test the generated wrapper with a real API key (use a test environment).
- Add logging for debugging production issues.
- Handle pagination if the API uses it (the prompt doesn’t auto-handle).
- Add async support if needed (separate prompt).
API Name: GitHub API
Base URL: https://api.github.com
Authentication Type: BEARER TOKEN (Personal Access Token)
Endpoints to Wrap: GET /user (get authenticated user), GET /user/repos (list user repos), GET /repos/{owner}/{repo} (get specific repo), POST /repos/{owner}/{repo}/issues (create issue)
Rate Limit: 5000 requests per hour
This framework improves outcomes by forcing:
- session management (efficiency)
- authentication handling (security)
- error handling (robustness)
- rate limiting (compliance)
- retry logic (resilience)
Great API wrappers don’t just work — they fail gracefully and handle edge cases.
Build Better AI Systems
Subscribe for advanced prompt engineering, AI coding tools, Python frameworks, and practical strategies for developers and engineers.