ChatGPT API Documentation Writer Prompt
Write clear complete API documentation that developers can actually use — endpoints, parameters, examples, and every error code.
Category
💻 Coding
Difficulty
Intermediate
Models
3
Last Updated
2026-06-29
Works with
📄 Example output
⚠️ Common Mistakes
❓ FAQ
⚙️ Fill in your variables
📋 Prompt
You are a senior technical writer who has documented APIs used by 100,000+ developers.
API name: [name]
Endpoint: [path and HTTP method]
Purpose: [what the endpoint does]
Authentication: [API key/OAuth/JWT/none]
Request parameters: [path, query, and body fields]
Response: [what it returns on success]
Error cases: [common error scenarios]
Task:
1. ENDPOINT OVERVIEW: One-sentence description
2. AUTHENTICATION: How to authenticate with a working example
3. REQUEST FORMAT: Method and URL, required headers, parameters table with name/type/required/description/example, curl example
4. RESPONSE FORMAT: Success response with annotated JSON, fields table with field/type/description
5. ERROR RESPONSES: Table of error codes, HTTP status, cause, and resolution
6. CODE EXAMPLE: Working snippet in Python or JavaScript
7. RATE LIMITS AND NOTES: Any limitations developers need to know
API name: [name]
Endpoint: [path and HTTP method]
Purpose: [what the endpoint does]
Authentication: [API key/OAuth/JWT/none]
Request parameters: [path, query, and body fields]
Response: [what it returns on success]
Error cases: [common error scenarios]
Task:
1. ENDPOINT OVERVIEW: One-sentence description
2. AUTHENTICATION: How to authenticate with a working example
3. REQUEST FORMAT: Method and URL, required headers, parameters table with name/type/required/description/example, curl example
4. RESPONSE FORMAT: Success response with annotated JSON, fields table with field/type/description
5. ERROR RESPONSES: Table of error codes, HTTP status, cause, and resolution
6. CODE EXAMPLE: Working snippet in Python or JavaScript
7. RATE LIMITS AND NOTES: Any limitations developers need to know
# GET /v1/prompts
Retrieve a paginated list of AI prompts. Supports filtering by category and model.
## Authentication
Authorization: Bearer YOUR_API_KEY
Get your API key: ToolsNova dashboard -> Settings -> API.
## Request
URL: https://api.toolsnova.net/v1/prompts
| Parameter | Type | Required | Default | Description |
|-----------|------|----------|---------|-------------|
| category | string | No | - | One of: writing, coding, marketing, seo, business, trading, finance, career, education, social, productivity, image |
| model | string | No | - | One of: chatgpt, claude, gemini, grok, deepseek |
| limit | integer | No | 20 | Max: 100 |
| offset | integer | No | 0 | Pagination offset |
curl 'https://api.toolsnova.net/v1/prompts?category=coding&limit=5' -H 'Authorization: Bearer YOUR_API_KEY'
## Response 200 OK
{"total": 18, "limit": 5, "offset": 0, "data": [{"id": 158, "slug": "sql-advanced-queries", "category": "coding", "best_model": "deepseek"}]}
## Errors
| Status | Code | Cause | Resolution |
|--------|------|-------|------------|
| 401 | INVALID_KEY | Missing or invalid API key | Check key in dashboard |
| 400 | INVALID_CATEGORY | Category not in allowed list | Use one of the 12 valid categories |
| 429 | RATE_LIMITED | Over 100 requests per minute | Implement exponential backoff |
Retrieve a paginated list of AI prompts. Supports filtering by category and model.
## Authentication
Authorization: Bearer YOUR_API_KEY
Get your API key: ToolsNova dashboard -> Settings -> API.
## Request
URL: https://api.toolsnova.net/v1/prompts
| Parameter | Type | Required | Default | Description |
|-----------|------|----------|---------|-------------|
| category | string | No | - | One of: writing, coding, marketing, seo, business, trading, finance, career, education, social, productivity, image |
| model | string | No | - | One of: chatgpt, claude, gemini, grok, deepseek |
| limit | integer | No | 20 | Max: 100 |
| offset | integer | No | 0 | Pagination offset |
curl 'https://api.toolsnova.net/v1/prompts?category=coding&limit=5' -H 'Authorization: Bearer YOUR_API_KEY'
## Response 200 OK
{"total": 18, "limit": 5, "offset": 0, "data": [{"id": 158, "slug": "sql-advanced-queries", "category": "coding", "best_model": "deepseek"}]}
## Errors
| Status | Code | Cause | Resolution |
|--------|------|-------|------------|
| 401 | INVALID_KEY | Missing or invalid API key | Check key in dashboard |
| 400 | INVALID_CATEGORY | Category not in allowed list | Use one of the 12 valid categories |
| 429 | RATE_LIMITED | Over 100 requests per minute | Implement exponential backoff |
🏆
💡 Pro Tips
Best model for this prompt
Claude
Claude (Opus 4 / Sonnet 4)
Always include a working curl example — it is the first thing most developers try
Error responses are as important as success responses — document every error with its resolution
Parameter tables with defaults and constraints prevent the most common support tickets
Code examples should be copy-paste ready — test them yourself before publishing
Documenting only the happy path — developers need to know what happens when things go wrong
Vague parameter descriptions — the ID of the user is insufficient; specify format, constraints, and where to find it
No authentication example — auth is always the first place developers get stuck
Inconsistent naming conventions — if you use snake_case in one response, use it everywhere
- What format for API docs?Markdown is standard for developer docs. OpenAPI/Swagger YAML for machine-readable specs that auto-generate interactive docs. Many teams use both: OpenAPI for the spec, Markdown for narrative guides.
- Should I auto-generate docs from code?Auto-generated docs ensure accuracy but are often sparse. The best docs combine auto-generated reference with hand-written guides, tutorials, and examples.
- Best model?Claude produces the most technically accurate and readable API docs — understands REST conventions deeply and uses correct technical terminology consistently.
- How to keep docs up to date?Treat docs as code — update in the same PR as the API change. Automated tests that verify example requests against the live API are the most reliable way to catch stale documentation.