API documentation

Free public API, no login required. Generates code against 126,529 real API signatures from 805 libraries. Rate limits and request format documented here.

Endpoint

POST https://sdkagent.dev/api/generate
Content-Type: application/json

Request

{
  "question": "Write an httpx request with exponential backoff",
  "lang": "auto",
  "model": "poolside/laguna-s-2.1:free",
  "history": []
}

Field	Required	Description
`question`	Yes	Your question. Up to 4,000 characters.
`lang`	No	`auto` (default) detects the language from the question. Can be set explicitly to `python` and others.
`model`	No	Model to use. Defaults to a code-specialised model.
`history`	No	Conversation history as an array of `{"role","content"}` (last 6 entries).

Response

{
  "answer": "generated code and explanation",
  "context": "the real APIs passed to the model",
  "api_count": 14,
  "langs": ["python"],
  "auto_detected": true,
  "quota": { "used": 3, "limit": 20, "remaining": 17 }
}

Rate limits

The API is free and requires no login, but the following limits apply to keep the service running.

Window	Limit
Per day	20 requests
Per 60 seconds	5 requests
Question length	4,000 characters

Exceeding a limit returns 429. Remaining quota is reported in the quota field and the X-RateLimit-Remaining header.

Example

curl -X POST https://sdkagent.dev/api/generate \
  -H "Content-Type: application/json" \
  -d '{"question":"group and aggregate a DataFrame with pandas"}'

Errors

Code	Meaning
400	Malformed request
429	Rate limit exceeded
502	Upstream model error
503	Knowledge base or configuration unavailable

Terms of use

This API is provided free of charge with no guarantee of availability, latency or output accuracy. It may change or be discontinued without notice. Commercial use is permitted, but verifying generated code is your responsibility. See the terms of service.