Create a backend

Use POST /v1/generate to start a backend generation job.

Request

curl -X POST "$AXTL_API_BASE_URL/v1/generate" \
  -H "Authorization: Bearer $AXTL_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Lead Qualification API",
    "prompt": "Create an API that accepts a lead profile, scores fit, lists missing fields, and returns next actions.",
    "region": "auto",
    "autoDeploy": false
  }'

Fields

Field	Required	Notes
`name`	Yes	Human-readable backend name. Verified schema: 2 to 120 characters.
`prompt`	Yes	Plain-English requirements. Verified schema: 10 to 100000 characters.
`model`	No	Model selection is managed by AXTL unless your workspace exposes advanced model settings.
`region`	No	Use `auto` unless your workspace settings specify otherwise.
`autoDeploy`	No	Defaults to `false` in current schema.

Response

AXTL returns 202 Accepted.

{
  "data": {
    "agent": {
      "id": "agent_...",
      "slug": "lead-qualification-api",
      "name": "Lead Qualification API",
      "status": "generating"
    },
    "job": {
      "id": "job_...",
      "status": "queued",
      "progress": 0,
      "provider": "pending",
      "model": "gpt-5.2"
    }
  }
}

The exact ID format is not part of the public contract yet.

Check status

curl "$AXTL_API_BASE_URL/v1/generate/{jobId}" \
  -H "Authorization: Bearer $AXTL_TOKEN"

The status response includes data.id, data.agentId, data.agentSlug, data.status, data.progress, data.logs, and optional artifact, validation, and failure fields.

Write better requirements

Good requirements include:

The input payload your API should accept.
The output shape you expect.
External systems the backend should call, if any.
Validation rules.
Failure behavior.