generation_failed — Cascade generation failed

What causes it

• The designer rejected the prompt as off-topic or outside what a cascade can express (the message explains the rejection).
• The generated definition failed validation even after the automatic repair pass; `details[]` lists the errors on `definition`.
• No data products are enabled for the account, so there is nothing to ground a live cascade in.
• The model could not be reached or returned an unexpected response. These are transient; the message says to try again.

How to fix it

1. Reword the prompt to name the screening products and the decision rules explicitly; short, concrete prompts generate cleaner cascades.
2. Send `create: false` to preview the generated definition and warnings, adjust, then create it with `POST /workflows`.
3. Retry transient failures with backoff; if products are the problem, enable them on the account first.

Operations that return it

• POST /workflows/generate Generate a workflow from a prompt

Example response

HTTP/1.1 422 Unprocessable Content · Content-Type: application/json

{
  "error": {
    "code": "generation_failed",
    "message": "We couldn't reach the cascade designer. Please try again.",
    "details": [
      {
        "field": "definition",
        "message": "node \"decision_1\" references unknown input \"credit_score\""
      }
    ],
    "request_id": "6f1c2a3e-8d4b-4c5e-9f0a-1b2c3d4e5f60",
    "doc_url": "https://cascades.crscreditapi.com/docs/api/errors/generation_failed"
  }
}

Related

• Guides › Ai generate a cascade