asset_analyze

Analyze a Scenario asset and/or text with Scenario LLM — flexible, instruction-driven analysis (caption, describe, classify, extract, answer questions, translate). For dedicated/typed outputs use asset_caption, asset_describe, or asset_detect instead. Args: - instruction: required — what to do (e.g. "Caption this image", "List the objects", "Translate to French") - images: optional — image asset IDs for vision (up to 10) - text_inputs: optional — extra text context (documents, examples) (up to 10) - num_outputs: optional — number of distinct results (1–5) - model: optional — underlying LLM (default gemini-2.5-flash) - thinking_level: optional — reasoning depth (Gemini 3 Flash Preview only) - wait: optional — wait for completion, else returns job_id to track via jobs_wait (or job_get for a status snapshot) - dry_run: optional — true returns a cost estimate without running or charging credits - team_id, project_id: required for OAuth callers Returns: a model-run job result (output text asset(s) + job_id), or status='in_progress' with a job_id if still running. With dry_run=true, returns a cost estimate. Don't use when: you need a control-map image (use asset_detect) or a final user-facing generation (use model_run).

{
  "instruction": "Caption this image and list the main objects.",
  "images": [
    "asset_img001"
  ],
  "team_id": "team_abc123",
  "project_id": "proj_xyz789"
}

{
  "status": "success",
  "job_id": "job_abc",
  "model_id": "model_scenario-llm",
  "assets": [
    {
      "id": "asset_txt001",
      "app_url": "https://app.scenario.com/..."
    }
  ],
  "message": "Completed. Call asset_display with an asset_id to show the result to the user."
}