Results

get_prompt_results

The latest run's result for each tracked question on each engine: won or lost, rank, sentiment, endorsement and quality.

readReads only. Annotated read-only, so a client can call it without asking.

Plan: Grow and above
Rate cost: 1 unit
REST: GET /projects/{project_id}/results

Try asking

“Which questions did we lose in the latest run?”

get_prompt_results

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_prompt_results",
    "arguments": {
      "project_id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
      "page_size": 50
    }
  }
}

structuredContent

{
  "content": [
    {
      "type": "text",
      "text": "…the same JSON, as text…"
    }
  ],
  "structuredContent": {
    "probes": [
      {
        "id": "5b1f2c3e-8d4a-4f6b-9c2e-1a7d3e9f0b21",
        "prompt_id": "0e9a7c55-2b3d-4c1e-8f6a-9d0b1c2e3f4a",
        "prompt_ref": 12,
        "prompt_text": "best invoicing software for freelancers in India",
        "platform": "perplexity",
        "won": false,
        "brand_position": 2,
        "sentiment": "neutral",
        "endorsement": "neutral",
        "quality": 0.429,
        "top_competitor": "Northwind Billing",
        "citation_count": 6,
        "competitor_count": 3,
        "reasoning_card_id": "a3c9e1f0-4b2d-4e8a-9f1c-7d6b5a4e3c2b"
      }
    ],
    "page": {
      "page": 1,
      "page_size": 50,
      "total": 40
    }
  },
  "isError": false
}

Arguments

Passed as the tool call's arguments.

project_idstringrequired

The project to read, from list_projects.

prompt_idstring

Only this prompt's answers.

pageintegerdefault 1

Page number, from 1.

page_sizeintegerdefault 50

Rows per page, up to 100.

Result

The same object as List results, returned as structured output.

probesobject[]

Results from the latest run.

+Show child attributes

idstring

Result id. Pass it to Get a reasoning card.

prompt_idstring

The tracked prompt that was asked.

prompt_refinteger

The prompt's short number, as the dashboard shows it (#12). 0 when not available.

prompt_textstring

The prompt that was asked.

platformstring

Engine id.

Values:chatgptperplexitygeminicopilotai_overviewsclaude

wonboolean

True when the answer recommended the brand with a quality of at least 0.5.

brand_positionintegernullable

Where the brand is first named among itself and the tracked competitors, from 1. Null when the answer doesn't name it.

sentimentstringnullable

Tone toward the brand. Null when it isn't mentioned.

Values:positiveneutralnegative

endorsementstringnullable

How the answer treats the brand, from named as the one to choose to advised against. Null when the brand isn't mentioned.

Values:strongrecommendedneutralreference_onlydiscouraged

qualitynumber

The answer's quality for the brand, 0–1: endorsement weighted by position. 0.5 or more is a win.

top_competitorstringnullable

The first tracked competitor the answer names.

citation_countinteger

Distinct URLs the answer cites.

competitor_countinteger

Tracked competitors the answer names.

reasoning_card_idstringnullable

The “Why?” card explaining the result. Null when there isn't one.

pageobject

Where these rows sit in the full list.

+Show child attributes

pageinteger

This page's number, from 1.

page_sizeinteger

Rows per page.

totalinteger

Rows across every page.

Errors

404 not_found	The project or item doesn't exist in this workspace. A project from another workspace also returns 404, never 403.
422 invalid_request	A parameter or body field is missing or has the wrong type or value. The message names the field. Also returned when a write would pass the plan's prompt limit.
401 unauthorized	No key was sent, or the key is unknown, revoked or expired, or the member who created it has left the workspace or been deactivated.
402 plan_required	The workspace's plan doesn't include API access. It's included on Grow, Scale and Custom.
403 workspace_not_approved	The workspace is still waiting for approval, or wasn't approved.
429 rate_limited	Too many requests this minute. Wait for the number of seconds in the `Retry-After` header, then retry.

Every code, and what to retry, is in Errors.

Failures arrive as a tool error — isError: true with code: message as text. See Using the tools.