Trustpilot MCP server

How an agent would call it

Two steps: search returns refs, then data_fetch opens one of them. The arguments below are the arguments that were sent.

{
  "tool": "search",
  "arguments": {
    "query": "ryanair.com",
    "sources": [
      "trustpilot"
    ],
    "limit": 2,
    "runId": "<from run_begin>"
  }
}

{
  "tool": "data_fetch",
  "arguments": {
    "sourceId": "trustpilot",
    "ref": "<the ref object the search returned, verbatim — the review travels inside it>",
    "runId": "<from run_begin>"
  }
}

This runs against an account — send Authorization: Bearer <token> with the call. Get one at app.bedrocktap.com/dashboard/tokens; the free tier needs no card, and it reaches trustpilot like every other source.

What it costs. 1.5 credits per review row returned; a fetch costs nothing, because the row already carries the review. The search above returned 2 rows and cost 3 credits.

What came back

{
  "ref": {
    "id": "6aa5414d952fd0eb5569ebac",
    "url": "https://www.trustpilot.com/reviews/6aa5414d952fd0eb5569ebac",
    "title": "If i could give half a star"
  },
  "text": "If i could give half a star\nIf i could give half a star, I would",
  "fetchedAt": "2026-09-12T15:01:19.601Z"
}

Each review ref carried rating, date, verified, replied and the review body itself. Both rows in this search were one-star, posted within the same hour, and neither carried a company reply. Because the body travels with the ref, the fetch composed its payload from the row it already held and dispatched nothing.

Observed 2026-09-12. One call against a live source — ask again tomorrow and the rows move.

What a row contains

One row is one review: the review id, its permalink, and the review’s own title. meta carries the star rating, the publication date, whether the reviewer was verified, whether the company replied, and the full review text. The query must be a company domain or a Trustpilot URL — free text returns nothing at all, so a product name has to be resolved to a domain first. There is no per-review endpoint here, so `data_fetch` takes the whole ref object the search returned — the review lives in its `meta`. An id on its own carries no review and is refused.

A question it answers. What are this brand’s customers complaining about this week, and does the company answer them in public?

Routing guidance

What an agent reads to decide whether trustpilot is the right source, and what to reach for when it is not.

Answers

Reviews of a brand, shop or service the general public buys from. The most feature-complete review source here — native star filter, date range, sort, language filter, verified-only, and a real per-review permalink, so nothing needs post-filtering.

Query shape

A company domain or a full review-page URL.

Price

Priced in credits, per row, per query or per call depending on the source. Every call is quoted before it runs — `cost_estimate` returns the exact figure and spends nothing. See /pricing.

Latency

Moderate (~30-90s).

Prefer instead

`g2` or `capterra` for B2B software. `software-reviews` only when you need several platforms at once — it costs many times more per row.

Cannot answer

B2B software reviews with reviewer job role, or app store reviews (use `app-store-reviews`).

Related sources