Connect Trustpilot to Claude: Search Business Units & Analyze Ratings

If your team needs to connect Trustpilot to Claude to automate review analysis, extract SKU-level insights, or manage customer reputation workflows, you need a Model Context Protocol (MCP) server. This server acts as the translation layer between Claude's tool calls and Trustpilot's REST APIs. You can either build and maintain this infrastructure yourself, or use a managed integration platform like Truto to dynamically generate a secure, authenticated MCP server URL.

If your team uses ChatGPT, check out our guide on /connect-trustpilot-to-chatgpt-automate-review-management-replies/ or explore our broader architectural overview on /connect-trustpilot-to-ai-agents-sync-catalog-data-invitations/.

Giving a Large Language Model (LLM) read and write access to a specialized reputation management ecosystem like Trustpilot is an engineering challenge. You have to handle OAuth 2.0 or API key token lifecycles, map massive JSON schemas to MCP tool definitions, and deal with Trustpilot's distinct separation between public and private data models. Every time Trustpilot updates an endpoint or deprecates a legacy resource, you have to update your server code, redeploy, and test the integration.

This guide breaks down exactly how to use Truto to generate a secure, managed MCP server for Trustpilot, connect it natively to Claude Desktop, and execute complex workflows using natural language.

The Engineering Reality of the Trustpilot API

A custom MCP server is a self-hosted integration layer. While the open MCP standard provides a predictable way for models to discover tools, the reality of implementing it against specialized B2B APIs is painful. Trustpilot enforces strict domain logic around privacy, business unit hierarchy, and localized asset routing.

If you decide to build a custom Trustpilot MCP server, here are the specific integration challenges you will face:

Public vs. Private Endpoint Fragmentation Trustpilot splits its data access into public and private tiers based on strict privacy guidelines. If you query the public trustpilot_business_units_list_all_reviews endpoint, you will receive standard review text and star ratings. However, you will not receive the customer's email address or the internal reference order ID. To access that PII, the LLM must explicitly call a separate private endpoint (trustpilot_business_units_list_private_reviews), which requires higher OAuth scopes and returns a completely different JSON structure. A poorly designed MCP server will confuse the LLM by combining these tools without clear schema separation.

SKU-Level Batch Summaries and Pagination Boundaries Analyzing product reviews across a massive catalog requires querying specific SKUs. Trustpilot provides dedicated endpoints like trustpilot_product_reviews_batch_summaries to get aggregated data (star distribution, average rating) for multiple SKUs in one request. However, if an LLM tries to query thousands of imported reviews without batching, it will hit pagination boundaries. Trustpilot's paginated endpoints return an empty array if requested beyond the available range. Your MCP server must inject limit and next_cursor schemas dynamically, and explicitly instruct the model to pass cursor values back unchanged to prevent hallucinated pagination loops.

Strict Rate Limits and Stateless Execution Trustpilot strictly enforces API quotas. Truto does not retry, throttle, or apply backoff on rate limit errors. When the upstream Trustpilot API returns an HTTP 429 (Too Many Requests), Truto passes that error directly to the caller. Truto normalizes upstream rate limit information into standardized headers (ratelimit-limit, ratelimit-remaining, ratelimit-reset) per the IETF spec. The caller (or the agentic framework wrapping Claude) is fully responsible for implementing its own retry and backoff logic. Your MCP architecture must handle these raw errors gracefully without crashing the active agent session.

Generating a Trustpilot MCP Server

Rather than hand-coding JSON-RPC 2.0 protocol handlers and manually mapping Trustpilot's OpenAPI spec to MCP tools, Truto derives tools dynamically from the integration's resource definitions and documentation records.

Each MCP server is scoped to a single connected instance of a Trustpilot account. The generated server URL contains a cryptographic token that securely encodes the target account, allowed methods, and expiration configurations.

You can generate this server via the Truto UI or programmatically via the API.

Method 1: Via the Truto UI

For internal operations or manual agent deployments, generating the server via the UI takes seconds.

1. Navigate to the Integrated Accounts page in your Truto dashboard and select your connected Trustpilot account.
2. Click the MCP Servers tab.
3. Click Create MCP Server.
4. Select your desired configuration (e.g., restrict to read methods only, apply support tags, or set a 7-day expiration).
5. Copy the generated MCP server URL (e.g., https://api.truto.one/mcp/a1b2c3d4e5f6...).

Method 2: Via the Truto API

For production applications, you can generate MCP servers programmatically. This is useful for spinning up ephemeral AI agents per customer or per workflow.

Make an authenticated POST request to the Truto API:

curl -X POST https://api.truto.one/integrated-account/{integrated_account_id}/mcp \
  -H "Authorization: Bearer YOUR_TRUTO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Trustpilot Reputation Analysis MCP",
    "config": {
      "methods": ["read", "write"]
    },
    "expires_at": "2026-12-31T23:59:59Z"
  }'

The API evaluates the Trustpilot integration's available documentation, generates a hashed secure token, and returns the endpoint payload:

{
  "id": "mcp_srv_9x8y7z6",
  "name": "Trustpilot Reputation Analysis MCP",
  "config": { "methods": ["read", "write"] },
  "expires_at": "2026-12-31T23:59:59Z",
  "url": "https://api.truto.one/mcp/a1b2c3d4e5f67890"
}

Connecting the MCP Server to Claude

Once you have the Truto MCP URL, you can connect it directly to Claude. All communication happens over HTTP POST with JSON-RPC 2.0 messages.

Method A: Via Claude or ChatGPT UI

If you are using the consumer-facing AI interfaces that support remote MCP servers:

For Claude:

1. Open Claude and navigate to Settings.
2. Go to Integrations -> Add MCP Server.
3. Paste your Truto MCP URL and click Add.

For ChatGPT:

1. Open ChatGPT and navigate to Settings -> Apps -> Advanced settings.
2. Enable Developer mode.
3. Under Custom connectors, click add, enter a name (e.g., "Trustpilot by Truto"), and paste the Truto MCP URL.

Method B: Via Claude Desktop Configuration

If you are building custom agents or using Claude Desktop natively, you can route the HTTP MCP server through the standard @modelcontextprotocol/server-sse bridge. This allows Claude Desktop (which expects local stdio communication) to talk to Truto's remote SSE endpoint.

Edit your claude_desktop_config.json file:

{
  "mcpServers": {
    "trustpilot_truto": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-sse",
        "https://api.truto.one/mcp/a1b2c3d4e5f67890"
      ]
    }
  }
}

Restart Claude Desktop. The model will initialize the connection, perform a handshake, and dynamically list all available Trustpilot tools.

Hero Tools for Trustpilot

Truto automatically generates descriptive, snake_case tool names derived from Trustpilot's resources. Because query and body parameters share a flat input namespace during tool execution, Truto strictly separates the schema definitions under the hood.

Here are the highest-leverage hero tools for automating Trustpilot workflows:

search_trustpilot_business_units

Searches for a Trustpilot business unit by name or partial match. This is almost always the first tool Claude must call to acquire the internal business_unit_id required for subsequent operations.

"Find the Trustpilot business unit for 'Acme Corp' and return its internal ID, display name, and current trust score."

get_single_trustpilot_business_unit_by_id

Retrieves public metrics for a specific business unit, including total number of reviews, overall score, and website URL.

"Get the detailed profile and review statistics for business unit ID '50b4b3c...' to see how many total reviews they have accumulated."

trustpilot_business_units_list_private_reviews

Lists private reviews for a Trustpilot business unit. Unlike the public equivalent, this tool exposes sensitive metadata such as the consumer's email address and internal order ID (if verified). Highly restricted and typically used in authenticated support workflows.

"Fetch the latest 50 private reviews for our business unit. Identify any 1-star or 2-star reviews and output the associated customer email addresses and order IDs for the support team."

trustpilot_product_reviews_batch_summaries

Gets aggregated review summaries for multiple specific SKUs in one request. It returns the star average, distribution, and total review count per SKU.

"Analyze the product review summaries for SKUs 'TSHIRT-BLK-L' and 'TSHIRT-WHT-M'. Compare their average star ratings and total review volume."

create_a_trustpilot_review_reply

Posts a public reply to a Trustpilot service review on behalf of the business. Requires the review_id and the text of the reply.

"Draft a professional, empathetic response to review ID '65a12b...' acknowledging their shipping delay, and post the reply directly to Trustpilot."

trustpilot_invitations_create_link

Generates a unique Trustpilot product or service review invitation link that can be emailed or texted directly to a consumer.

"Generate a service review invitation link for business unit ID '50b4b3c...' that we can include in our post-purchase SMS sequence."

For the complete schema definitions and the full list of supported operations, view the Trustpilot integration page.

Workflows in Action

Once the MCP server is connected, Claude can orchestrate multi-step API calls autonomously. Here are two real-world scenarios.

Scenario 1: Automated Review Triage & Support Handoff

Customer Support teams spend hours daily matching negative Trustpilot reviews to internal CRM profiles. An AI agent can automate this discovery and draft initial responses.

"Check our business unit for any new private reviews posted in the last 24 hours. Filter for reviews with 3 stars or fewer. For each negative review, extract the order ID and customer email, then draft a polite public reply apologizing for the friction and asking them to check their email for a resolution."

How the agent executes this:

1. Calls search_trustpilot_business_units to verify the company's internal ID.
2. Calls trustpilot_business_units_list_private_reviews to retrieve the latest private reviews, including orderId and customer emails.
3. Evaluates the stars parameter locally in context to filter for <= 3.
4. Drafts contextual reply text based on the review's text.
5. Calls create_a_trustpilot_review_reply for each matching review_id to post the response.

sequenceDiagram
  autonumber
  participant Claude as "Claude AI"
  participant TrutoMCP as "Truto MCP Server"
  participant Trustpilot as "Trustpilot API"

  Claude->>TrutoMCP: Call tools/call (list_private_reviews)
  TrutoMCP->>Trustpilot: GET /v1/private/business-units/{id}/reviews
  Trustpilot-->>TrutoMCP: Return JSON (stars, orderId, email)
  TrutoMCP-->>Claude: Return MCP result block
  
  Note over Claude: Agent identifies 2-star<br>review and drafts reply.
  
  Claude->>TrutoMCP: Call tools/call (create_review_reply)
  TrutoMCP->>Trustpilot: POST /v1/private/reviews/{id}/reply
  Trustpilot-->>TrutoMCP: 204 No Content
  TrutoMCP-->>Claude: Return success status

Scenario 2: SKU Reputation Auditing

E-commerce Managers need to track product sentiment at the SKU level to identify manufacturing defects or sizing issues.

"Fetch the product review batch summaries for our new fall line (SKUs: FALL-JAC-01, FALL-B00T-02). Compare their star distributions. If any SKU has an average below 4.0, generate a unique review invitation link so we can solicit more feedback from recent buyers."

How the agent executes this:

1. Calls search_trustpilot_business_units to get the BU ID.
2. Calls trustpilot_product_reviews_batch_summaries passing the array of requested SKUs.
3. Analyzes the starsAverage and distribution arrays returned in the response.
4. Identifies that FALL-B00T-02 has an average of 3.8.
5. Calls trustpilot_invitations_create_link to generate a dedicated URL to distribute to recent buyers of that specific SKU.

Security and Access Control

Exposing a reputation management platform like Trustpilot to an LLM requires strict boundary control. Truto's MCP servers enforce security at the token level, meaning the client cannot bypass these restrictions.

• Method Filtering: You can restrict a Trustpilot MCP server to only allow read operations. If an agent hallucinates a request to call create_a_trustpilot_review_reply, the MCP router blocks it before it ever reaches the proxy layer.
• Tag Filtering: Limit the available tools to specific integration resource tags. For example, you can grant an agent access to reviews and business_units, but completely block access to invitations or deletions.
• API Token Authentication: By toggling require_api_token_auth: true, the MCP server requires the client to pass a valid Truto API token in the Authorization header. Possession of the MCP URL alone is no longer sufficient, preventing unauthorized execution if the URL leaks.
• Automatic Expiration: Set an expires_at timestamp when generating the server. Once the timestamp passes, Truto automatically schedules a Durable Object alarm to purge the token and all associated KV data, instantly killing the agent's access to Trustpilot.

Moving Beyond Manual Review Management

Connecting Trustpilot to Claude via a managed MCP server transforms reputation management from a manual, reactionary process into an automated, proactive system. By offloading the complexities of API versioning, token management, and localized data routing to Truto, your engineering team can focus on writing better agent prompts rather than debugging JSON-RPC payloads.

Stop writing custom integrations for LLMs. Generate a secure, production-ready MCP server for Trustpilot in seconds, and let your AI agents handle the rest.