MCP Integration

Model Context Protocol (MCP) integration guide for connecting AI agents to Vouch reputation data and actions.

What is MCP?

The Model Context Protocol (MCP) is an open standard developed by Anthropic for connecting AI systems to external data sources and tools. Vouch implements MCP to allow AI agents—including Claude, custom LLM applications, and workflow automation tools—to:

  • Query review data, sentiment trends, and reputation metrics
  • Trigger review solicitation campaigns
  • Draft and submit review responses
  • Access location and contact information
  • Monitor compliance and audit trails

Architecture Overview

Vouch's MCP server exposes a set of tools and resources that AI agents can discover and invoke:

  • Tools: Executable functions like send_solicitation, draft_response, get_vouch_score
  • Resources: Read-only data endpoints for reviews, locations, and analytics
  • Prompts: Pre-configured prompt templates for common reputation management tasks

The MCP server runs as a separate service that authenticates requests using API keys scoped to specific tenants and permissions.

Getting Started

Step 1: Generate an API Key

From your Vouch dashboard, navigate to Settings → API Keys and create a new key with MCP access. Choose the appropriate scope (read-only, write, or admin).

Step 2: Configure Your MCP Client

Add the Vouch MCP server to your MCP client configuration. For Claude Desktop, edit your claude_desktop_config.json:

{
  "mcpServers": {
    "vouch": {
      "command": "npx",
      "args": ["-y", "@vouch/mcp-server"],
      "env": {
        "VOUCH_API_KEY": "your-api-key-here",
        "VOUCH_API_URL": "https://api.tryvouch.io"
      }
    }
  }
}

Step 3: Test the Connection

In your MCP client, verify that Vouch tools are available by listing available tools or invoking a simple read operation like get_locations.

Available Tools

The Vouch MCP server exposes the following tools:

get_locations

List all locations in the tenant with their review platform connections.

get_reviews

Query reviews by location, date range, rating, or platform. Supports filtering and pagination.

get_vouch_score

Retrieve the Vouch Score (composite reputation metric) for a location or tenant.

send_solicitation

Trigger a review solicitation to a contact via email, SMS, or WhatsApp. Requires write permission.

draft_response

Generate an AI-drafted response to a review. Returns draft text for human approval.

submit_response

Post an approved response to a review platform. Requires write permission.

get_recommendations

Retrieve AI-generated recommendations from the Recommendation Engine.

get_audit_trail

Query the audit log for compliance and troubleshooting. Requires admin permission.

Example Use Cases

Automated Reputation Reporting

An AI agent queries get_reviews and get_vouch_score weekly, generates a natural-language summary of reputation trends, and emails it to the management team.

Proactive Service Recovery

When a low-rated review is detected, an AI agent drafts a response using draft_response, flags it for manager approval, and creates a follow-up task in the CRM.

Multi-Location Benchmarking

An AI agent compares Vouch Scores across all locations, identifies underperformers, and suggests specific actions based on review themes and recommendation engine output.

Compliance Monitoring

An AI agent monitors the audit trail for consent violations, frequency cap breaches, or review gating patterns, and alerts the compliance team if anomalies are detected.

Security & Permissions

MCP API keys are scoped to specific permissions:

  • Read: Query reviews, locations, scores, and recommendations
  • Write: Send solicitations, draft and submit responses
  • Admin: Access audit trails and modify configuration

Keys can be further restricted to specific locations or date ranges. All MCP requests are logged in the audit trail.

Rate Limits & Quotas

MCP requests are subject to the following limits:

  • Read operations: 1,000 requests per hour per API key
  • Write operations: 100 requests per hour per API key
  • Solicitation sends: Subject to your plan's monthly send quota

Contact support@aartha.ai if you need higher limits for production AI agent deployments.

Additional Resources