---
title: "MCP and AI agents"
description: "Connect an agent to CrawlConsole, choose scopes, discover tools, and query owned property telemetry safely."
section: "Agents"
version: "latest"
locale: "en"
updated: "2026-07-18T20:52:38.000Z"
---

# MCP and AI agents

Connect an agent to CrawlConsole, choose scopes, discover tools, and query owned property telemetry safely.

CrawlConsole exposes a remote Model Context Protocol server at `https://crawlconsole.com/mcp`. Agent API keys control access to backlink intelligence, owned properties, telemetry, and WebMCP analytics.

## Create an Agent API key

Open **Settings** and **API Keys**, then create a key for the agent or integration. Give each client its own key so access can be reviewed and revoked independently.

Copy the raw secret when it is created. CrawlConsole stores only the protected key record and cannot show the raw secret again.

## Configure a client

Use a remote HTTP MCP configuration with a bearer token.

```json
{
  "mcpServers": {
    "crawlconsole": {
      "url": "https://crawlconsole.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_AGENT_API_KEY"
      }
    }
  }
}
```

Client configuration formats vary. Use the client's secret storage when it provides one.

## Discover available tools

Call `dataset_catalog` to list the tools and active limits available to the key. Common workflows include:

- `domain_authority` for Domain Score and graph authority;
- `referring_domains` for domain-level backlink sources;
- `competitor_link_gap` for domains that link to competitors but not the target;
- `list_properties` for canonical owned property ids;
- `get_crawler_analytics` for verified crawler page demand;
- `get_webmcp_analytics` for WebMCP registrations and calls;
- `search_docs` for grounded CrawlConsole product instructions.

## Work with property tools

Property tools require the canonical id returned by `list_properties`. Do not pass a domain name where a `propertyId` is required.

```text
Call list_properties. Find the property whose domain is example.com, then call get_property_telemetry_status and get_crawler_analytics for the last 30 days.
```

## Protect keys and scopes

- Create one key per agent or integration.
- Grant only the scopes the client needs.
- Keep keys out of browser code and public repositories.
- Revoke keys that are no longer in use.
- Review Agent usage for unexpected tools or request volume.

OAuth discovery is also available from CrawlConsole's well-known authorization metadata for clients that support the remote authorization flow.