MCP

The DockRay integration with the Model Context Protocol (MCP) lets you securely connect AI tools and assistants to your application monitoring data. This allows MCP-compatible clients to search projects and errors, retrieve diagnostic information, and perform permitted operations directly in DockRay.

Application Monitoring Available to AI

The MCP server provides AI tools with controlled access to DockRay data. An assistant can use information about detected issues when analyzing applications, diagnosing errors, or working on solutions.

Depending on the permissions granted, an MCP client can:

• search projects and errors,
• retrieve details about detected issues,
• access diagnostic context for errors,
• change error statuses,
• perform other available operations within the scope of the key.

Controlled Access to DockRay

Each client or automation can be assigned a separate MCP key. The administrator defines its permissions, ensuring that the AI tool can access only the operations it actually needs.

Keys are assigned to the account and can be revoked independently without affecting other integrations.

Secure Operations Performed by AI

Every request is authenticated before an operation is performed. DockRay verifies the account, key, and required permission before providing access to data or allowing its state to be changed.

Operations performed through MCP are also recorded in the audit log, allowing you to review actions taken by connected tools.

Model Context Protocol Standard

The integration uses the Model Context Protocol standard, allowing DockRay to connect to compatible AI clients and tools without requiring a separate integration for each one.

You can configure access through the Model Context Protocol in the DockRay dashboard under Settings → MCP.

1. Create a new MCP key and give it a name that identifies the client or automation that will use it.
2. Select only the permissions required by that tool.
3. Copy the generated token. For security reasons, it is displayed only when the key is created.
4. Set the MCP endpoint in your client to https://dockray.io/api/mcp.
5. For authentication, use the Authorization: Bearer {token} header.

It is recommended to create separate keys for different clients and automations. This allows you to manage their permissions independently and revoke access for a single tool when necessary.