Your AI coding assistant already knows your codebase. Now it can know your production environment too.

AppSignal's MCP server gives AI agents and AI code editors direct access to your monitoring data — errors, performance metrics, and more — so they can help you debug, investigate and resolve issues without switching context.

And with our new public endpoint, getting started is simpler than ever.

What Changed: No Docker Required

Previously, connecting an AI agent to AppSignal meant running our MCP server as a Docker container. It worked, but it added setup overhead — especially for developers who just wanted to try it out.

Now, our MCP server is available as a hosted endpoint at https://appsignal.com/api/mcp. Point your agent at it, authenticate with a token, and you're done. No containers, no infrastructure, no fuss.

The Docker image is still available for teams that need a local MCP server, but for most setups the public endpoint is all you need.

Getting Started in Two Steps

1. Generate an MCP Token

To generate your token:

1. Click your profile icon in the top right
2. Go to Account Settings
3. From there, navigate to MCP Tokens to create a new token

Tokens have configurable permissions per toolset — by default you get write access to exceptions and dashboards, and read access to app discovery. You can lock down or expand those permissions per token.

2. Add It to Your Agent

Add the endpoint and token to your agent's MCP configuration. Here's an example for Claude:

{
  "mcpServers": {
    "appsignal": {
      "type": "streamable-http",
      "url": "https://appsignal.com/api/mcp",
      "headers": {
        "Authorization": "Bearer your-mcp-token"
      }
    }
  }
}

That's it. Your agent can now query AppSignal directly.

Where It Works

Our MCP server works with any MCP-compatible client — including standalone AI agents and AI code editors with built-in MCP support:

• Claude — AI agent, via Claude Desktop or Claude Code
• Cursor — AI code editor with built-in MCP support
• Windsurf — AI code editor with native MCP integration
• Zed — AI code editor with MCP support
• VS Code — via GitHub Copilot's MCP support

What Can Your Agent Do?

Once connected, your AI agent has access to 18 tools across five areas. That means it can:

The read/write split is worth noting: this isn't a read-only integration. Your agent can close incidents, leave notes, assign handlers, and build dashboards — the same actions you'd take in the AppSignal UI.

A Few Things You Can Ask

Here are some concrete prompts to get a feel for what's possible:

• "Show me all open exceptions in the background namespace from the last 24 hours"
• "Close all incidents related to ActionMailer — they were fixed in the last deploy"
• "What anomaly triggers are configured for response_time in production?"
• "What was the p95 response time for my API endpoints yesterday between 2pm and 4pm?"
• "Create a dashboard called 'API Health' and add a line chart of p95 response time by action"

Try It Out

AppSignal's MCP server is currently in beta, and we'd love your feedback.

1. Read the MCP documentation to get set up.
2. Try asking your agent to investigate a recent error or check your app's performance.
3. Let us know what you think — drop us a message on Discord or reach out to support@appsignal.com.