MCP Server Overview | Atatus Docs

The Atatus MCP server links your Atatus observability data directly to AI assistants that support the Model Context Protocol (MCP), including Claude, Claude Code, and Cursor. Simply ask a question in plain English, and your assistant will fetch the answer straight from your Atatus account. This means you can skip clicking through dashboards or learning complex query syntax.

Instead of manually navigating to the Errors tab, applying filters, and reading through stack traces, you can simply ask:

"Are there any unhealthy Kubernetes workloads right now?"

"Why is the checkout-service slow this week?"

"What did we deploy in the last 6 hours, and did it break anything?"

Your AI assistant will locate the relevant data, trace the issue from cause to effect, and deliver a clear answer in seconds.

Note: The MCP server is read-only by default. A Read-scoped API key allows you to query everything without making any changes. If you need to perform write actions, such as resolving or ignoring an error, you will need a Read & Write key. See the Installation guide for details.

What you can ask

The server equips your assistant with 67 tools spanning ten monitoring domains, all mapping directly to the telemetry data you already collect in Atatus.

Category	Tools	Ask things like
APM / Application	17	"How healthy is order-service?", "Which transactions are slowest?", "Did the deploy break anything?"
Browser Monitoring	5	"What are our Core Web Vitals?", "Which JavaScript errors are users hitting?", "Show me that error's stack trace."
Mobile Monitoring	7	"What's the crash-free rate?", "What's crashing the Android app?", "How is v2.3.1 doing?"
Kubernetes	11	"What's unhealthy in the cluster?", "Why is this pod failing?", "Any pods stuck pending?"
Infrastructure	8	"Which hosts are inactive?", "Top processes on app-prod-1 by CPU?", "Why did this container restart?"
Logs	2	"Show me error logs from host Zenitsu in the last 7 days.", "What are the top recurring log patterns?"
Distributed Tracing	4	"Find traces slower than 2s", "Where did this request spend its time?", "Show the service map."
Database Monitoring	4	"Which database is slowest?", "Which queries should I optimise?", "Does this query use an index?"
Synthetics and Uptime	5	"Are all uptime checks passing?", "What was our uptime in March?", "What went down, and for how long?"
Service Topology	4	"If payments-db goes down, what breaks?", "What does checkout depend on?"

For a complete breakdown of each tool, including its specific function and sample questions, see the Tool Reference.

Alerts, Incidents, and SLOs do not have MCP tools yet — see Not yet available.

Supported clients

The MCP server works with any client that supports remote MCP over HTTP. Documented, ready-to-paste setups:

Editors & IDEs

CLIs

Desktop apps

Claude Desktop

Other MCP clients connect in a similar way. You can point them to https://mcp.atatus.com/mcp over HTTP and include your key in the X-API-KEY header. Refer to the Installation guide for specific configuration details for each client.

Get started in three steps

Create an Atatus Application Key in Settings » Account Settings » API Keys.
Connect your AI client (Claude Code, Cursor, VS Code, and more).
Ask your first question.

Tip: Each API key reads only its own account's data, and a Read-scoped key can never change anything. Keep Read as your default.

Next steps

Installation: Get your Application Key and connect your AI client in under 5 minutes.
Tool Reference: Explore the full list of tools, their functions, sample queries, chaining options, time ranges, and response limits.
Troubleshooting & FAQ: Find solutions to common setup issues and answers to frequently asked questions.