Sematext MCP Server is now generally available.

When things fail or error in production the root of the problem is typically not the service that initially manifests itself as having issues. Normally, finding the cause means jumping between traces, spans, logs, metrics, events, alerts, and code. With Sematext MCP Server, you now have an alternative that can make this a lot faster.

You can connect the Sematext MCP Server to AI agents like Claude Code, Cursor, Codex, Windsurf, Gemini, OpenCode, etc. Then, instead of manually looking at dashboards and gathering context from your observability data, you can tell your AI agents to pull the relevant signals together and ask questions like:

• Why did requests start failing around 11:04?
• Which service degraded after deployment this morning?
• What alerts fired in the last hour?
• Why is checkout latency increasing in Europe?
• What changed before the spike at 12:07?
• Create an incident summary and suggest probable root causes.

You can also use it for other things, such as:

• Finding how often different types of alerts are fired and finding the ones that are too noisy.
• Performance investigations while still developing a feature.
• Checking the overall resource usage of your fleet to plan provisioning.

The Sematext MCP Server currently exposes 13 tools and you can see usage examples in the docs. We plan on adding support for Network Monitoring, Service Map, API Monitoring, Root Cause Analysis, and others.

Here’s an example of our own usage of Sematext MCP Server from last week. The context is that one of our OpenSearch clusters started increasing in size. That cluster contains a number of indices. Instead of manually exploring the size of each index in the Sematext Cloud dashboards we simply asked Claude a pretty simple and imprecise question, even made some typos along the way, like this:

Claude used the Sematext MCP Server and quickly found a set of indexes that were responsible for the cluster growth we observed:

Reminder: regressions and production issues often arise after deployments. If you are not sending your deployment events to Sematext yet, see our Events docs to learn how to do that. It’s simple, completely free, and having this signal next to your logs, traces, metrics, etc. is an excellent way to make your troubleshooting much more effective, whether manually in the UI via dashboards or using the MCP Server with AI agents.

The Sematext MCP Server was designed to use the minimal number of tokens. We’ve spent weeks minimizing its token usage. We have also worked hard to ensure our MCP Server tools generate optimal responses so your AI agents can be both efficient and effective.

For more information, check out Sematext MCP Server docs to get started.

Start Free Trial

---