Sentry MCP Server

by Sentry

The Sentry MCP Server gives your AI coding assistant direct access to your Sentry error monitoring data. Instead of switching between your editor and the Sentry dashboard to look up stack traces, you can ask Claude, Cursor, or VS Code Copilot to pull issue details, search for errors, and even trigger root cause analysis -- all without leaving your workflow. The server exposes 16 tools organized into groups you select during setup. At the core, you get issue retrieval and search across all your projects, detailed error investigation with full stack traces, and integration with Sentry's Seer AI agent for automated root cause analysis and fix suggestions. Beyond debugging, you can manage projects, create DSNs, check release health, and query performance data. The recommended setup uses the remote OAuth server at mcp.sentry.dev/mcp. You authenticate with your existing Sentry org credentials and you are done -- no API tokens to manage, no packages to install locally. If you run a self-hosted Sentry instance or prefer local execution, install @sentry/mcp-server via npm and configure a SENTRY_AUTH_TOKEN. Transport-wise, the server supports Streamable HTTP with automatic fallback to Server-Sent Events (SSE), which means it works reliably across different MCP client implementations. OAuth scoping lets you control exactly which tool groups are available to your LLM, keeping context windows focused on what matters for your current task. One practical pattern: chain multiple tool calls to gather context before triggering Seer analysis. Ask your assistant to find recent unresolved issues in a project, pick the highest-impact one, pull the full event data, then run Seer to get a fix proposal. The MCP server handles the Sentry API orchestration while your LLM handles the reasoning. MCP is still a developing protocol, so expect occasional rough edges. But for teams already on Sentry, this is probably the fastest way to get error context into your AI-assisted development workflow.

FAQ