Sentry MCP Server

What it does

The Sentry MCP Server connects AI coding agents to your Sentry organization. Agents can fetch issues, error details, project configuration, and Seer analysis directly from the IDE. Sentry hosts the server remotely with OAuth, so no local installation is required. During setup you can choose which tool groups to enable to keep the context window focused.

Available tools

Tool	What it does
get_issue	Get full details for a Sentry issue
list_issues	List and filter issues by project, status, or query
get_project	Get project configuration and metadata
list_projects	List projects in the organization
get_seer_analysis	Retrieve Seer AI analysis for an issue
search_events	Search event data and stack traces

Tool availability depends on which groups you enable during setup. Self-hosted installations may expose different tools via stdio.

Setup by tool

Cursor

Add via one-click installation or manually. For the hosted server:

{
  "mcpServers": {
    "sentry": {
      "url": "https://mcp.sentry.dev/mcp"
    }
  }
}

Or use the add-mcp command:

npx add-mcp https://mcp.sentry.dev/mcp

On first connection, you will be prompted to authenticate via OAuth in the browser.

Claude Code

Add the same configuration to your Claude MCP config. The OAuth flow will open in a browser when you first use the server.

VS Code / GitHub Copilot

Add the url to your MCP client configuration. Authentication is handled through the standard OAuth flow.

Self-hosted Sentry

For self-hosted Sentry, use the stdio transport with @sentry/mcp-server and a User Auth Token. See Sentry docs for the exact command and env vars.

When to use this

• Debugging: Pull error details, stack traces, and repro steps into coding sessions
• Issue triage: List and filter issues while implementing fixes
• Seer context: Use Seer analysis to understand root cause and suggested fixes
• Post-deploy verification: Check that new deployments did not introduce new issues

Security considerations

• OAuth scopes control what the server can access; review during setup
• Choose only the tool groups you need to limit exposure
• For self-hosted stdio, use a User Auth Token with minimal scopes
• Sentry data may contain PII; ensure your AI tool's data handling meets your policy
• Manually approve tool calls before execution when your client supports it