Sentry MCP 🔌 MCP server Freemium

What it does

Sentry MCP is Sentry's official server for connecting coding agents to your error and performance data. It is built for developers working alongside an agent, not as a general interface to every part of Sentry. A typical use: you paste an issue link or ask about the latest crash, and the agent pulls the stack trace, breadcrumbs and related events, then fixes the code with real context. Sentry hosts it at mcp.sentry.dev and you sign in with OAuth. A stdio version is also available for self-hosted Sentry. The source is published under the Functional Source License (FSL-1.1), which converts to Apache 2.0 later.

Tools it gives your agent

Tools are grouped into "skills" that you choose when you connect:

• Inspect (on by default, read-only): search issues and events, and get issue details, stack traces, breadcrumbs, traces, spans, replays, releases, profiles, dashboards, alert rules and monitors
• Seer (on by default): runs Sentry's AI debugger on an issue for root-cause analysis
• Triage (opt-in): resolve, assign and update issues, and add notes
• Project management (opt-in): create and change projects, teams, DSNs and uptime monitors
• Docs search and whoami

Works with

Any MCP client that supports remote HTTP servers, including Claude Code, Cursor, VS Code, Codex and Gemini CLI. Sentry also offers a Claude Code plugin that adds a sentry-mcp subagent. The stdio package @sentry/mcp-server runs locally with a user auth token.

How to connect

Scoping the server to one project is recommended:

claude mcp add --transport http sentry https://mcp.sentry.dev/mcp/{organizationSlug}/{projectSlug}

Or as a Claude Code plugin:

claude plugin marketplace add getsentry/sentry-mcp
claude plugin install sentry-mcp@sentry-mcp

Cost and limits

The server itself costs nothing, but you need a Sentry account, and your Sentry plan decides which data and features are available. In stdio mode, the natural-language search tools only work if you supply an LLM provider key: OpenAI, Azure OpenAI, Anthropic or OpenRouter.

Safety notes

Only the read-only Inspect and Seer skills are on by default. Turn on Triage or Project management only when you want the agent to change things. Scoping the URL to an organization or project keeps the agent to that data. For stdio mode, create a token with just the scopes the README lists.