# Sentry MCP

> Sentry's hosted MCP service for debugging — pull issues, events, traces, and releases into your agent, and trigger Seer root-cause analysis.

Sentry MCP connects agents to production reality: 38 tools for searching issues and events, pulling stack traces and trace details, updating and annotating issues, and invoking Seer root-cause analysis. For Claude Code, Sentry ships it as a plugin that installs a sentry-mcp subagent — errors get debugged against the actual telemetry instead of your memory of it.

Website: https://mcp.sentry.dev

Sentry MCP closes the loop between "an error is happening in production" and "the agent is fixing it." Instead of pasting a stack trace into chat, the agent queries Sentry itself — full issue context, event history, trace data — and can even hand the gnarly ones to **Seer**, Sentry's root-cause analysis, before writing the fix.

## Highlights

- **38 debugging-focused tools** — `get_issue_details`, `search_issues`, `search_events`, `get_trace_details`, `update_issue`, project/release/team lookups, and docs search; curated for coding agents, not a 1:1 API mirror.
- **Seer on demand** — `analyze_issue_with_seer` runs Sentry's AI root-cause analysis and returns the findings as agent context.
- **A real Claude Code plugin** — the install ships a `sentry-mcp` subagent that Claude Code auto-delegates Sentry investigations to, keeping noisy telemetry out of your main context.
- **Hosted or self-hosted** — OAuth remote at `mcp.sentry.dev/mcp` for SaaS; an npx stdio server with a user token for self-hosted Sentry.

## In an AI-assisted workflow

```bash
claude plugin marketplace add getsentry/sentry-mcp
claude plugin install sentry-mcp@sentry-mcp
# then:
# > Investigate the spike in SENTRY-1207, find the root cause, and propose a fix
```

The agent pulls the issue, reads representative events and the trace, correlates with the release, and comes back with a diagnosis grounded in telemetry — the workflow the [debugger agent](/agents/quality-security/debugger) runs, now with production data attached.

> [!NOTE]
> The subagent delegation pattern here is worth copying for any noisy data source: raw events stay in the subagent's context; only the diagnosis returns to yours. It's the same context discipline covered in [Managing Claude Code Memory & Context](/guides/configuration/claude-code-memory-context).

## Good to know

The hosted service is free with a Sentry account (Sentry itself is freemium). The code is source-available under FSL-1.1-Apache-2.0 — readable and self-hostable, not OSI open source. Self-hosted stdio mode needs token scopes (`org:read`, `project:read/write`, `team:read/write`, `event:write`), and its AI-search tools require your own LLM provider key.

---

_Source: https://agentscamp.com/tools/sentry-mcp — Tool on AgentsCamp._