MCP/Sentry Toolkit & MCP
Sentry Toolkit & MCP logo

Sentry Toolkit & MCP

Verified Blue (500+ GitHub stars)

Agentic tooling for Sentry: investigate production errors, search stack traces, diagnose performance issues, and triage alerts directly from your coding agent.

eval-safety•Observability & Error Tracking
GitHub944
License:Apache-2.0
VERIFIED
BySentry@Sentry

About this Server

The official Sentry Toolkit and Model Context Protocol (MCP) server connects AI coding agents (Claude Code, Cursor, Copilot, VS Code) directly to Sentry error and performance telemetry. It enables assistants to search production issues, retrieve stack traces with source context, analyze user session replays, query performance spans, and automate issue triage. Supports both hosted remote connections (mcp.sentry.dev) and local stdio execution via @sentry/mcp-server with optional Seer AI root-cause analysis.

Real-time Interaction

Direct low-latency RPC protocol communication with structured JSON schemas.

Scoped Sandboxing

Restricted capability bounds to prevent accidental modifications or leakage.

Security & Best Practices

User Token Scoping: Use restricted personal or organization tokens with minimal read-only permissions for inspection workflows.
Data Scrubbing Compliance: Inherits Sentry organization data scrubbing rules, PII masking, and sensitive header stripping automatically.
Self-Hosted Support: Supports self-hosted Sentry instances via --host=sentry.example.com and optional --insecure-http for private VPCs.

When to Use (and When Not To)

Best For
  • •Investigating production bugs, unhandled exceptions, and regression errors reported in Sentry
  • •Fetching full stack traces and localized source lines directly into an IDE coding assistant
  • •Diagnosing slow transaction spans, database query bottlenecks, and API latency spikes
  • •Automating bug triage, issue assignment, and error resolution during sprint cycles
Not For / Limitations
  • •Local unit and integration testing without remote Sentry instrumentation
  • •General open-web documentation searches (use Context7 or Firecrawl instead)
  • •Continuous real-time metrics streaming outside Sentry telemetry

Install & Setup

terminal command
$claude mcp add sentry -- npx -y @sentry/mcp-server@latest --access-token=$SENTRY_AUTH_TOKEN

Exposed Tools & Endpoints

7 tools

Functions your AI agent can invoke via this Model Context Protocol server

search_errors

Searches recent Sentry issues and errors matching query tags, release versions, environments, and timeframes.

get_issue_details

Retrieves comprehensive details for a specific Sentry issue including latest event payload and error classification.

get_stacktrace

Fetches source-mapped stack traces, frame arguments, and surrounding source code context for an event.

search_events

Searches raw event telemetry across transactions, error occurrences, and performance spans.

triage_issue

Updates issue status (resolved, unresolved, ignored) or assigns the issue to a team member.

get_session_replay

Retrieves user session replay recordings, DOM events, and console logs leading up to an exception.

seer_analyze_issue

Invokes Sentry Seer AI to perform automated root-cause analysis and generate actionable code fix suggestions.