Sentry MCP
An official MCP server from Sentry that connects AI coding assistants to Sentry error tracking, enabling natural language queries over issues, events, traces, and performance data for AI-assisted debugging workflows.
Score Breakdown
⚙ Agent Friendliness
🔒 Security
Auth token with project/org scopes. Internal integrations for server-side. SOC2 Type II, ISO27001. Error data may contain sensitive stack traces, user data. Data scrubbing available for PII in events.
⚡ Reliability
Best When
Your team uses Sentry and wants AI assistants to have direct, read-capable access to error data during debugging sessions, reducing context-switching between IDE and Sentry dashboard.
Avoid When
You use a different error tracking platform (Datadog, Rollbar, Bugsnag) or need a generic observability MCP that spans multiple backends.
Use Cases
- • Investigate production errors and stack traces directly within Claude Code or Cursor without switching to the Sentry UI
- • Use natural language to search Sentry issues converted automatically to Sentry query syntax
- • Automate error triage by giving AI agents access to issue history, event counts, and performance traces
- • Integrate Sentry error context into AI-driven code review and debugging sessions
- • Deploy as a remote HTTP MCP server for team-wide AI-to-Sentry integration with OAuth authentication
Not For
- • Teams not using Sentry for error monitoring — no cross-platform observability support
- • Scenarios requiring write access beyond creating issues/comments (e.g., modifying alert rules)
- • Self-hosted Sentry instances that have not enabled the Seer AI feature if AI search is needed
Interface
Authentication
Requires Sentry User Auth Token with specific scopes: org:read, project:read, project:write, team:read, team:write, event:write. AI search tools additionally require an OpenAI or Anthropic API key. OAuth app credentials needed for remote deployment.
Pricing
Open source MCP server; requires active Sentry subscription. AI-powered search requires separate LLM API key (OpenAI or Anthropic) with associated costs.
Agent Metadata
Known Gotchas
- ⚠ AI-powered search requires a second LLM API key (OpenAI or Anthropic) on top of Sentry credentials — easy to misconfigure
- ⚠ MCP_DISABLE_SKILLS env var needed to disable Seer on self-hosted Sentry without Seer enabled
- ⚠ OAuth setup for remote deployment requires additional app registration steps
- ⚠ Subagent delegation routing is automatic and may not be transparent to the calling agent
- ⚠ Token scope requirements are narrow and specific — missing any scope silently breaks specific tools
Alternatives
Full Evaluation Report
Detailed scoring breakdown, competitive positioning, security analysis, and improvement recommendations for Sentry MCP.
Scores are editorial opinions as of 2026-03-06.