Sentry MCP — Error Monitoring Server for AI Agents

What it is

Sentry MCP is a Model Context Protocol server that connects AI agents to Sentry's error monitoring platform. It gives agents like Claude Code the ability to query issues, analyze stack traces, track regressions, and access crash context directly from the development environment. Instead of switching between your editor and the Sentry dashboard, the AI agent retrieves error data and suggests fixes inline.

The server targets developers who use Sentry for error tracking and want their AI coding assistant to have awareness of production errors while writing code.

How it saves time or tokens

Sentry MCP eliminates the context switch between your editor and the Sentry dashboard. When debugging, the AI agent can pull the exact stack trace, affected users count, and error frequency without you navigating the Sentry UI. The agent correlates error data with your codebase to suggest targeted fixes, reducing the investigation cycle from multiple tool switches to a single conversation.

How to use

1. Add the Sentry MCP server to your .mcp.json:

{
  "mcpServers": {
    "sentry": {
      "command": "npx",
      "args": ["-y", "@sentry/mcp-server"],
      "env": {
        "SENTRY_AUTH_TOKEN": "sntrys_your_token",
        "SENTRY_ORG": "your-org"
      }
    }
  }
}

1. Get your auth token at sentry.io/settings/auth-tokens.

1. Restart Claude Code and ask about your errors: 'What are the top unresolved issues in the backend project?'

Example

Debugging workflow with Sentry MCP:

# Ask the agent to investigate errors
'Show me the top 5 unresolved errors in the API project'

# Agent calls Sentry MCP, returns:
# 1. TypeError: Cannot read property 'id' of null (342 events)
# 2. ConnectionTimeoutError: Redis connection timed out (128 events)
# ...

# Drill into a specific error
'Show me the stack trace for the TypeError issue'

# Agent retrieves full stack trace and suggests a fix
'The null check is missing in UserService.getProfile(). Add:
if (!user) return null;'

Related on TokRepo

• AI Tools for Monitoring — Monitoring and observability tools for production systems
• AI Tools for Testing — Testing tools that catch errors before they reach production

Common pitfalls

• The auth token needs project-level read access. Organization-level tokens without project scope return empty results.
• Rate limits apply to the Sentry API. Frequent queries during a debugging session may hit throttling. Cache results when iterating on a fix.
• The MCP server exposes error data to the AI agent. Ensure your Sentry data does not contain sensitive user information that should not be sent to LLM providers.