Sentry

Sentry’s MCP server lets an agent pull real production error data directly into a debugging session — recent issues, full stack traces, event frequency, and affected releases — instead of relying on a developer to copy-paste a stack trace into the conversation manually.

The practical workflow it enables is closing the loop from alert to fix: an agent can look up an issue by ID or query recent errors for a project, read the actual stack trace and breadcrumbs, and go straight to proposing or making the code change, all inside the same session that started with “why is this failing in prod.”

Most useful paired with the Git or GitHub MCP servers in this catalog, so root-causing an error and opening the fixing commit or PR happens in one continuous agent workflow rather than two disconnected tools.