Summary

The stacktrace arrives with the question.

The usual loop is read the alert, open Sentry, copy the trace, paste it somewhere. Here the URL is the input and the trace is already in context, which covers most of the distance between an alert and a diagnosis.

What it is

A Python MCP server for Sentry.io. It retrieves one issue by id or by pasted URL — title, status, level, first-seen and last-seen timestamps, event count and the full stacktrace — and lists a project's issues.

What you get

One issue retrieved by id or by a Sentry URL you paste
Title, status, level, first-seen and last-seen timestamps, and event count
The full stacktrace for that issue
A project's issues listed, given the project and organisation slugs
A sentry-issue prompt that drops the issue details into the conversation as context

Requirements

A Sentry auth token, with the project and organisation slugs, passed on the launch command. Runs through uvx, pip or Docker.

Setup effort

One command plus a key — pip install mcp-sentry, then supply credentials