Debug production issues without leaving your editor
Pull logs, traces, and bug reports into Cursor, find the root cause with AI, and fix the code in one continuous flow
Overview
A bug comes in on Slack. You open your logging tool, filter for the error, scroll through traces, copy what looks relevant. Then you switch to your editor to find the related code. Then back to logs to check another case. By the time you understand the pattern, you have lost 30 minutes to tab-switching. And you still have to reproduce it locally, fix it, and write up the MR.
How it works
MCP pulls the context you need directly into Cursor: the original bug report from Slack or Jira, related merge requests from GitLab. You iterate with the AI to understand the pattern, identify root cause, and figure out how to reproduce locally. Once you have fixed the issue, the AI drafts your merge request description with investigation notes, repro steps, and QA details. You review, push, and ship.
Who this is for
Engineers who debug production issues and are tired of switching between their editor, logging tools, and issue trackers. Works best in Cursor where you can investigate and fix in the same session.
Suggested prompt
Help me debug a production issue. Start by asking me for: (1) the error message or customer email so you can pull related logs, (2) where the bug was reported (Slack thread or Jira ticket), (3) which repo and branch I am working in, and (4) anything else you need to investigate. Once you have my answers, pull the logs and bug context, then walk me through what you are seeing. Help me identify the pattern and figure out how to reproduce it locally. When I have fixed the issue, draft a merge request description that includes what we found, how I reproduced it, and how to QA the fix. Show me the MR before pushing to GitLab.
Frequently asked questions
What observability and bug tracking tools does this work with?
This template connects to GitLab for merge requests and code changes, Jira for bug reports and issue tracking, and Slack for the original bug reports or discussions. You pull all the context into your editor without switching tabs.
Does it actually fix the bug for me?
It helps you investigate and understand the issue, but you write the fix. The AI walks you through the logs and traces, helps identify the pattern and root cause, and then drafts the MR description when you are done. The code changes are yours.
Can I use this with tools other than GitLab and Jira?
The template is set up for GitLab and Jira, but the approach works with any combination of code hosting and issue tracking that Zapier supports. You would just swap the integrations when setting up your server.
Does it draft the merge request for me?
Yes. Once you have fixed the issue, the AI drafts your MR description with investigation notes, reproduction steps, and QA details based on everything you discussed during the debugging session. You review it before pushing to GitLab.