Queue maintenance and bug fixes for Cursor background agents
A developer authorized the GitHub app so Cursor background agents cloned repos into sandboxes, branched, and worked through a day's queue of maintenance, features and bug fixes.
Done withCursor
- What they did
- Zachary Proser authorized Cursor's GitHub app for his organizations. Cursor then cloned the repos into sandboxed containers, where the agents ran commands and created branches. Each morning he queued 5-7 maintenance, bug-fix and small feature tasks, each aimed at a chosen repo and branch. He reviewed the changes, sent follow-ups in a per-task chat, or opened the work in the Cursor editor.
- How it went
- By the time his coffee was done, most tasks were finished or ready for review. He estimates one-shot success at 70-80%, against 40-60% for Codex. Creating a new branch that didn't exist yet errored, and multi-turn branch handling was clunky.
- Worth knowing
- Agents come with Cursor Pro at $20/month and ran on Claude Sonnet 4. He still did larger refactors and features directly in the editor.
Try it yourself with Cursor
Using the GitHub repo [owner/repo], start background agents for each item in this queue: [bug 1], [maintenance task], [small feature]. For each one, clone into a sandbox, create a separate branch, make the change, run the tests, and open a draft pull request describing what changed. Stop when every item has a draft PR, and do not merge any of them.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
Source: zackproser.com · Undated
Related use cases
Keep coding agents running autonomously for long stretches
Cursor's write-up on scaling long-running autonomous coding.
Draft personalized outbound, recaps and forecasts for sales
Cursor's internal ChatGTM drafts signal-based outbound messages with personalized landing pages, call recap emails and follow-ups, and assembles the deal forecast.
Debug prod pods and docker-compose with Cursor
A DevOps engineer uses Cursor with a Kubernetes MCP to pull prod logs and suggest fixes, and to repair a broken docker-compose file.
Build a C compiler with a team of parallel agents
Anthropic's engineering write-up on tasking Opus 4.6 agent teams with building a C compiler.
Build and deploy a Hacker News scraper app
Replit Agent generated a scraper, database models, API endpoints and front end that stores and shows Hacker News items, then deployed it.
Resolve a Slack-reported issue with a coding agent
A Codex agent resolved an issue passed to it in about 40 minutes, and an ops agent verified the fix back in Slack.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.