Send migration and fix tickets to a coding agent
Reviewers handed Devin a steady stream of tickets across two production projects and let it draft code, test changes, and open pull requests.
- What they did
- For three weeks, reviewers assigned Devin work from Slack, Linear, and Jira on two live systems: a Next.js monorepo and a Python data pipeline. The queue included migrations, data engineering items, bug fixes, and refactors. Devin sorted out the task, made the edits in its cloud workspace, checked the code, and created pull requests. Reviewers were able to steer its plan and leave PR feedback for it to act on.
- How it went
- It performed well on repetitive tasks with clear rules, especially migrations and framework upgrades. It often handled review feedback correctly on the first pass, but vague instructions produced middling or bad results, and it could spend roughly 20 minutes pursuing the wrong solution before anyone noticed. Cost also varied a lot: two migration jobs used 5 and 28 ACUs, and Core charged $2.25 per ACU.
- Tips
- Give the agent specific acceptance criteria and break work into small, self-contained tickets. Expect to review early, because unclear requests and hidden detours can waste time and money.
Try it yourself with Devin
You are working in [repo URL]. I will give you [3-5 tickets] for [framework upgrades, migrations, bug fixes, or small refactors]. Handle them one at a time, keep each change on its own branch, run the appropriate checks, and open one pull request per ticket. Before doing anything that would commit, send, delete, or submit work, stop and ask me to confirm. After each pull request, write a short summary of the change, any risks, and what I should inspect, then wait for my approval before starting the next ticket.
Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.
Get 3 agent uses like this every morning
With the day's top AI news. Free, and one click to unsubscribe.
Related agent uses
Use Devin to build and ship a small app
They used Devin to turn an empty repo into a deployed City Explorer app in about 90 minutes.
Use Devin to draft pull requests on a codebase
A team tested Devin on their codebase for PR creation, but slow output and full review needs made it feel like a junior developer.
Use Devin to build, debug, and ship app setup work
Devin handled app scaffolding, repo debugging, and deployment setup in demos and early user reports.
Use Devin for one integration, one migration, and a review
Answer.AI tested Devin on three coding tasks, but the migration stalled and the security pass mixed valid findings with mistakes.
Start in Devin CLI, then hand off to cloud
Use Devin CLI to build context locally, then pass the same session to Devin Cloud so it can keep working without a reset.
Use Devin to push backlog work into pull requests
Ryan Carson used Devin across several tools to turn backlog items into pull requests faster.