Hand narrow code changes to Devin for pull requests
A team used Devin for a month on small fixes like chart formatting and API edge cases, with tests and PRs; a complex full-stack task went poorly.
Done withDevin
- What they did
- The RunLLM team connected Devin to their GitHub repo. It scanned the code and tried the setup steps. They typed short task descriptions, such as a pie chart format fix, an API edge case, and a default update schedule. Devin planned, edited, ran linters and tests, and opened a PR. The team spent about five minutes testing each result, and PR comments could steer further changes.
- How it went
- Small single-component tasks took a minute or two to write up and five minutes to test, against roughly an hour by hand. A full-stack task with a new table, API call and button failed, and the team rewrote it themselves.
- Worth knowing
- Small tasks cost 1-5 ACUs ($2-10) at $500 for 250 ACUs. The three larger tasks averaged about 20 ACUs, and two of them gave nothing usable.
Try it yourself with Devin
In the repo [repo URL], fix [small, well-defined issue, e.g. chart axis formatting or an API edge case with null input]. Add or update tests, make sure they pass, and open a pull request with a clear description of the change. Keep the scope narrow and don't refactor unrelated code; stop and ask me if the task looks bigger than described.
Source: frontierai.substack.com · Undated
Five of these in your inbox every morning
The best things people got an AI agent to do, each with the prompt to try it.
More like this
Add a third-party API integration via Devin PR
Devin opened a PR with the endpoint and backend wiring that was merged, but missed API key logic, code style and abstraction, needing review fixes.
Try Devin on research tasks (mostly failed)
Answer.AI had Devin fetch an author's recent papers and research transcript summarization; it got stuck parsing HTML and returned generic results, rated a failure.
Build a Pong game and website from scratch
A tester asked Devin to create a basic Pong-style game and a website, and it finished both in under 20 minutes.
- Claude
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.
- Custom agent
Audit a codebase for redundant dependencies and polyfills
An MCP server lets an agent check a codebase for redundant dependencies and polyfills, usable remotely or as a local stdio server.
Build a budgeting app from a single prompt
A low-coder used Replit Agent to build a 'Budget Buddy' with category budgets, a spending dashboard and expense log in about 20 minutes, using half their monthly credits.