Turn GitHub issues into draft PRs with a long-running agent

A developer used Dots to read GitHub issues and produce rough pull request drafts, mostly as a starting point for later work.

Done with DotsCodexClaude Code· Intermediate· Oct 2026

What they did
They used Dots together with Codex for tasks that involved a lot of reading and took a while to finish. The agent scanned GitHub issues and turned them into draft pull requests. No extra setup details are given, and the workflow description does not include a person stepping in during the run.
How it went
The pull requests were often not very strong, but they still provided an initial version to build on. They reported better practical results from a Claude Code-style workflow that kept a human involved. Using regular plan limits was a dealbreaker, and they also wanted tighter control when the agent started drifting from the intended context.
Tips
Treat this as a rough drafting tool, not a replacement for review. If you use it, pause before any branch is opened or any draft PR is submitted so you can check the plan first.

Try it yourself with Dots

Go through the open issues in [GitHub repo URL] and select the [number] clearest items tagged [label or topic]. For each one, inspect the relevant code, create a branch, and write a draft pull request that includes an initial fix attempt, tests if the repository has them, and a short note explaining what changed and what remains uncertain. Before you create any PRs, show me the full list of planned drafts and wait for my approval. Stop and ask before any action that would send, publish, merge, or delete anything.

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.