Finds failing tests in a repo, splits each failure into independent tickets, stores them, hands them out one at a time.
That's it. No fixing. No PRs. No auto-anything past detection. This is the "notice the bug" half of a bigger pipeline — not the "fix the bug" half.
Bugs in OSS sit unfixed because someone has to notice them first, then write an issue, then wait for a maintainer to have time. That chain has a weak first link — noticing.
IssueLoop skips it. It reads the test suite directly. A failing test is the bug report. No human has to spot it or write it up.
- Runs a repo's test commands
- Catches what fails
- Asks one question per failure: is this one bug or several bundled together? Splits accordingly
- Stores each as a ticket
- Hands tickets out one at a time — claimed tickets don't get handed out twice
- Doesn't propose a fix
- Doesn't touch a file
- Doesn't open a PR
- Doesn't run anything not already in your test manifest
Fixing is a separate system's job. This repo's only output is a clean, prioritized ticket queue.