BugLens

Lots of bug reports in open-source projects show up as a screenshot and a single line of text. Before anyone can fix the bug, a maintainer has to rewrite the report in the project’s issue format and figure out which file is responsible. That’s often where new contributors get stuck.

BugLens takes a public GitHub repo, a screenshot of the bug, and one sentence describing it. From those it gives you four things:

An issue draft that follows the repo’s own issue template and only uses labels the repo actually has. If something wasn’t stated, it’s marked [please confirm] rather than guessed.
The most likely culprit files, each with a reason, a confidence score, and the code snippet that matched, linked to the exact lines.
A plain-language brief for someone making their first contribution.
Similar existing issues, so you can spot a duplicate before you file.

Gemma 4 looks at the screenshot, ranks the candidate files, and writes the draft. Code search runs locally with an open-source embedding model and blends three signals: meaning, exact text visible on screen, and file names from stack traces. The model can only pick from files the search actually found, so it can’t invent a path.

We tried it on a real, already-fixed bug in File Browser. Out of 353 files, BugLens ranked first the one file the maintainers changed in their fix.

BugLens points you to likely files. It doesn’t fix the bug, and everything it produces is a suggestion, so check it before you file.