A THOUGHTFUL FIRST PASS

Less sorting.
More solving.

A little local AI for your issue queue. Find a starting label, spot possible duplicates, and keep the final say.

Open-model inferenceHuman-reviewed changes

01Bring a small queueStart with synthetic issue JSON

02Take a closer lookReview labels and similar issues

03Make a reversible changeApply locally, export, or undo

YOUR WORKSPACE

A calmer queue starts here.

Local session

Issues in this run—A manageable starting point

Ready to review—Clearer label suggestions

Needs a closer look—Ambiguous or low similarity

The review queue

Your judgment belongs in the loop.

Awaiting a run

A fresh perspective, one issue at a time.

Load the synthetic sample or add your own synthetic JSON. Run a first pass to see suggestions here.

0 issues selectedOnly your checked choices will be applied.

A change you can take back.

Selections update this local copy. Your GitHub repository is never touched.

01 / INPUT

Bring your issues

Try the demo queue, paste JSON, or open a local file. Use synthetic data for this private prototype.

ISSUE EXPORTJSON

1–100 issues, as an array or an “issues” object. Use a unique positive id (or number) and a title. Only id, title, body, and label names are retained.

Local after setup. Initial model setup may need a download.

A SUGGESTION, WITH CONTEXT

See the reason. Keep the choice.

Each label comes with a nearby reference example. Similarity is a comparison score, not a probability or a guarantee.

Possible duplicates are clues to investigate. Nothing is merged or closed automatically.

The honest benchmark

Checking for a saved evaluation…

Developer-made synthetic cases, not an independent holdout. These results do not establish performance on real repositories.

Less sorting.More solving.

A calmer queue starts here.

The review queue

A fresh perspective, one issue at a time.

Less sorting.
More solving.