Original community demo · Video hosted on X
Ranking & evaluation
A PR-review experiment evaluates a diff against typed checks and maps the results to a review outcome, escalating ambiguous cases.
Opens the author's original post in a new tab.
The implementation ideas below are our interpretation of the community demo. Check the original post for the author’s actual setup.
Help maintainers prioritize a review queue and identify changes that deserve a deeper security or correctness check.
Provide the diff and relevant review criteria, collect bounded judgments, then map them to a review queue. Retain human review and repository checks before merge.
These examples explain a possible workflow; they are not recorded demo results or API request and response formats.
A diff alone may omit important surrounding behavior. The model’s assessment is a triage signal, not evidence that a change is secure or correct.
Define the context and a small set of choices in the playground, then use the API guide to connect your application.
View demoA browser-extension experiment identifies advertising elements in the DOM and hides the elements it flags.
View demoA related Qwen-based experiment predicts categorical outputs for multiple JSON keys together on an Apple Silicon laptop.
View demoA PostgreSQL extension applies natural-language criteria to rows, exploring semantic filtering inside the database.
