Show HN: Verification of Human Understanding of LLM-Generated Work

[jbethune]

Built this back in March because of fears my vibe coding was getting away from me. Basically it just analyzes PRs and asks a question to the developer about them to verify understanding. Turned into a larger project with my consulting firm and clients over the last month. Going to leave the original basic implementation open source.

[missmoss]

Curious if there's a demo/screenshot on what questions developers should expect and what acceptable answers look like. Checked out the repo but didn't find the showcase.

[jbethune]

The SaaS version has a short video demonstrating the UX: https://ninchi.ai

[missmoss]

Cool! I watched the demo. The concept makes sense. Quick thought: free-text answers might be high friction for adoption. Not sure if you'd consider various verification method, but something like multiple choice questions or "highlight the relevant code chunk" could be lower effort for the developer, and easier to evaluate programmatically too.