Hey Product Hunt! I’m Oleg, the creator of Lintic.

If your engineers don’t write code without AI, why are you still interviewing like they do? LeetCode feels like a ghost of the past. Modern engineering is about direction, decomposition, and iteration. I built Lintic to assess how people actually work in 2026.

Features I’m excited about:

The Constraint System: This is the core signal. You can limit tokens and interactions to see whether a candidate uses AI strategically or just brute-forces prompts.

Simulated Infrastructure: We already support a mock Postgres service that runs in the browser, with realistic behavior designed to test judgment under pressure. The broader vision is to simulate more real-world infrastructure over time.

WebContainers Runtime: I’m excited that we can provide a full Node.js environment in the browser. That means zero server-side compute costs for you.

Agent Interface Protocol: Lintic is agent-agnostic. You can plug in OpenAI, Anthropic, or your own custom models to see how candidates handle different AI personalities.

Adversarial Profiles: You can toggle settings that stress-test code with slow queries or queue backpressure to see how a candidate recovers from errors.

AI-Native Review Process: Reviewing is AI-forward too. For each session, you can inspect what happened and chat with the agent about the candidate’s approach, tradeoffs, mistakes, and recovery. It gives you a much richer signal than just reading the final code.

Lintic is entirely self-hostable via a single Docker image, so you keep full control over your data and your API budget.

I’d love to hear your thoughts.

What signals would you use to evaluate someone in an AI-native coding interview?

Looking forward to your feedback!