Launching today

Nyx

Launching today

Your own team of AI security engineers, on demand.

68 followers

Your own team of AI security engineers, on demand.

68 followers

AI agents fail in ways traditional software doesn't. Nyx is your own team of AI security engineers, on demand, that helps you find all the ways in which your AI agents fail. Point it at any AI agent, or multi-agent system, and it launches 1,000+ attack strategies that adapt to your system in real time - pure blackbox, no integration needed. Built by ex-Meta and Monzo engineers.

Payment Required

Launch tags:Developer Tools•Artificial Intelligence•Security

Launch Team / Built With

EOL Dataset by HeroDevs — Find every EOL dependency in your stack. Free. In 5 minutes.

Find every EOL dependency in your stack. Free. In 5 minutes.

Promoted

Maker

📌

Hey Product Hunt 👋 We built this because we saw that Al agents present different failure modes from traditional software. Agents are non-deterministic, the same input can pass one run and fail the next. Vulnerabilities surface in those gaps, and attackers exploit these edge cases. So fixing one doesn't mean you're protected from the next. The old alternatives don't scale either. Manual adversarial testing is six figures per engagement and it's a frozen snapshot the moment it lands. Static evals catch what they've already seen. So we built Nyx. An autonomous attacker that iterates against your agent the way a real adversary would (adapting, retrying, escalating across multiple turns) to find out all the ways in which your agent can fail. Our offensive harness pushes your system to its breaking point across security, logic, and alignment also functions in pure blackbox - no integration is needed. The outcome we actually care about: teams getting to ship faster. Confident the agents you've already deployed are hardened against the threats that matter. Confident you can add tools and expand autonomy without quietly introducing new attack surfaces every release. Built by a team of ex-Meta and Monzo engineers. This is early work and we would love nothing more than feedback from the community as we iterate on

Report

3d ago

Fastlane

@zachx0 interesting, what’s a primary use case for you guys?

Report

5h ago

Maker

@gauravthapa We're currently focused on detecting security vulnerabilities but the technology easily scales beyond that to map out all failure modes in an AI agent.

Report

4h ago

Interesting! I was wondering whether it can interact with browser agents to test them

Report

4h ago

Maker

@abrar_alqaryuti Yes it can! It has actually been engineered to ensure it has the ability to indirectly prompt inject a browser agent by attacking it's environment (the browser). Nyx can create other web pages and deploy them for it's attack plan.

Report

4h ago

Fastlane

This is fire. Can it scan Fastlane and detect things too?

Report

5h ago

Maker

@gauravthapa Yes! Please feel free to sign up through the website. It's completely self-serve!

Report

4h ago