AgentsThe story, in brief

Google Open-Sources Mantis: A Modular Skills Toolkit That Lets Coding Agents Find, Reproduce and Patch Vulnerabilities

Google open-sourced Mantis: a modular toolkit that turns coding agents into autonomous security engineers—finding, reproducing, and patching vulnerabilities end-to-end.

Illustration of two anonymous hands arranging task cards around an amber tool on a shared desk.
People, judgement and the changing nature of work.AI illustration by KeyNews
The KeyNews take

Why it matters

Mantis demonstrates agent autonomy scaling into real security workflows. This is the first production-grade open toolkit for agents to execute the full vulnerability lifecycle without human intervention—a blueprint for how agents move from narrow tasks to complex, multi-step professional work.

The key facts

7 to know
  1. Google open-sourced Mantis under Apache 2.0

  2. Stack-agnostic security skills toolkit for AI coding agents

  3. Full vulnerability lifecycle automation: sweep → filter → reproduce → patch → re-attack → score

  4. Sandbox-based reproduction and testing

  5. Risk scoring built in

  6. Released as demonstration-only (production readiness flagged)

  7. Published September 9, 2026

Go to the source

MarkTechPostmarktechpost.com

Publisher excerpt: Google has open-sourced Mantis, a stack-agnostic toolkit of security review skills for AI coding agents. It runs the full vulnerability lifecycle: sweep the code, filter false positives, reproduce the bug in a sandbox, patch it, re-attack the patch, then score the risk. Apache 2.0, and documented…
Read original report
Back to today's editionMore agents news

Keep reading

Related stories

More from Agents