# 0sec Labs open-sources 0, an AI security harness for code review, pentesting and fixes

> It runs locally through a browser console, a headless CLI or an MCP server, with whatever model provider you bring.

- URL: https://devisdead.ai/news/0sec-labs-open-sources-0-an-ai-security-harness-for-code-review-pentesting-and-f
- Published: 2026-11-09T12:42:03Z
- Author: DevIsDead Newsroom
- Section: Dev
- Tags: security, open source, mcp, pentesting
- Originally reported by MarkTechPost (https://www.marktechpost.com/2026/11/08/0sec-labs-open-sources-0-ai-security-harness/).

Another AI security tool just went open source. 0sec Labs has released 0, an AI security harness for code review, penetration testing, reverse engineering and fixes, according to [MarkTechPost](https://www.marktechpost.com/2026/11/08/0sec-labs-open-sources-0-ai-security-harness/).

It runs locally in three ways: a browser console, a headless command-line tool for scripts and CI, and an MCP server so other agents can call it. You bring your own model provider.

Details are thin: the summary doesn't say which models it supports best, how it's licensed or how it performs. But the shape is notable. Local-first, provider-agnostic security harnesses are the defender's answer to [free models that now hack nearly as well as gated ones](/news/free-chinese-ai-models-now-hack-almost-as-well-as-the-locked-up-frontier-ones-a-): if attackers can point open tools at your code, you want the same capability running in your own pipeline first, without sending your source to a third party.

**[Scoreboard](/scoreboard) update:** AI +1. Defenders get a free, local AI security harness to run on their own code.

Stay undead.