0sec Labs open-sources 0, an AI security harness
Original title: 0sec Labs Open-Sources 0, an AI Security Harness With Web UI, CLI and MCP
The MIT or Apache-2.0 research preview uses a bring-your-own model and self-reports 93 of 95 XBOW solves.
Zurich-based, Y Combinator-backed 0sec Labs released 0, an open-source harness for vulnerability investigation, code review, pentesting, reverse engineering, and fix generation. Users run it from a local browser console, a headless CLI, or an MCP server on Apple Silicon macOS, Linux, or a container, with inference at one of 14 listed providers. The lab reports strong CTF scores and more than 30 merged Linux kernel fixes, while noting scores are self-reported and not pass@1. A hosted 0cloud service is on a waitlist.
Unconfirmed: Benchmark scores and the count of merged Linux kernel fixes are self-reported by 0sec Labs.
- Best cited XBOW result is 93/95 on a gpt-5.4 cohort using any-success grouping.
- Black-box XBOW union is 81/104; a 2026 cold run scored 45/104.
- Cybench is listed at 36/40 with Azure gpt-5.4 and three retries.
- 0verse wraps Ghidra, angr, and AFL++ and exposes nine MCP tools.
- 0cloud lists $100 in trial credits and $5 per pull-request security test.
- Vendors
- 0sec Labs
- Organizations
- Y Combinator
- AI models
- gpt-5.4
From the original article
7 min read at marktechpost.com0sec Labs has open-sourced 0, an AI security harness for code review, pentesting, reverse engineering and fixes. It runs locally through a browser console, a headless CLI and an MCP server, with your own model provider. The post 0sec Labs Open-Sources 0, an AI Security Harness With Web UI, CLI and MCP appeared first on MarkTechPost.
By Michal Sutter · marktechpost.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.