Zurich-based 0sec Labs has released 0, an open-source AI security harness. It runs vulnerability investigations, code review, pentesting, reverse engineering and fix generation. Developers drive it from a local browser console, a headless CLI or an MCP server. The 0sec Labs open-source AI security harness is licensed MIT OR Apache-2.0 and ships as a research preview.
TL;DR
- Wrapper: It is a software harness. Bring your own LLM from 14 listed providers.
- Runs on: Apple Silicon macOS, x64 or ARM64 Linux, or the
ghcr.io/0sec-labs/0container. No GPU needed. Model inference runs at your provider. - Performance: Strong self-reported CTF scores, plus a public record of upstream Linux kernel fixes.
- Best: 93/95 (97.9%) on XBOW web CTFs, gpt-5.4 cohort, any-success grouping.
- Worst: Black-box-only XBOW union of 81/104 (77.9%). A 2026 cold run scored 45/104 (43.3%).
- Bottom line: Best: one harness spans web, source, packages, binaries and kernels. Worst: headline scores are self-reported and not pass@1.
What is 0 from 0sec Labs?
0 is a control plane around LLMs and security tools. It is not a model. You plug in a provider, and 0 runs scoped workflows against authorized targets.
The research team describes it as a self-evolving, multi-model harness. 0sec Labs is backed by Y Combinator.
How does the 0 harness work?
The architecture docs state one rule: reproduce before trusting. Agentic scans follow a conceptual pipeline:
- Plan, Discover, Attack: a research agent picks vulnerability classes, maps endpoints and tests hypotheses.
- Triage: filters remove weak findings and check evidence completeness.
- Verify: a separate session re-exercises each candidate, with 5 turns per finding.
- Report: outputs include HTML, PDF, SARIF, Markdown and JSON.
Web pentesting is shell-first. The agent uses bash with curl, python3 and sqlmap, not a fixed HTTP tool set. Evidence carries a proof grade from candidate to impact-proven.
Compiled targets go through 0verse, a separate in-repo engine. It wraps Ghidra, angr and AFL++. Its local MCP server exposes 9 tools for scanning, triage, PoVs and reports.
How do developers use it?
Installation is one script. Running 0 opens the local browser console without an account. The console has chat, a findings queue and board, workflow templates, skills and memory.
Coding agents connect through 0 mcp-server --workflows. The CLI exposes the same workflow runner with commands like 0 workflow run and 0 runs list.
Model access is bring-your-own. The API keys page lists OpenAI, Anthropic, DeepSeek, OpenRouter, Azure, Qwen, Kimi, GLM and Grok. It also supports Claude, Codex and Gemini CLI runtimes.
Benchmarks: what do the numbers say?
All scores below are self-reported.
| Benchmark | Score | Conditions |
|---|---|---|
| XBOW web CTFs | 93/95 (97.9%) | gpt-5.4 cohort, any retained success counts |
| XBOW union, white-box | 102/104 | Retained artifacts, not additive |
| XBOW union, black-box | 81/104 | Retained artifacts, not additive |
| Cybench | 36/40 (90.0%) | Azure gpt-5.4, 3 retries |
| npm audit, 81 packages | F1 0.973 | 100% TPR, FPR 0.11, self-built ground truth |
The recorded cost was about $5.20 per solved XBOW challenge.
| Feature | 0 (0sec Labs) | Shannon (Keygraph) | Strix | CAI (Alias Robotics) |
|---|---|---|---|---|
| License | MIT OR Apache-2.0 | AGPL-3.0 | Apache-2.0 | MIT + proprietary research-only |
| Status | Research preview | Active | Active | Archived Aug 28, 2026 |
| Interfaces | Web console, CLI, MCP server | CLI, CI actions | CLI, local web viewer, agent skills | CLI/REPL, MCP support |
| Scope | Web, LLM, MCP, source, packages, binaries, kernel | Web apps and APIs, source required | Code, web apps, APIs | General offensive and defensive automation |
| Binary reverse engineering | Yes, 0verse (Ghidra) | Not disclosed | Not disclosed | Not disclosed |
| Models | BYO, 14 listed providers | BYO, Anthropic, OpenAI, xAI, Bedrock, local | BYO via LiteLLM | 300+ via LiteLLM |
| Requirements | macOS (Apple Silicon) or Linux | Docker, Node.js 18+ | Docker | Python 3.12 recommended |
What has 0 found in the wild?
The research page lists 30+ merged Linux kernel 0-days. Disclosures span Bluetooth, NFC, mac802154, TIPC and mlx5e subsystems. Library CVEs include jsonata, node-forge, protobuf.js and Wireshark.
Availability and pricing
The open-source harness is free. You pay your model provider. A hosted version, 0cloud, is on a waitlist. The pricing page lists $100 in trial credits and $5 per PR security test.
Key Takeaways
- 0 is an MIT/Apache-2.0 AI security harness from Y Combinator-backed 0sec Labs.
- It ships a browser console, headless CLI and MCP server for coding agents.
- Bring your own model: 14 providers plus Claude, Codex and Gemini CLIs.
- Self-reported XBOW score is 93/95; black-box union is 81/104.
Check out the GitHub repo, docs and research disclosures. Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.
Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us
Michal Sutter
Michal Sutter is a data science professional with a Master of Science in Data Science from the University of Padova. With a solid foundation in statistical analysis, machine learning, and data engineering, Michal excels at transforming complex datasets into actionable insights.
