Security and governance for autonomous AI agents.
We build open-source tools so developers can run AI agents with confidence — protected by real infrastructure defenses, not just prompts.
| Project | What It Does | Language | OWASP ASI |
|---|---|---|---|
| 🏰 AI-Bastion | 8-layer defense-in-depth for AI agents on Linux. Firewalls, canary tokens, anti-injection, integrity checks, SOAR response. | Shell | 10/10 (infra) |
| 🛡️ AI-Bastion-Guardian | Windows-side security for agents in WSL2. Firewall rules, egress control, credential vault, port exposure detection. | PowerShell | 8/10 (perimeter) |
| ⚙️ rust-ai-governance-pack | Make AI coding agents produce Rust that is verifiable, secure by default, and repeatable. Rules, skills, katas, verification gates. | Rust | — |
┌─────────────────────────────────────────────────────────────┐
│ rust-ai-governance-pack │
│ "The rules AI follows when writing code" │
│ │
│ ┌───────────────────────────────────────────────────────┐ │
│ │ AI-Bastion-Guardian (Windows) │ │
│ │ "The outer wall" │ │
│ │ │ │
│ │ ┌─────────────────────────────────────────────────┐ │ │
│ │ │ AI-Bastion (Linux) │ │ │
│ │ │ "The fortress" │ │ │
│ │ │ │ │ │
│ │ │ [ Your AI Agent ] │ │ │
│ │ │ │ │ │
│ │ └─────────────────────────────────────────────────┘ │ │
│ └───────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────┘
Never attack. Always defend. Always share.
Defense > Offense. Intelligence > Retaliation.
Every layer exists because something actually happened. Built during the OpenClaw security crisis of February 2026 as a response to real CVEs, real malware campaigns, and real infrastructure threats.
OpenClaw · NanoClaw · LangChain · CrewAI · AutoGPT · Claude Code · Ollama · Any AI agent on Linux
Created by the Trinidad Operativa (Cerebro + Manos + Jefe)
"The butler is brilliant. Just make sure he remembers to lock the door." — Jamieson O'Reilly, Dvuln