Grith AI

We ran Claude Code and Codex under strace for a benchmark. Claude Code attempted 752 reads of /proc/*/environ. 256 succeeded. Codex: 0. Not malice. Just an agent with your whole machine as ambient authority. I built a supervisor that sits underneath. It is live.

🔥Hot take 🔥 If your AI agent can run shell commands, read files, and hit the network… …and you’re relying on prompts or “alignment” for security… you don’t have a security model. You have a demo. 🧵👇 #AI #infosec

Alibaba's AI Agent Hijacked GPUs and Dug Reverse SSH Tunnels grith.ai/blog/alibaba-rome-a…

The Trivy Supply Chain Attack Reached LiteLLM grith.ai/blog/litellm-compro…

Meta's Rogue AI Agent Gave Engineers Access They Shouldn't Have Had grith.ai/blog/meta-ai-agent-… #meta #ai #aiagents

Google's A2A Protocol Has Zero Defenses Against Prompt Injection grith.ai/blog/a2a-protocol-z…

A GitHub Issue Title Compromised 4,000 Developer Machines grith.ai/blog/clinejection-w…

Permission Fatigue Is Not a UX Problem. It Is a Security Failure. grith.ai/blog/permission-fat… #ai #claudecode #codex #gemini

NemoClaw vs grith: Sandbox for One Agent vs Security for All grith.ai/blog/nemoclaw-vs-gr…

AI Agent Backdoors Trivy Security Scanner, Weaponizes a VS Code Extension grith.ai/blog/ai-agent-compr… #pwned #aiagent #ai #vibecoding