Talos - AI agent with a permission kernel between model and shell
Show HN: Talos – An AI agent with a permission kernel between model and shell
Talos is an AI agent that runs shell commands through a deterministic security kernel, authorizing each effect individually with exact arguments, valid once for 30 seconds. It offers a real shell, a confined Claude worker for building and testing, four chat channels, and 23 gated tools. The kernel enforces a path floor, hardline, and sandbox, with 2063 unit tests and 179 adversarial cases. Installer proves claims, starts nothing, and requires explicit identity configuration.
A brake that has to think first is not a brake. A model that is hallucinating cannot prevent its own shutdown.
- tomconnors
The world badly needs more options for running agents securely. I applaud you for making an attempt. But this landing page does nothing to win my trust. The copy is full of LLM-style language that is at best off-putting and at worst impenetrable. Even your comment in this thread looks like it was written by an LLM. If you don't care enough even to describe the thing yourself, why should we trust that you care enough to make the thing work well?
- ericpauley
Trademark C&D in 3, 2, 1...
- bakies
I run my agents on talos... But not this talos