c2siorg / agentic-cognitive-firewallView on GitHub
Application-level policy layer for LLM agents. It blocks unauthorised actions by checking what a task may do and where each value came from, not by trying to recognise attack text. Policies in Rego, decided in an isolated sidecar with no model in the loop, so protection holds for small local models too.
☆20Oct 26, 2026Updated this week

Alternatives and similar repositories for agentic-cognitive-firewall

Users that are interested in agentic-cognitive-firewall are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.

Sorting:

Are these results useful?