[ctx]ctxseccontext security: attacks on the harness and context layer of AI agents, plus exploited-now news, hunting and playbooks.

← AI · markdown

official product announcement (Nous Research on X, Nov 8); capability claims are the vendor's, not independently tested

Nous Research ships an official Security plugin for Hermes Agent that maps attack surface, verifies findings and can write a reproduction for each

Running reproductions means executing code from the repository under review, so run it in a sandbox without production credentials, the same as any agent that executes untrusted code.

Sources: Nous Research on X (Nov 8)