vervet is an adversary-emulation instrument. Run it only against systems you are explicitly authorized to test. Every state-changing technique requires an Ed25519-signed scope manifest; acting outside that scope is a hard, typed refusal by design. You are responsible for holding valid authorization for any engagement.

If you find a security issue in vervet itself — for example, a way to mint a Grant outside the authorization gate, forge or replay a scope manifest, bypass the CIDR/technique/time checks, or leak credential material into evidence — please report it privately. Do not open a public issue.

Email don@codetestcode.io with:

• a description of the issue and its impact,
• steps or a proof of concept to reproduce it,
• the commit or version affected.

We aim to acknowledge reports within a few days and will coordinate a fix and disclosure timeline with you.

In scope: the authorization spine (vervet-scope), the evidence/audit chain (vervet-core, vervet-engage), and anything that could cause a technique to act without a valid Grant or to write secrets into a receipt.

Out of scope: misuse of the tool against unauthorized targets (that is on the operator, not a vulnerability in vervet), and findings that require an attacker to already hold the signing key.

vervet is pre-1.0; security fixes land on main and the latest release.