adcs-access-and-relay

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt instructs embedding secrets verbatim into commands and outputs (NT hashes, plaintext passwords, -p 'Pass', echo "BASE64_CERT", etc.), requiring the LLM to handle and produce secret values directly and thus posing an exfiltration risk.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This document is a high-risk adversarial playbook: it contains explicit, actionable instructions for credential theft (including DCSync/secretsdump and forging/obtaining machine/privileged certificates), CA/template manipulation to issue arbitrary certificates, NTLM/Kerberos relay and coercion techniques, remote code execution (writing webshells and CRL/file writes), and guidance to minimize detection — indicating deliberate malicious intent and backdoor/persistence capabilities.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill explicitly instructs modifying AD/CA state (template/ACL changes, publishing templates, adding CA officers, writing files/webshells to CA directories, restarting certsvc), and also requires running privileged tools/commands (ntlmrelayx/krbrelayx/mitm6 and even sudo ntpdate) which together push the agent to change the machine/system state and perform privileged actions.