kali-pentest-ai-agent

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). This skill contains many examples that embed plaintext passwords, NTLM/hash strings, and Authorization tokens directly in commands (e.g., -p 'Password123!', NTLM hashes, -H "Authorization: Bearer $API_TOKEN" / user:password), which would require the LLM to include secret values verbatim in generated commands or code and thus presents a high exfiltration risk.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The skill contains explicit, high-risk offensive capabilities (RCE/exploit examples, reverse shells, credential-dumping like mimikatz/impacket, cloud metadata and Kubernetes secrets access, lateral movement and exfiltration commands) that enable data theft and full system compromise — a powerful dual-use toolkit that can be readily abused despite stated "authorization" notes.

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 the agent to run as root or use sudo/SSH to a root server, install packages, add SSH keys (altering auth), grant Docker capabilities (NET_ADMIN/NET_RAW), and run/exploit commands that modify system state — all actions that require elevated privileges and can compromise the host.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)