securing-systems

Pass

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill contains numerous code snippets and command-line examples illustrating various security testing techniques. For instance, red-team.md and pentest.md provide templates for shellcode execution, SQL injection, and lateral movement. These are explicitly presented as educational templates for "Defense Validation" and "Security Testing" purposes.
  • [REMOTE_CODE_EXECUTION]: red-team.md documents the installation of the Sliver C2 framework using the command curl https://sliver.sh/install | sudo bash. This pattern is used to install a well-known open-source security tool and is intended for use in authorized red-team environments.
  • [EXTERNAL_DOWNLOADS]: The skill references several external resources and tools. For example, blue-team.md provides instructions to install the sigma-cli package via pip, and threat-intel.md references various online threat intelligence platforms. These downloads target established security tools and well-known industry services.
  • [PROMPT_INJECTION]: While the skill contains instructions for the AI agent on how to handle security assessments, these instructions include significant safety constraints in SKILL.md. These constraints mandate the use of RFC 5737 reserved IP addresses, placeholders for credentials, and strict adherence to authorized scopes.
Audit Metadata
Risk Level
SAFE
Analyzed
May 18, 2026, 02:55 PM
Security Audit — agent-trust-hub — securing-systems