The Agent Skills Directory

[SAFE]: The skill implements robust security guardrails defined in the frontmatter and SAFETY.md, which explicitly prohibit malicious behaviors such as creating Active Directory persistence (golden tickets, ACL modifications) or bypassing safety filters.
[SAFE]: A mandatory credential redaction policy is enforced, ensuring that sensitive material like NTLM hashes or passwords are never displayed in full in the agent's response, but are instead directed to secure output files.
[SAFE]: The execution script active_directory.py requires an explicit --i-have-authorization flag for live actions and performs scope validation against a scope.json file to prevent unauthorized target testing.
[SAFE]: External references in references/tools.md point to well-known, industry-standard security tools from reputable GitHub organizations (e.g., BloodHoundAD, Impacket).
[SAFE]: The skill restricts all sensitive outputs to a user-defined --output directory, preventing the storage of findings in world-readable locations like /tmp.

pentest-active-directory-hardened