The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to install and execute various third-party security tools using the Bash tool, including nmap, sqlmap, semgrep, bandit, and gitleaks.
[PRIVILEGE_ESCALATION]: The references/penetration-testing.md file provides specific commands for identifying privilege escalation vectors on Linux systems, such as searching for SUID binaries (find / -perm -4000) and checking sudo permissions (sudo -l).
[DATA_EXFILTRATION]: The skill provides instructions for local credential harvesting under its lateral movement section. This includes searching for passwords in other users' home directories (grep -r "password" /home/*/) and inspecting shell history for secrets (cat ~/.bash_history | grep -i "pass\|pwd\|secret").
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data (such as source code and configuration files) while possessing high-impact shell execution capabilities.
Ingestion points: The skill reads project source code, infrastructure files, and logs via the Read, Grep, and Glob tools for auditing purposes.
Boundary markers: The instructions include process-level requirements for authorization but lack technical delimiters or "ignore" instructions for the untrusted content being audited.
Capability inventory: The agent has full access to Bash for command execution and file system manipulation across the project environment.
Sanitization: The documentation does not specify methods for sanitizing or escaping the untrusted code snippets before they are analyzed or processed by the agent.
[METADATA_POISONING]: Automated scans identified the documentation URL (jeffallan.github.io) and the SKILL.md file as having a poor reputation, likely due to the concentration of offensive security terminology and penetration testing command patterns.

security-reviewer