The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The workflow instructions require the use of sudo to modify system-level kernel parameters (net.ipv4.ip_forward), manage firewall rules via iptables, and execute network interception tools such as bettercap and arpspoof. These operations are consistent with the skill's primary purpose of network security testing and simulation.
[COMMAND_EXECUTION]: The assessment script (scripts/agent.py) uses subprocess.run to execute the curl utility to inspect HSTS headers and redirect chains. The skill also documents the use of external tools like Bettercap, sslstrip2, and Wireshark for intercepting and analyzing network traffic.
[REMOTE_CODE_EXECUTION]: Automated scanning flagged a remote code execution pattern in SKILL.md. Analysis confirms this is a false positive: the command curl -s "https://hstspreload.org/api/v2/status?domain=example.com" | python3 -m json.tool uses the Python standard library's json.tool module to pretty-print JSON data fetched from a well-known service; it does not execute the remote content as code.
[INDIRECT_PROMPT_INJECTION]: The agent.py script processes data from external URLs, representing a potential surface for indirect prompt injection.
Ingestion points: Data is ingested via curl from user-specified target URLs and the official hstspreload.org API (in scripts/agent.py at lines 21, 41, 62, and 79).
Boundary markers: The script does not utilize boundary markers or delimiters when processing output from external tools.
Capability inventory: The script's capabilities are limited to executing curl via subprocess.run (lines 14, 42, 57, 72, and 85) for assessment purposes and does not have file-writing or persistent network capabilities.
Sanitization: Content retrieved from external sources is printed to the console without sanitization or escaping.

performing-ssl-stripping-attack