The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple sensitive system commands for network manipulation and monitoring. \n
SKILL.md includes commands for tcpdump, radvdump, alive6, nmap, and ip6tables. \n
scripts/agent.py uses subprocess.run() to execute ip6tables -L -n --line-numbers to inspect firewall rules. \n- [PRIVILEGE_ESCALATION]: The skill requires administrative privileges (sudo) for the majority of its operations, including network sniffing, packet injection, and firewall modification. \n- [INDIRECT_PROMPT_INJECTION]: The scripts/agent.py script and various tools in SKILL.md (like scapy and mitm6) ingest raw, untrusted network traffic (ICMPv6, DHCPv6, NDP) without sanitization. \n
Ingestion points: scapy.sniff() and scapy.sr() in scripts/agent.py. \n
Boundary markers: None present to distinguish between legitimate and malicious network packets. \n
Capability inventory: The script can read network traffic, execute shell commands via subprocess, and write findings to a local JSON file. \n
Sanitization: No validation or filtering is applied to the data extracted from network packets before processing or reporting. \n- [CREDENTIALS_UNSAFE]: The skill automates the capture and local storage of sensitive authentication data. \n
The impacket-ntlmrelayx command in SKILL.md is configured to save relayed NTLM credentials to /tmp/loot. \n- [METADATA_POISONING]: There is an inconsistency in the skill's ownership metadata. The SKILL.md frontmatter identifies the author as mahipal, whereas the LICENSE file attributes copyright to mukul975.

exploiting-ipv6-vulnerabilities