The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to execute the nmap binary through a Python wrapper ($NMAP_MCP_SCRIPT). It constructs and runs shell commands based on user-provided parameters such as target and ports for various scan types (SYN, TCP, UDP, Ping).
[PRIVILEGE_ESCALATION]: Several tools (nmap_arp_discovery, nmap_syn_scan, and nmap_udp_scan) require the cap_net_raw Linux capability to function. While more granular than full root/sudo access, this involves granting the skill's environment elevated permissions to interact with raw network sockets.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external, potentially untrusted network sources.
Ingestion points: Data is ingested from network targets during host discovery and port scanning (SKILL.md).
Boundary markers: The instructions do not define specific delimiters or warnings to prevent the agent from interpreting tool output as new instructions.
Capability inventory: The skill has the capability to execute shell commands (nmap) and write results to logs (SKILL.md).
Sanitization: There is no mention of sanitizing or escaping network-returned data (such as service banners or hostnames) before it is returned to the agent's context.

nmap-network-scan