skills/mukul975/anthropic-cybersecurity-skills/performing-wireless-security-assessment-with-kismet/Gen Agent Trust Hub
performing-wireless-security-assessment-with-kismet
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PRIVILEGE_ESCALATION]: The skill instructs the user to use
sudofor several high-privilege operations necessary for wireless monitoring. - Evidence:
SKILL.mdcontains commands such assudo airmon-ng check kill,sudo ip link set wlan0 down, andsudo kismet -c wlan0to enable monitor mode and start the capture server. - [COMMAND_EXECUTION]: The skill provides numerous shell commands to configure the host system's wireless stack and run the Kismet assessment suite.
- Evidence: Preparation steps in
SKILL.mdinvolve executingiwconfig,iw list, andip linkto manage physical hardware interfaces. - [INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by ingesting and displaying untrusted data from the wireless environment.
- Ingestion points: Wireless frames (SSIDs, MAC addresses, and manufacturer data) are captured by Kismet and processed by
scripts/agent.pyand the analysis scripts inSKILL.mdvia JSON/REST API. - Boundary markers: Absent; external SSIDs and device names are interpolated directly into reports and printed to the terminal.
- Capability inventory:
scripts/agent.pyhas network capabilities via therequestslibrary to communicate with the local Kismet API;SKILL.mdcontains multiple subprocess calls for environment configuration. - Sanitization: Absent; device metadata is processed as strings without validation or escaping, which could technically be used to deliver malicious payloads to an agent reviewing the assessment results.
Audit Metadata