net
Pass
Audited by Gen Agent Trust Hub on Nov 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies heavily on executing external binaries via the
subprocessmodule to perform its functions. While arguments are passed as lists to mitigate shell injection, the paths to the tools (liketshark) are configurable. - Evidence:
scripts/net_runtime.pycontainsresolve_tool_pathwhich uses values fromconfig.json. Scriptsnet_capture.py,net_analyze.py,net_ping.py,net_scan.py, andnet_stats.pyall invoke system commands. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources, specifically live network traffic and pcap files, which could contain malicious payloads designed to influence the agent.
- Ingestion points: Network interface captures in
scripts/net_capture.pyand file reading inscripts/net_analyze.py. - Boundary markers: None explicitly implemented to separate untrusted packet data from agent instructions.
- Capability inventory: The skill has extensive capabilities including file system writes, network connectivity testing, and packet analysis.
- Sanitization: Data is structured into JSON, but raw strings from banners or packet summaries are included.
- [DYNAMIC_EXECUTION]: The skill uses dynamic importing for certain platform-specific or standard library modules.
- Evidence:
scripts/net_ping.pyuses__import__("time").scripts/net_runtime.pyusesimport msvcrtwithin thejson_write_lockfunction.
Audit Metadata