gtrace-path-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes the gtrace binary using the $MCP_CALL Python wrapper. It supports multiple sub-commands including traceroute, mtr, and globalping for network path analysis.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests and displays hostnames and IP addresses from external network routers which could theoretically contain malicious strings.
- Ingestion points: Network hop data (hostnames, IP addresses) returned by traceroute and MTR tools as described in SKILL.md.
- Boundary markers: None present in the description of data interpolation.
- Capability inventory: Subprocess execution of the gtrace binary which requires cap_net_raw for raw socket probes.
- Sanitization: No explicit sanitization or filtering of hop hostnames is mentioned before display or recording in GAIT.
Audit Metadata