root-cause-tracing-process
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data, including incident records, logs, and external failure reports, which are processed through the $ARGUMENTS input and various file-reading capabilities (SKILL.md lines 7, 34-45, 59-67).\n
- Ingestion points: Untrusted data enters the agent context through the
<tracing_input>argument and by reading external logs or artifacts during the tracing process.\n - Boundary markers: The skill uses explicit tags to delimit input and provides clear instructions to separate reported observations from agent-verified evidence (SKILL.md line 62).\n
- Capability inventory: The skill utilizes shell access, diagnostic tools, and file system operations to investigate failure mechanisms (SKILL.md lines 48-52, 78-90).\n
- Sanitization: The skill employs a rigorous 'Evidence-chain protocol' (SKILL.md lines 19-32) that requires every material claim to be verified against direct observations (e.g., command output or source code), providing a strong reasoning-based defense against malicious instructions embedded in processed data.
Audit Metadata