results-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external experiment logs and data files (CSV, JSON, TSV). While this introduces an ingestion surface for untrusted data, the skill includes explicit validation steps and a 'quarantine' policy for contradictory data, effectively mitigating risks associated with untrusted data processing.
- [COMMAND_EXECUTION]: The skill mentions a command layer (
/analyze-results) and instructions to generate figures, which typically involve local tool execution. However, these are strictly scoped to the primary analysis purpose and include safe defaults like a 'read-only audit mode' that prevents file writes when evidence is incomplete.
Audit Metadata