incident-commander
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted incident descriptions and event logs which are then interpolated into severity classification reports and Post-Incident Review (PIR) documents. This creates a surface where malicious instructions embedded in logs could influence an agent's analysis or subsequent actions. \n
- Ingestion points:
scripts/incident_classifier.pyreads incident descriptions;scripts/timeline_reconstructor.pyandscripts/pir_generator.pyingest JSON-formatted incident and timeline data. \n - Boundary markers: The skill templates do not utilize explicit boundary markers or delimiters to isolate untrusted data from generated report instructions. \n
- Capability inventory: The tools generate high-impact incident response artifacts, including severity classifications and RCA documents, which steer the agent's incident management strategy. \n
- Sanitization: The implementation uses simple string interpolation into templates without escaping or filtering content for potential LLM instructions.
Audit Metadata