chaos-engineering-resilience
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing untrusted external data.\n
- Ingestion points: System metrics (latency, error rates) and command outputs from chaos injection tools are read into the agent context as seen in SKILL.md and evals/chaos-engineering-resilience.yaml.\n
- Boundary markers: The skill lacks explicit instructions to treat tool outputs or system logs as untrusted data or to ignore embedded instructions.\n
- Capability inventory: The skill coordinates infrastructure-altering tasks such as pod termination, network partitioning, and resource exhaustion.\n
- Sanitization: No input validation or sanitization mechanisms are defined for data ingested from monitoring tools or external chaos agents.\n- [COMMAND_EXECUTION]: The skill facilitates the execution of powerful administrative commands.\n
- Evidence: SKILL.md and scripts/validate-config.json reference the use of tc, kubectl, and various chaos engineering tools to manipulate system state. This functionality is essential to the skill's primary purpose of resilience testing.
Audit Metadata