fear

Pass

Audited by Gen Agent Trust Hub on Apr 27, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill defines a cognitive risk-management methodology (pre-mortem analysis) using metaphorical framing. It contains no executable code, scripts, system commands, or external dependencies.
  • [PROMPT_INJECTION]: Ingestion points: project plans or decisions are ingested as input (SKILL.md, Procedure step 2). Boundary markers: the agent is instructed to restate targets and 'no-touch boundaries' before beginning analysis (SKILL.md, Procedure step 1). Capability inventory: the agent relies on its standard toolset for analytical tasks; no specialized code-execution or network capabilities are introduced by the skill. Sanitization: mitigation is provided through explicit scoping and the analytical (rather than execution-oriented) nature of the procedure.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 27, 2026, 09:14 AM
Security Audit — agent-trust-hub — fear