skills/simota/agent-skills/field/Gen Agent Trust Hub

field

Pass

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted participant data during qualitative and quantitative analysis phases, creating an indirect prompt injection surface.
  • Ingestion points: Data enters the agent context during the ANALYZE phase, which processes interview transcripts, session notes, and recordings as described in SKILL.md and reference/analysis-and-synthesis.md.
  • Boundary markers: The skill contains explicit instructions to separate observation from interpretation and requires the use of a cognitive bias checklist (reference/bias-checklist.md) to maintain analytical integrity.
  • Capability inventory: The skill produces synthesis artifacts like personas and journey maps and communicates with downstream agents (Cast, Echo, Spark). It lacks high-privilege capabilities such as arbitrary file-system writes or network execution.
  • Sanitization: The skill mandates PII anonymization, consent management, and data governance standards (e.g., SOC 2 Type II, GDPR) for any research platforms utilized.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 22, 2026, 01:06 PM