analyze-experiments

Pass

Audited by Gen Agent Trust Hub on Aug 19, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by ingesting and processing untrusted data from external sources.
  • Ingestion points: The skill retrieves experiment descriptions via Amplitude:get_from_url (Step 0) and user feedback content via Amplitude:get_feedback_insights (Step 7).
  • Boundary markers: The instructions do not define clear delimiters or provide instructions to the agent to ignore potential malicious prompts embedded within the external metadata or feedback strings.
  • Capability inventory: The skill possesses write capabilities through Amplitude:create_experiment, allowing it to modify platform state based on its analysis.
  • Sanitization: There is no evidence of filtering, escaping, or validation of the content retrieved from these external sources before it is processed for insights or recommendations.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 19, 2026, 02:13 AM
Security Audit — agent-trust-hub — analyze-experiments