discover-opportunities

Pass

Audited by Gen Agent Trust Hub on Apr 2, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill ingests untrusted data from customer feedback sources, creating a surface for indirect prompt injection.\n
  • Ingestion points: The agent retrieves customer feedback insights, specific user quotes, and raw comments via the get_feedback_insights, get_feedback_mentions, and get_feedback_comments tools as described in Phase 2d of SKILL.md.\n
  • Boundary markers: The instructions do not define clear delimiters or specific instructions for the agent to treat retrieved feedback as data only or to ignore any commands contained within the feedback text.\n
  • Capability inventory: The skill is primarily focused on data analysis and synthesis using Amplitude-specific tools (e.g., query_charts, get_experiments). While it does not explicitly invoke shell commands or external network tools like curl, it acts as an analytical processor for untrusted strings.\n
  • Sanitization: There is no mention of sanitizing, escaping, or validating the feedback content before it is processed and summarized by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 2, 2026, 08:19 PM
Security Audit — agent-trust-hub — discover-opportunities