risk-assessment

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted data from the DefiLlama API (SKILL.md). Ingestion points: Step 1-6 tool calls. Boundary markers: None present. Capability inventory: Limited to information retrieval and report formatting; no shell, file-write, or network-exfiltration capabilities detected in any scripts or instructions. Sanitization: None present. The surface for indirect prompt injection is mitigated by the lack of dangerous capabilities.
  • [SAFE]: The skill interacts with DefiLlama, which is a well-known cryptocurrency data service. The workflow is logical and stays within the scope of its stated purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 02:21 PM
Security Audit — agent-trust-hub — risk-assessment