analyse-defi

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data via a DataPackage JSON object.
  • Ingestion points: Data fields defined in SKILL.md for protocol revenue, TVL, and fee distribution.
  • Boundary markers: Absent.
  • Capability inventory: None (the skill has no access to tools, scripts, or network requests).
  • Sanitization: Implicitly handled by instructions to strictly follow a defined JSON schema and ignore other fields.
  • [SAFE]: No malicious patterns, obfuscation, or remote code execution risks were detected. The skill functions purely as an analytical prompt.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 02:21 PM
Security Audit — agent-trust-hub — analyse-defi