logic-explain

Pass

Audited by Gen Agent Trust Hub on Jul 5, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by design, as it is instructed to ingest and analyze untrusted code snippets from the user or project context.
  • Ingestion points: The skill processes user-provided functions and methods as described in the execution explain guide.
  • Boundary markers: The instructions lack explicit requirements for using delimiters or isolation markers to distinguish untrusted code from the agent's logic instructions.
  • Capability inventory: The skill does not implement a restricted execution environment or specific capability limits for the agent while it is processing untrusted content.
  • Sanitization: There are no steps defined to sanitize or filter input code for embedded malicious instructions before the analysis process begins.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 5, 2026, 09:37 AM
Security Audit — agent-trust-hub — logic-explain