NYC

manim-composer

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • SAFE (SAFE): No security issues detected.
  • Prompt Injection: No instructions attempting to override agent safety or system prompts were found. The instructions are focused solely on the intended use case of educational video planning.
  • Data Exposure & Exfiltration: No access to sensitive file paths, credentials, or unauthorized network operations were detected.
  • Unverifiable Dependencies: The skill does not install external packages or execute remote scripts. It only references standard libraries (ManimCE, ManimGL) for documentation purposes.
  • Dynamic Execution: There are no instances of eval(), exec(), or runtime code generation. The provided code blocks are static documentation examples.
  • Indirect Prompt Injection (Surface Analysis): While the skill performs web research (Phase 1), which involves ingesting untrusted data, the output is restricted to a static markdown file (scenes.md). There are no high-risk capabilities (like automated script execution) that could be triggered by malicious content in search results.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:34 PM