lp-manga

Pass

Audited by Gen Agent Trust Hub on Mar 2, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [NO_CODE]: The skill consists entirely of Markdown prompt templates and architectural documentation. No executable scripts, binaries, or source code files are included, significantly reducing the attack surface.
  • [COMMAND_EXECUTION]: The skill does not utilize any tools or instructions that execute shell commands or interact with the operating system beyond standard file management.
  • [DATA_EXFILTRATION]: No network-related tools (curl, wget, etc.) or functions are present. The skill operates exclusively on local project files within the user's workspace.
  • [PROMPT_INJECTION]: The prompt templates use advanced engineering techniques such as Role-Play, Chain-of-Thought (CoT), and XML-based structured formatting. No malicious overrides or safety bypass instructions were detected.
  • [SAFE]: The skill implements a robust safety design by requiring explicit user confirmation and approval at the end of every phase (Requirements, Concept, Design) before proceeding to the next step.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 2, 2026, 04:32 AM