skills/lijigang/ljg-skills/ljg-map/Gen Agent Trust Hub

ljg-map

Pass

Audited by Gen Agent Trust Hub on Sep 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill orchestrates a multi-step generation process that involves executing a local Python script (assets/gen_illustration.py) for image generation and a Node.js utility (capture.js) for rendering HTML templates into PNG images.
  • [EXTERNAL_DOWNLOADS]: The skill interacts with an external service at api.marswave.ai to generate thematic terrain maps using AI. This is a functional requirement for the skill's primary purpose.
  • [INDIRECT_PROMPT_INJECTION]: The skill utilizes a 'deep research' phase where it ingests data from external web sources via research tools. This information is subsequently used to construct image generation prompts and populate card metadata. This creates a surface where malicious content on processed web pages could potentially influence the output or image generation prompts, though the risk is assessed as low given the application context.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 26, 2026, 12:12 AM
Security Audit — agent-trust-hub — ljg-map