laohan-fengmianzhuzige

Pass

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONMETADATA_POISONINGEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/build_cover.py executes shell commands and local binaries to perform image processing and retrieve configuration.
  • Evidence: Spawns zsh to retrieve the ARK_API_KEY environment variable and calls the remotion CLI for rendering output.
  • [DYNAMIC_EXECUTION]: The skill uses computed paths to locate its execution runtime, defaulting to local directories specific to the author's workstation if environment variables are not set.
  • Evidence: The RUNTIME variable in scripts/build_cover.py is derived from os.environ.get("LAOHAN_REMOTION_RUNTIME") or a hardcoded fallback path.
  • [METADATA_POISONING]: The skill instructions and sample configuration files contain hardcoded absolute paths that reveal the author's local username and directory structure.
  • Evidence: Absolute paths like /Users/hanzhmacbookair/Documents/... are present in SKILL.md, samples/brief-example.json, and scripts/build_cover.py.
  • [EXTERNAL_DOWNLOADS]: The skill fetches generated background images from Volcengine's API (a well-known cloud service) during the image creation process.
  • Evidence: urllib.request.urlretrieve is used to download content from https://ark.cn-beijing.volces.com.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 9, 2026, 03:38 AM