skill-cross-platform-diff
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized access to sensitive system files were detected. The skill functions as intended for content analysis.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data via the '话题/内容主题' (Topic/Content Theme) field in SKILL.md. While the instructions lack explicit boundary markers and sanitization logic to separate this input from system prompts, the risk is assessed as safe because the skill's capability inventory is restricted to text analysis and local file output, with no access to network tools or shell execution environments.
- [REMOTE_CODE_EXECUTION]: The metadata file EASEL-META.md references external GitHub repositories (yiyangcpa/Cross-Platform-Comparison and NanmiCoder/MediaCrawler) for research attribution. These are documented as static references; the skill contains no logic to download, install, or execute code from these or any other remote sources.
Audit Metadata