lov-wechat-article-operator
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONOBFUSCATIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external content from WeChat articles, creating a surface for indirect prompt injection. Ingestion points: WeChat article content (referenced in SKILL.md). Boundary markers: Presence or absence is unverifiable due to the encrypted logic. Capability inventory: Reading, editing, saving, and reloading articles. Sanitization: Verification of content sanitization is impossible due to encryption.
- [OBFUSCATION]: The core logic is distributed as encrypted files (AES-256-GCM), preventing static analysis of the scripts and instructions. Evidence: Encrypted files include scripts/validate_skill.py.enc and scripts/verify_article_state.py.enc.
- [DYNAMIC_EXECUTION]: The skill decrypts and loads logic into the agent's context at runtime. Evidence: The decryption process described in SKILL.md using the lovstudio-skill-helper tool.
- [EXTERNAL_DOWNLOADS]: The skill installation and operation depend on external tools from the vendor lovstudio. Evidence: Usage of npx lovstudio and uvx lovstudio-skill-helper commands.
Audit Metadata