soia-pkm-clip-douyin
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill performs its stated purpose of clipping Douyin videos. Its file system access is restricted to the user's Obsidian vault, the
~/Downloadsdirectory, and a standard configuration path in the user's home directory. All operations are transparently documented. - [EXTERNAL_DOWNLOADS]: The skill downloads MP4 files from Douyin's infrastructure to the local machine. As these downloads target a well-known service and are part of the skill's primary function, they are considered safe.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it ingests untrusted text from Douyin video descriptions and author metadata.
- Ingestion points: Video descriptions and author nicknames are retrieved via Playwright response interception in
archive_douyin.py. - Boundary markers: Content is delimited by standard Markdown headers (e.g.,
## 原文) and YAML frontmatter keys, though no explicit instructions are provided to the agent to ignore potentially malicious commands within the ingested text. - Capability inventory: The skill has permissions to write to the file system and perform network requests to Douyin domains.
- Sanitization: The script implements filename sanitization to prevent path traversal or invalid characters, but the content of the video description is written directly to the vault without additional escaping or filtering.
Audit Metadata