soia-pkm-interpret-article-analysis
Pass
Audited by Gen Agent Trust Hub on Jul 16, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill exposes an indirect prompt injection surface by reading and processing the content of articles from a local directory without sanitization.
- Ingestion points: Article files are read from the '/' as described in the '定位文章' section of SKILL.md.
- Boundary markers: The prompt instructions do not define delimiters or provide 'ignore' directives to prevent the agent from executing instructions that might be embedded within the article text.
- Capability inventory: The skill is capable of reading local article files and writing new interpretation documents ('<原文件名>-AI解读.md') to the local file system.
- Sanitization: There is no evidence of filtering, escaping, or validation of the data ingested from external articles before it is processed by the AI for summary generation.
- [EXTERNAL_DOWNLOADS]: The documentation provides an installation command 'npx skills add soia-team/soia-open-skills' which fetches content from a remote repository. This is a standard procedure for installing skills within this framework and references the author's own repository.
Audit Metadata