chinese-article-writing
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill instructions are focused entirely on linguistic rules and document structure. There are no attempts to override agent behavior, bypass safety filters, or extract system prompts.
- [DATA_EXFILTRATION]: No network tools (curl, wget, etc.) or file system operations are requested or used. There is no evidence of credential harvesting or sensitive data access.
- [REMOTE_CODE_EXECUTION]: The skill does not perform any external downloads or execute remote scripts. It is composed entirely of static markdown documentation and YAML configuration.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted user input (articles to be polished). While it lacks explicit boundary markers in its default prompt, it also lacks any exploitable capabilities like file-writing or network access, rendering the risk negligible.
- Ingestion points: User-provided text content in 'SKILL.md' usage flow.
- Boundary markers: Not explicitly defined in the default prompt template.
- Capability inventory: None. The skill only performs text transformation and analysis.
- Sanitization: None identified.
- [DYNAMIC_EXECUTION]: While 'references/sample-article.md' contains illustrative pseudo-code for a Vite plugin, it is provided as a writing sample and is not intended for execution by the agent.
Audit Metadata