xiaohongshu-search
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The Python script
scripts/fetch_xhs_hot_articles.pymakes network requests toredfox.hkto retrieve content, which is a core function of the skill. \n- [COMMAND_EXECUTION]: The agent executes a local Python script with arguments (keywords) extracted from user prompts. This presents a minor risk of argument manipulation or path traversal during the generation of the HTML report if the agent does not strictly validate the extracted terms. \n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes and displays content from an external API. \n - Ingestion points: External data enters the context through API responses fetched by
fetch_xhs_hot_articles.py. \n - Boundary markers:
SKILL.mdenforces strict output formatting (Markdown tables) and explicitly restricts the agent from interpreting or analyzing the data, which helps mitigate the risk of embedded malicious instructions. \n - Capability inventory: The skill has the capability to perform network requests, write local files, and create calendar entries. \n
- Sanitization: The script includes basic HTML entity escaping for its file output, but no further sanitization is applied to the data shown in the agent's response.
Audit Metadata