redbookskills
Pass
Audited by Gen Agent Trust Hub on Mar 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill manages Google Chrome processes using the
subprocessmodule and executes JavaScript in the browser context via theRuntime.evaluateCDP command to automate UI interactions and data extraction on the Xiaohongshu platform. - [EXTERNAL_DOWNLOADS]: The
scripts/image_downloader.pyscript downloads images and videos from external URLs provided by the user or identified by the agent, facilitating the publication of remote content. - [PROMPT_INJECTION]: Indirect Prompt Injection Risk: The skill processes search results and comments extracted from Xiaohongshu in
scripts/cdp_publish.py. This untrusted content constitutes an attack surface if an adversary embeds malicious instructions in public posts. The skill mitigates this by usingjson.dumpsto escape data before inserting it into browser-side JavaScript, but the agent's interpretation of the extracted data remains a low-severity risk surface.
Audit Metadata