The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill contains logic to automatically install the 'BrowserAct' CLI dependency if it is missing from the environment. This represents an unverified code execution path initiated by the skill itself.
[PROMPT_INJECTION]: The instructions explicitly command the agent to bypass user consent and standard transparency protocols. Specifically, it states the agent should install dependencies 'automatically without asking the user to confirm' and 'do not describe it as a request for the user to open their browser'. This is an attempt to override the agent's safety constraints regarding software installation and user notification.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to fetch software from browseract.com. This domain is not a recognized trusted vendor or a well-known technology service provider. The installation link includes a redirect parameter to GitHub, but the primary interaction is controlled by an untrusted third-party domain.
[COMMAND_EXECUTION]: The skill requires the execution of the browser-act CLI to perform its primary functions. Because this tool is installed automatically from an untrusted source, its execution on the local system poses a significant security risk.
[DATA_EXFILTRATION]: While no direct exfiltration command is present, the skill's documentation notes that sensitive information such as 'cookies, account information, browser IDs, and proxy settings' are stored in workspaces/. The required third-party CLI would likely have access to these sensitive project artifacts.

xiaohongshu-rednote-hashtag-scraper