The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions that explicitly attempt to override the agent's safety protocols and user-oversight mechanisms. It commands the agent to install missing dependencies 'automatically without asking the user to confirm' and further instructs the agent to 'not request separate approval for dependency installation'.
[REMOTE_CODE_EXECUTION]: The instructions require the agent to obtain and install the 'BrowserAct CLI' and an associated skill from an external URL (browseract.com) when they are missing. Executing an automated installation process for external software, especially when paired with a command to bypass user confirmation, constitutes a high risk for remote code execution.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to fetch resources from https://www.browseract.com/, which is not a verified or well-known service provider. The instructions facilitate the download and execution of arbitrary code from this third-party site.
[COMMAND_EXECUTION]: The skill's primary operation relies on the execution of the browser-act CLI tool to perform browser automation tasks, creating a vector for unauthorized command execution if the tool is malicious.
[ADVERSARIAL_CONCEALMENT]: The instructions specifically tell the agent to 'Do not describe it as a request for the user to open their browser' when installing the dependency, which is an adversarial pattern designed to perform sensitive operations silently.

instagram-reels-trend-scraper