The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The browser-use python command allows for the execution of arbitrary Python code within a persistent session, providing a direct path for executing malicious logic.
[REMOTE_CODE_EXECUTION]: The browser-use eval command allows for the execution of arbitrary JavaScript code within the context of the active web page.
[DATA_EXFILTRATION]: The skill provides features for exporting browser cookies to local files (browser-use cookies export) and syncing local Chrome profile cookies to the Browser-Use cloud (browser-use profile sync), which poses a high risk of sensitive session token exposure.
[EXTERNAL_DOWNLOADS]: The installation instructions require downloading and installing the browser-use package and its dependencies (Chromium) from external repositories.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to interact with the command line, enabling it to perform various system-level operations.
[PROMPT_INJECTION]: The skill is highly susceptible to indirect prompt injection due to its core function of processing untrusted web content.
Ingestion points: Web page data is ingested through commands like browser-use state, browser-use extract, and browser-use run.
Boundary markers: No specific delimiters or safety instructions are used to separate untrusted web content from agent instructions.
Capability inventory: The skill possesses high-impact capabilities including arbitrary code execution (python, eval) and sensitive data handling (cookies export).
Sanitization: There is no evidence of sanitization or filtering of the web content before it is processed by the AI agent.

browser-use