The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents the use of the agent-browser CLI for automating browser interactions. The commands are limited to browser control (navigation, clicking, typing) and are consistent with the tool's intended use.
[PROMPT_INJECTION]: The skill operates on content retrieved from external websites, which presents a surface for indirect prompt injection. This is a known risk for any tool that processes untrusted external data, and the skill itself does not contain malicious instructions.
[DATA_EXFILTRATION]: Commands for taking screenshots and reading element text are included for automation purposes. No patterns of sending sensitive local data to external endpoints were identified.

core-agent-browser