The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill exposes an evaluate endpoint (/api/v1/executor/evaluate) that allows for the execution of arbitrary JavaScript code within the browser context. This dynamic code execution capability presents a significant risk if an agent is manipulated into running malicious scripts.
[COMMAND_EXECUTION]: The skill utilizes shell-based curl commands to interact with a local browser automation API. It provides extensive control over browser sessions, including tab management, file uploads, and page evaluation.
[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection as it processes untrusted data from external websites and possesses powerful interactive capabilities. * Ingestion points: The skill ingests external web data through GET /snapshot, POST /extract, GET /page-text, and GET /page-content. * Boundary markers: The instructions lack requirements for the agent to use specific delimiters or to ignore instructions embedded in the extracted web data. * Capability inventory: The skill has the ability to execute code in the browser, interact with forms, and navigate the web. * Sanitization: There are no explicit instructions to sanitize or validate the content retrieved from external URLs.

browserwing-executor