The Agent Skills Directory

[DYNAMIC_EXECUTION]: The execute function allows the agent to run arbitrary JavaScript within the browser context, as documented in references/commands.md. This is a core feature for web automation and data extraction.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from web pages, creating a surface for potential instruction injection. Ingestion points: The open and snapshot functions retrieve page elements and text. Boundary markers: The skill does not use specific delimiters to distinguish between system instructions and data extracted from the web. Capability inventory: The skill can interact with page elements, execute scripts, and run shell commands via the belt tool. Sanitization: No sanitization of extracted DOM content or text is provided.
[DATA_EXFILTRATION]: The skill provides patterns for accessing sensitive browser information; references/authentication.md demonstrates how to use execute to retrieve document.cookie.
[EXTERNAL_DOWNLOADS]: The skill directs users to install a CLI tool from https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md. This is documented as a necessary setup step for the vendor's platform.

agent-browser