The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the agent-browser package from the NPM registry. This package provides the core automation capabilities and is maintained by a well-known organization.
[COMMAND_EXECUTION]: The skill operates by executing shell commands for the agent-browser tool. This includes commands for navigation (open), interaction (click, fill), and state management (state save). These are standard operations for a browser automation utility.
[INDIRECT_PROMPT_INJECTION]: The skill represents an attack surface for indirect prompt injection as it retrieves and processes content directly from external websites via snapshots and text extraction.
Ingestion points: Data enters the agent's context through commands like agent-browser snapshot, agent-browser get text, and agent-browser get html in SKILL.md.
Boundary markers: The skill does not define specific delimiters or instructions to ignore potential commands embedded in the fetched web content.
Capability inventory: The agent has the ability to execute CLI commands, browse the web, and write local files (e.g., screenshot, state save).
Sanitization: There is no explicit sanitization or validation mentioned for the content retrieved from web pages before it is presented to the agent.

agent-browser