The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from arbitrary external websites via the browser_read_page tool, which presents a surface for instructions hidden in web content to influence agent behavior.
Ingestion points: The browser_read_page tool is used across all core workflows (Search, Login, E-commerce) to retrieve content into the agent's context.
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat external content as untrusted data.
Capability inventory: The skill possesses extensive capabilities including browser_click, browser_type, file_write, file_read, schedule_create, and memory_store.
Sanitization: No sanitization or validation logic for external site content is specified in the prompt instructions.
[PRIVILEGE_ESCALATION]: The HAND.toml file includes environment setup instructions for Linux distributions that utilize sudo to install system-level binaries (python3 and chromium-browser). These are documented as standard administrative steps for dependency management.
[COMMAND_EXECUTION]: The skill utilizes browser automation tools that allow the agent to perform actions such as browser_click and browser_type on remote websites. This enables the agent to execute specific interactions within a web session based on its interpretation of site content.

browser-automation