The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the playwright-cli tool to interact with web browsers, as well as npx and npm for running tests and managing dependencies.- [REMOTE_CODE_EXECUTION]: Arbitrary JavaScript execution is supported through the run-code and eval commands, allowing for complex browser interactions and automation logic. This includes access to the automation runner's environment variables for handling secrets.- [DATA_EXFILTRATION]: The skill provides commands to read and save sensitive browser data, including cookie-get, localstorage-get, and state-save. These tools can be used to export session state and authentication tokens, though the skill provides warnings on how to handle this data securely.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection from external web data.
Ingestion points: Web page snapshots, DOM text, and console output are processed by the agent.
Boundary markers: The skill includes a 'Security rules' section explicitly instructing the agent to treat page content as data and not as instructions.
Capability inventory: Includes shell command execution, JavaScript execution, file writing, and network routing.
Sanitization: The skill provides instructions for the agent to manually identify and report suspicious embedded instructions rather than implementing automated sanitization.- [EXTERNAL_DOWNLOADS]: Instructions are provided to install automation tools from the official npm registry, which is a well-known service.

playwright-cli