The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands for setup and execution, including 'npm install' and 'npx playwright install'. These are standard procedures for initializing the Playwright environment and its required browser binaries.
[REMOTE_CODE_EXECUTION]: Browser binaries are downloaded from Microsoft's official Playwright registry during the setup process. Additionally, the 'run.js' script allows for the dynamic execution of generated automation scripts via Node.js.
[DYNAMIC_EXECUTION]: The 'run.js' wrapper implements a dynamic execution pattern by wrapping provided code in an async template and loading it using 'require()'. This is the intended functionality for custom web automation.
[DATA_INGESTION]: The skill interacts with untrusted external websites and extracts data (e.g., text, tables). This constitutes an attack surface for indirect prompt injection.
Ingestion points: Functions like 'extractTexts' and 'extractTableData' in 'lib/helpers.js' that pull content from the current page context.
Boundary markers: None implemented in the extraction helpers or script templates.
Capability inventory: The skill can execute shell commands ('run.js' setup), perform dynamic JavaScript execution ('run.js' main), and conduct network operations through the Playwright API.
Sanitization: No sanitization or escaping is performed on extracted web content before it is passed to the agent.

playwright-skill