The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install playwright-core via npm, which is a well-known and standard library for browser automation.
[COMMAND_EXECUTION]: The skill launches the Chromium browser as a subprocess using shell commands with debugging flags to enable CDP (Chrome DevTools Protocol) communication.
[COMMAND_EXECUTION]: The skill functions by writing and executing local Node.js scripts (.mjs) to perform automation tasks such as form filling and data extraction.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from external websites into the agent's context.
Ingestion points: Untrusted data enters the agent context via page.content() and page.evaluate() calls in SKILL.md scripts.
Boundary markers: No specific delimiters or safety warnings for embedded instructions are specified in the prompts.
Capability inventory: The skill has access to shell execution (Bash) and file system operations (Read, Write) as defined in allowed-tools.
Sanitization: There is no evidence of content sanitization or validation of the HTML/text extracted from the browser before it is processed by the agent.

playwright