The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill uses a wrapper script run.js that dynamically generates JavaScript files and executes them.
Evidence: run.js captures input code (either from a file or a string argument), wraps it in a template, writes it to a temporary file in the skill directory (.temp-execution-${Date.now()}.js), and executes it using require(). While this is the primary mechanism for the automation functionality, it represents a runtime code generation and execution vector.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to navigate to and interact with external websites, which creates a surface for indirect prompt injection attacks.
Ingestion points: SKILL.md and lib/helpers.js provide patterns for navigating to URLs using page.goto(TARGET_URL) and reading page content (e.g., extractTableData, extractTexts).
Boundary markers: No explicit delimiters or boundary markers are suggested for processing content retrieved from external sites.
Capability inventory: The skill environment has network access, file system access (for screenshots and temporary script storage), and the ability to execute generated code.
Sanitization: No technical sanitization of external web content is implemented; the skill relies on a manual warning to the agent in SKILL.md regarding untrusted content.
[COMMAND_EXECUTION]: The skill instructions and scripts involve the execution of shell commands and network probing.
Evidence: SKILL.md instructs the execution of npm run setup and node run.js. Additionally, lib/helpers.js contains a detectDevServers function that performs automated network probing of common localhost ports (e.g., 3000, 3001, 8080) using the http module.

playwright-skill