The Agent Skills Directory

[DYNAMIC_EXECUTION]:
The skill documentation and reference files (specifically references/running-code.md) promote the use of playwright-cli run-code and playwright-cli eval to execute arbitrary JavaScript code within the browser context to handle complex workflows like geolocation, permission management, and custom scraping.
These commands allow the execution of logic generated at runtime, which is a powerful capability necessary for the skill's primary function but inherently risky if misused.
[INDIRECT_PROMPT_INJECTION]:
The skill is designed to navigate to and process content from external, untrusted web pages through commands like playwright-cli goto and playwright-cli snapshot.
Ingestion points: Data enters the agent's context via DOM snapshots (playwright-cli snapshot), console logs (playwright-cli console), and network activity logs (playwright-cli requests) as described in SKILL.md and references/tracing.md.
Boundary markers: The skill documentation uses standard markdown and YAML formatting to present browser data to the agent, providing structure but not preventing the agent from interpreting embedded instructions.
Capability inventory: The skill possesses powerful capabilities including arbitrary JavaScript execution (run-code), file writing (state-save, screenshot, pdf), and active browser interaction (click, fill, type).
Sanitization: The instructions do not explicitly guide the agent to sanitize or ignore instructions potentially embedded within the web content it retrieves.
[EXTERNAL_DOWNLOADS]:
The SKILL.md file provides instructions for installing the Playwright CLI using npm install -g @playwright/cli@latest, which fetches resources from the official NPM registry.
[COMMAND_EXECUTION]:
The skill extensively utilizes shell commands such as playwright-cli, npx playwright test, and npm to perform browser automation and testing tasks, as specified in the allowed-tools metadata.

playwright-cli