The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a bash wrapper script (scripts/playwright_cli.sh) to execute browser automation commands via the system shell.
[EXTERNAL_DOWNLOADS]: Uses npx to dynamically download and run the @playwright/cli package from the npm registry. This package is part of the well-known Playwright project maintained by Microsoft.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it interacts with and processes untrusted data from external websites.
Ingestion points: Untrusted web content is ingested through the open and snapshot commands in SKILL.md and references/cli.md.
Boundary markers: No explicit delimiters or boundary markers are used to separate web content from agent instructions.
Capability inventory: The skill can execute shell commands, perform network requests through the browser, and write files to the local system (e.g., screenshots and traces).
Sanitization: No sanitization or filtering of browser-rendered content is mentioned prior to processing element snapshots.
[REMOTE_CODE_EXECUTION]: The skill exposes eval and run-code functionality in references/cli.md, which allows for the execution of arbitrary JavaScript code strings within the automated browser context.

playwright