The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to use Playwright to drive a browser and interact with live web applications. Content from these applications is untrusted and could potentially contain instructions designed to influence the agent's behavior.
Ingestion points: The agent interacts with live production URLs as described in the 'After deploy, test the live URL' section of SKILL.md.
Boundary markers: The instructions do not specify any delimiters or safety markers to differentiate between the agent's instructions and the content retrieved from the web browser.
Capability inventory: The skill uses browser automation (Playwright), can capture screenshots, and utilizes shell commands for environment setup and package installation.
Sanitization: No mechanisms for sanitizing or filtering the content retrieved from the browser are mentioned before the agent processes it.
[EXTERNAL_DOWNLOADS]: The skill mentions the installation of the @playwright/test package and the download of browser binaries (Chromium). These are well-known and standard components of the Playwright testing framework.

playwright