The Agent Skills Directory

[Indirect Prompt Injection Surface]: The skill navigates and crawls content from external web pages (up to a 25-page cap). While this is necessary for its testing purpose, it creates a surface where content on a tested site could potentially contain instructions intended to influence the agent. The skill follows a structured procedural flow to mitigate this risk.
[Dynamic Script Execution]: The skill uses browser_evaluate to run JavaScript within the browser context to extract links and inspect API response shapes. It also utilizes node -e shell commands to write structured JSON reports to the local filesystem. These operations are scoped to the skill's defined testing and reporting functionality.
[Local Script Invocation]: The skill executes several utility scripts located within the plugin's local directory (e.g., check-version.js, check-activation-status.js). These are standard helper tools for managing the Power Pages testing environment.
[Sensitive Data Handling]: The skill includes explicit instructions for handling site verification tokens during API replay testing. It mandates that tokens be kept in memory and never recorded in logs or reports, demonstrating a security-conscious approach to credential handling.
[User Consent for Operations]: Critical actions such as logging into a private site or submitting forms that might modify environment data are gated by explicit user confirmation prompts, ensuring the user remains in control of the agent's actions.

test-site