The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill instructs the user to disable security protections by starting the agent with --sandbox danger-full-access. This removes the execution environment's restrictions, granting the agent full access to the host system.
[COMMAND_EXECUTION]: The skill provides instructions to execute shell commands for environment setup and dependency management, including npm install playwright, npm install --save-dev electron, and verification via node -e.
[DYNAMIC_EXECUTION]: The skill's core functionality relies on the js_repl tool to execute arbitrary JavaScript code. It uses page.evaluate(), electronApp.evaluate(), and appWindow.evaluate() to run scripts within the context of the target application, which can lead to unintended actions if interacting with malicious content.
[EXTERNAL_DOWNLOADS]: The setup process involves downloading and installing third-party packages and browser binaries via npm and npx during runtime.
[INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to instructions embedded in external content it processes.
Ingestion points: External URLs and web content loaded via page.goto() and mobilePage.goto() in SKILL.md.
Boundary markers: None identified. The instructions do not specify delimiters or warnings for the agent to ignore instructions found on the web pages.
Capability inventory: The skill can execute shell commands (npm install), perform network operations (page.goto), and execute arbitrary code via js_repl and evaluate calls across multiple files.
Sanitization: No sanitization or validation of the content retrieved from external URLs is specified before the agent processes or renders it.

playwright-interactive