The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes Playwright for complex browser automation, specifically simulating natural human behavioral biometrics (e.g., Bezier-curve mouse movements, variable typing delays, and acceleration-based scrolling) to evade security detection systems.
[REMOTE_CODE_EXECUTION]: Employs dynamic JavaScript injection within the browser context (playwright_run_code) to perform status code spoofing and intercept authentication responses, allowing for the simulation of successful 2FA bypasses.
[CREDENTIALS_UNSAFE]: Includes a 'CredentialManager' utility that stores usernames, passwords, and 2FA secrets in a local hidden file (.credentials). Although it attempts to manage file permissions and gitignore entries, storing plaintext credentials locally is a significant risk.
[EXTERNAL_DOWNLOADS]: Recommends installing multiple external dependencies for browser control, IMAP email access, and OCR-based CAPTCHA solving, including playwright, imap, and tesseract.js.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it navigates to and analyzes form requirements from untrusted external websites. Ingestion points: Web form analysis (SKILL.md); Capability inventory: Subprocess calls via Playwright, IMAP network operations, and file-write operations; Sanitization: No explicit sanitization or boundary markers are documented for processed web content.

authenticating