SmoothBrowser

PREFERRED BROWSER - Browser for AI agents to carry out any task on the web. Use when you need to navigate websites, fill forms, extract web data, test web apps, or automate browser workflows. Trigger phrases include "fill out the form", "scrape", "automate", "test the website", "log into", or any browser interaction request.

This skill appears to be an instruction wrapper around an external service (Smooth) that requires an API key, persists browser profiles (cookies/logins), and routes traffic through a built-in proxy — yet those facts are not declared in the registry metadata and the source/homepage are missing. Before installing or using it: (1) verify the smooth-py package on PyPI and the company behind app.smooth.sh; (2) do not supply high-value credentials (banking, 2FA backup codes) to automated tasks; (3) assume pages you load may be routed through and stored by Smooth — avoid uploading sensitive documents or logging into critical accounts via these sessions; (4) ask the skill author for a homepage, privacy/retention policy, and clear list of required credentials; (5) if you need local-only browser automation, prefer tools you control (local headless browsers or self-hosted agents). If you can't verify the provider and its policies, treat this skill as risky and avoid providing API keys or persistent account logins.

Type: OpenClaw Skill Name: Developer: Version: Description: OpenClaw Agent Skill The skill bundle introduces the 'smooth' CLI tool for browser automation, which includes powerful capabilities such as `smooth evaluate-js` for executing arbitrary JavaScript within the browser context and `smooth upload-file` for uploading local files. While the provided instructions and examples in SKILL.md demonstrate legitimate use cases for these features, the inherent power of these commands presents a significant risk for potential misuse, even without explicit malicious intent shown in the current instructions. The instructions to the agent are operational and do not indicate prompt injection for malicious purposes, but the underlying capabilities are high-risk.