opencli-browser

The opencli-browser skill enables an agent to control a live Chrome browser directly through the OpenCLI interface. It allows automated interaction with web pages, including inspecting page content, filling out forms, clicking through complex or logged-in workflows, and extracting data on demand. This skill addresses the challenge of programmatically navigating dynamic web pages while maintaining reliability and accuracy in element selection, making it ideal for tasks that require precise, real-time browser manipulation without relying on static adapters.

Key features of opencli-browser include a selector-first target contract, where each command targets a specific element via numeric references or CSS selectors, and a structured envelope response that reports matches and confidence levels. It supports compound form fields, handling of stale references, and network activity capture. Additional capabilities include managing tab leases to keep browser sessions alive, binding to existing user tabs for logged-in workflows, and controlling foreground visibility and live inspection modes. The skill emphasizes reliability with detailed feedback on element matching and DOM changes, helping agents navigate even in complex or evolving web environments.

This skill is primarily intended for agents needing direct browser automation rather than developers creating reusable adapters. Typical use cases include automating tasks on web applications with login flows, scraping structured data from interactive pages, or performing ad-hoc browser inspections for debugging or analysis. Users are expected to meet prerequisites such as having Chrome running with the necessary OpenCLI extensions and ensuring proper tab access. OpenCLI agents, automation engineers, and developers seeking precise, programmatic control of live web sessions are the main target audience for this tool.