Agent Browser Automation
Paste the following prompt into your AI chat to install this skill:
Please follow https://skillhub.cn/install/skillhub.md to install @user_a994e44e/agent-browser-plus.
About this skill
What Problem It Solves
agent-browser targets AI-agent browser automation where brittle selectors cause failures. It converts page state into inspectable snapshots and assigns refs such as @e1 and @e2, allowing agents to interact with elements by page-load references instead of hard-coded CSS or XPath expressions. It is useful for form submission, saved login state, tabs, screenshots, PDFs, cookie state, network inspection, and debugging.
How It Works
A typical workflow is:
- agent-browser open <url> to navigate to a page
- agent-browser snapshot -i to list interactive elements with refs
- Use the returned refs to click, fill, select, or otherwise operate on the page
- Re-run the snapshot after navigation or significant DOM changes
It also supports:
- State management for cookies and storage, including saved authentication flows
- Parallel sessions with isolated browser contexts
- JSON output for machine parsing
- Debugging controls such as headed mode, timeouts, and CDP connections
- Browser capabilities including screenshots, PDFs, video recording, mouse control, frames, dialogs, JavaScript, network, and tabs
Keep in mind:
- Refs change when the page reloads or navigates, so snapshot again after navigation
- Prefer fill over type to clear existing input text
- If an element is not found, inspect the snapshot first; if the page has not loaded, add a wait step
Use Cases
- QA engineers validate web forms by snapshotting elements and filling refs.
- Automation engineers save cookies and storage to reuse authenticated sessions.
- Product engineers inspect tabs, frames, and dialogs during UI acceptance.
- Data engineers export page info with JSON output for downstream parsing.
Best For
- QA engineers running web UI regression who need stable element selection.
- Product engineers doing automated acceptance who inspect tabs, dialogs, and frames.
- AI application engineers who need browser results as parseable JSON.
- Ops engineers needing saved cookies and storage for repeatable authenticated flows.
Related Skills
Run a grilling session to interact with or test AI agents.
A systematic prompt optimization skill that refines prompts using a four-step distillation framework (diagnose, structure, think, compress) and methodologies from four prompting masters.
Tencent Cloud OCR agent for K-12 exam grading with async full-sheet or single-question marking, handwritten answer recognition, correctness, knowledge points, and coordinates.
Quickly converts a user's input, list, or screenshot into a multi-page workbench, supporting template selection, custom builds, and responsive layouts.