web-research

Files (3)

• SKILL.md
• manifest.json
• model-card-research.md · references/

SKILL.md (first lines)

# Skill: Web Research

## Overview
Browser automation tools for web research, scraping, and navigation using Playwright.

## Workflow
1. **Model-card gate (required when it applies):** if the research target is any ML model, checkpoint, adapter, or hosted endpoint — LLM or vision-language, diffusion/image/video/audio, embedding, or reranking — read [references/model-card-research.md](references/model-card-research.md) and apply its protocol before drawing conclusions or making recommendations
2. Navigate to a URL
3. Take a screenshot to verify the page loaded
4. Interact with page elements as needed
5. Extract information from the rendered page
6. Close the browser when done

## Tools

### browser_navigate
Open a URL in the browser.
```bash
gearcore call playwright browser_navigate '{"url": "https://example.com"}'
```

### browser_screenshot
Capture the current page state.
```bash
gearcore call playwright browser_screenshot '{}'
```

### browser_click
Click a page element using CSS selector.
```bash
gearcore call playwright browser_click '{"selector": "#submit-btn"}'
```

### browser_type
Type text into a page element.
```bash
gearcore call playwright browser_type '{"selector": "#search-input", "text": "query"}'
```

### browser_tab_list
List open browser tabs.
```bash
gearcore call playwright browser_tab_list '{}'
```

### browser_close
Close the browser to free resources.
```bash
gearcore call playwright browser_close '{}'
```

## Best Practices
- Take a screenshot after navigation to verify the page loaded correctly
- Use CSS selectors for reliable element targeting
- Close browser tabs when done to free resources
- Be respectful of rate limits and robots.txt

From BlinkVoid/GearCore, MIT.

More from BlinkVoid/GearCore