refactor: remove outdated browser automation guide and add comprehensive browser automation rules documentation

- Deleted the obsolete browser-automation-guide.mdc file.
- Introduced a new browser-automation.rules.mdc file containing detailed instructions and examples for using the browser-use CLI tool, including key commands, common options, and provider-model references.
- Updated the documentation to enhance clarity and usability for users performing browser automation tasks.

Author: drumnation

.cursor/rules/browser-automation-guide.mdc

@@ -0,0 +1,82 @@
+---
+description: Guide for using the browser-use CLI tool. Request when performing browser automation tasks.
+globs:
+alwaysApply: false
+---
+# Browser Automation (browser-use) Guide
+# Last Updated: 2025-03-31 10:13:02 AM
+
+## Browser Automation with browser-use
+
+The browser-use CLI tool allows you to control a browser using natural language instructions.
+
+### Key Commands
+
+1. **Start a browser session**:
+
+   ```bash
+   browser-use start
+   ```
+
+2. **Run a task in an existing browser**:
+
+   ```bash
+   browser-use run "your task instruction" --url "https://example.com"
+   ```
+
+3. **Close the browser session**:
+
+   ```bash
+   browser-use close
+   ```
+
+### Common Options
+
+- `--provider` - Choose LLM provider: `Deepseek` (default), `Google`, `OpenAI`, `Anthropic`
+- `--vision` - Enable vision capabilities (automatically selects vision-capable models)
+- `--headless` - Run browser in headless mode
+- `--record` - Enable session recording
+- `--record-path` - Path to save recordings
+- `--max-steps` - Maximum number of steps per task (default: 10)
+- `--max-actions` - Maximum actions per step (default: 1)
+- `--add-info` - Additional context for the agent
+
+### Examples
+
+1. **Basic browsing task**:
+
+   ```bash
+   browser-use run "search for OpenAI" --url "https://www.google.com"
+   ```
+
+2. **Visual analysis with Google Gemini**:
+
+   ```bash
+   browser-use run "analyze the visual layout" --url "https://www.openai.com" --provider Google --vision
+   ```
+
+3. **Complex analysis with OpenAI**:
+
+   ```bash
+   browser-use run "analyze the layout and design" --url "https://www.example.com" --provider OpenAI --vision
+   ```
+
+4. **Debugging with recording**:
+
+   ```bash
+   browser-use run "test the login process" --url "https://example.com" --record --record-path ./debug_session
+   ```
+
+### Provider-Model Reference
+
+- **Deepseek**: `deepseek-chat`
+- **Google**: `gemini-1.5-pro` (default), `gemini-2.0-flash` (model-index 1)
+- **OpenAI**: `gpt-4o` (vision-capable)
+- **Anthropic**: `claude-3-5-sonnet-latest` (default), `claude-3-5-sonnet-20241022` (model-index 1)
+
+### Tips
+
+- Always include the full URL with protocol (https://)
+- For complex tasks, increase the `--max-steps` parameter
+- Use `--add-info` to provide additional context not covered in the main prompt
+- Enable `--vision` when visual analysis is needed