agent.yaml

apiVersion: adl.dev/v1
kind: Agent
metadata:
  name: browser-agent
  description: AI agent for browser automation and web testing using Playwright
  version: 0.2.1
spec:
  capabilities:
    streaming: true
    pushNotifications: false
    stateTransitionHistory: false
  config:
    browser:
      user_agent: "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36"
      viewport_width: 1920
      viewport_height: 1080
      header_accept: "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.7"
      header_accept_language: "en-US,en;q=0.9"
      header_accept_encoding: "gzip, deflate, br"
      header_dnt: "1"
      header_connection: "keep-alive"
      header_upgrade_insecure_requests: "1"
      data_dir: "/tmp/playwright/artifacts"
      args:
        - "--disable-blink-features=AutomationControlled"
        - "--disable-features=VizDisplayCompositor"
        - "--no-first-run"
        - "--disable-default-apps"
        - "--disable-extensions"
        - "--disable-plugins"
        - "--disable-sync"
        - "--disable-translate"
        - "--hide-scrollbars"
        - "--mute-audio"
        - "--no-zygote"
        - "--disable-background-timer-throttling"
        - "--disable-backgrounding-occluded-windows"
        - "--disable-renderer-backgrounding"
        - "--disable-ipc-flooding-protection"
  skills:
    - id: navigate_to_url
      name: navigate_to_url
      description: Navigate to a specific URL and wait for the page to fully load
      tags:
        - navigation
        - browser
        - playwright
      schema:
        type: object
        properties:
          url:
            type: string
            description: The URL to navigate to
          wait_until:
            type: string
            description: When to consider navigation succeeded (domcontentloaded, load, networkidle)
            default: load
          timeout:
            type: integer
            description: Maximum navigation timeout in milliseconds
            default: 30000
        required:
          - url
      inject:
        - logger
        - playwright
    - id: click_element
      name: click_element
      description: Click on an element identified by selector, text, or other locator strategies
      tags:
        - interaction
        - click
        - playwright
      schema:
        type: object
        properties:
          selector:
            type: string
            description: CSS selector, XPath, or text to identify the element
          click_count:
            type: integer
            description: Number of times to click
            default: 1
          button:
            type: string
            description: Mouse button to use (left, right, middle)
            default: left
          force:
            type: boolean
            description: Force click even if element is not visible
            default: false
          timeout:
            type: integer
            description: Maximum time to wait for element in milliseconds
            default: 30000
        required:
          - selector
      inject:
        - logger
        - playwright
    - id: fill_form
      name: fill_form
      description: Fill form fields with provided data, handling various input types
      tags:
        - form
        - input
        - automation
        - playwright
      schema:
        type: object
        properties:
          fields:
            type: array
            items:
              type: object
              properties:
                selector:
                  type: string
                  description: Selector for the form field
                value:
                  type: string
                  description: Value to fill in the field
                type:
                  type: string
                  description: Type of input (text, select, checkbox, radio)
              required:
                - selector
                - value
            description: List of form fields to fill
          submit:
            type: boolean
            description: Whether to submit the form after filling
            default: false
          submit_selector:
            type: string
            description: Selector for the submit button if submit is true
        required:
          - fields
      inject:
        - logger
        - playwright
    - id: extract_data
      name: extract_data
      description: Extract data from the page using selectors and return structured information
      tags:
        - scraping
        - extraction
        - data
        - playwright
      schema:
        type: object
        properties:
          extractors:
            type: array
            items:
              type: object
              properties:
                name:
                  type: string
                  description: Name for the extracted data field
                selector:
                  type: string
                  description: CSS selector or XPath to extract data from
                attribute:
                  type: string
                  description: Attribute to extract (text, href, src, etc.)
                  default: text
                multiple:
                  type: boolean
                  description: Extract all matching elements or just the first
                  default: false
              required:
                - name
                - selector
            description: List of data extractors to run
          format:
            type: string
            description: Output format (json, csv, text)
            default: json
        required:
          - extractors
      inject:
        - logger
        - playwright
    - id: take_screenshot
      name: take_screenshot
      description: Capture a screenshot of the current page or specific element
      tags:
        - screenshot
        - capture
        - visual
        - playwright
      schema:
        type: object
        properties:
          full_page:
            type: boolean
            description: Capture the entire scrollable page
            default: false
          selector:
            type: string
            description: Optional selector to screenshot specific element
          type:
            type: string
            description: Image format (png, jpeg)
            default: png
          quality:
            type: integer
            description: Quality for jpeg images (0-100)
            default: 80
        required: []
      inject:
        - logger
        - playwright
    - id: execute_script
      name: execute_script
      description: Execute custom JavaScript code in the browser context
      tags:
        - javascript
        - execution
        - custom
        - playwright
      schema:
        type: object
        properties:
          script:
            type: string
            description: JavaScript code to execute
          args:
            type: array
            items:
              type: string
            description: Arguments to pass to the script
          return_value:
            type: boolean
            description: Whether to return the script execution result
            default: true
        required:
          - script
      inject:
        - logger
        - playwright
    - id: handle_authentication
      name: handle_authentication
      description: Handle various authentication scenarios including basic auth, OAuth, and custom login forms
      tags:
        - authentication
        - login
        - security
        - playwright
      schema:
        type: object
        properties:
          type:
            type: string
            description: Authentication type (basic, form, oauth)
          username:
            type: string
            description: Username or email for authentication
          password:
            type: string
            description: Password for authentication
          login_url:
            type: string
            description: URL of the login page for form authentication
          username_selector:
            type: string
            description: Selector for username field in form authentication
          password_selector:
            type: string
            description: Selector for password field in form authentication
          submit_selector:
            type: string
            description: Selector for submit button in form authentication
        required:
          - type
      inject:
        - logger
        - playwright
    - id: wait_for_condition
      name: wait_for_condition
      description: Wait for specific conditions before proceeding with automation
      tags:
        - wait
        - synchronization
        - timing
        - playwright
      schema:
        type: object
        properties:
          condition:
            type: string
            description: Type of condition (selector, navigation, function, timeout, networkidle)
          selector:
            type: string
            description: Selector to wait for if condition is 'selector'
          state:
            type: string
            description: State to wait for (visible, hidden, attached, detached)
            default: visible
          timeout:
            type: integer
            description: Maximum time to wait in milliseconds
            default: 30000
          custom_function:
            type: string
            description: Custom JavaScript function to evaluate for 'function' condition
        required:
          - condition
      inject:
        - logger
        - playwright
    - id: write_to_csv
      name: write_to_csv
      description: Write structured data to CSV files with support for custom headers and file paths
      tags:
        - export
        - csv
        - data
        - file
      schema:
        type: object
        properties:
          data:
            type: array
            items:
              type: object
            description: Array of objects to write to CSV, each object represents a row
          filename:
            type: string
            description: Name of the CSV file (without path, will be saved to configured data directory)
          headers:
            type: array
            items:
              type: string
            description: Custom column headers for the CSV file (optional, will use object keys if not provided)
          append:
            type: boolean
            description: Whether to append to existing file or create new file
            default: false
          include_headers:
            type: boolean
            description: Whether to include headers in the CSV output
            default: true
        required:
          - data
          - filename
      inject:
        - logger
        - playwright
  agent:
    provider: ""
    model: ""
    systemPrompt: |
      You are an expert Playwright browser automation assistant. Your primary role is to help users automate web browser tasks efficiently and reliably.
      
      Your core capabilities include:
      1. **Web Navigation**: Navigate to URLs, handle redirects, and manage page loads
      2. **Element Interaction**: Click buttons, fill forms, select dropdowns, and interact with any web element
      3. **Data Extraction**: Scrape and extract structured data from web pages
      4. **Form Automation**: Fill and submit complex forms with validation
      5. **Screenshot Capture**: Take full-page or element-specific screenshots
      6. **JavaScript Execution**: Run custom scripts in the browser context
      7. **Authentication Handling**: Manage various authentication methods
      8. **Synchronization**: Wait for specific conditions and handle dynamic content
      
      Key expertise areas:
      - Modern web technologies (SPA, dynamic content, AJAX)
      - Selector strategies (CSS, XPath, text, accessibility)
      - Browser automation best practices
      - Error handling and retry mechanisms
      - Cross-browser compatibility (Chromium, Firefox, WebKit)
      - Performance optimization for automation scripts
      - Handling pop-ups, alerts, and iframes
      - File uploads and downloads
      - Network interception and modification
      - Mobile and responsive testing
      
      When helping users:
      - Always use robust selectors that won't break easily
      - Implement proper wait strategies for dynamic content
      - Handle errors gracefully with informative messages
      - Suggest efficient approaches for the task
      - Consider accessibility and best practices
      - Provide clear explanations of automation steps
      - Optimize for speed while maintaining reliability
      
      Your automation solutions should be maintainable, efficient, and production-ready.
  services:
    playwright:
      type: service
      interface: BrowserAutomation
      factory: NewPlaywrightService
      description: Playwright service for browser automation and web testing
  server:
    port: 8080
    debug: false
  language:
    go:
      module: github.com/inference-gateway/browser-agent
      version: "1.25"
  acronyms:
    - DOM
    - CSS
    - XPath
    - URL
    - API
    - SPA
    - AJAX
  scm:
    provider: github
    url: https://github.com/inference-gateway/browser-agent
    github_app: true
    issue_templates: true
  sandbox:
    flox:
      enabled: true
  artifacts:
    enabled: true