Navigate to a URL and get page content as structured markdown (default), plain text, or raw HTML

Navigate back to the previous page

Scroll the page up or down

Switch to a different browser tab

Wait for an element or timeout

Click on an element by CSS selector

Click at specific viewport coordinates (x, y). Use for canvas, maps, SVGs, or elements without stable CSS selectors

Fill a form field with a value

Select a dropdown option by value

Hover over an element

Press a keyboard key

Execute JavaScript on the page

Get comprehensive page structure: headings, landmarks, forms, interactive elements, and links

Extract text content from a specific page section by heading name or CSS selector

List page resources (links, images, forms)

Capture a screenshot of the current page or a specific element

Save a file to the output directory

Generate and execute a deterministic multi-step script without per-step LLM round-trips

Check the status of a running or completed script

Wait for one or more scripts to finish and collect their results

Abort a running script immediately

Save a script definition to disk for later reuse

List all saved scripts

Read a saved script definition

Execute a high-level crawl goal autonomously. The agent plans, navigates, and extracts data using its own LLM loop. Returns structured results when done. Use this for complex multi-page tasks; use individual tools (navigate, click, etc.) for fine-grained control.