domain

Documentation

Overview

Package domain holds the browser-use bounded context's contracts: the driver interface the tools consume and the wire/result types it produces. It is pure - stdlib imports only.

Index

• type BrowserDriver
• type BrowserScreenshotResult
• type BrowserTab
• type BrowserToolResult

Types

type BrowserDriver

type BrowserDriver interface {
	Navigate(ctx context.Context, url string) (BrowserToolResult, error)
	Click(ctx context.Context, selector string) (BrowserToolResult, error)
	// ClickAt clicks at viewport coordinates (CSS pixels), for use with a
	// screenshot. Not every backend supports it (the extension bridge does not).
	ClickAt(ctx context.Context, x, y float64) (BrowserToolResult, error)
	Type(ctx context.Context, selector, text string, pressEnter bool) (BrowserToolResult, error)
	// Read returns element text in Content and drained browser events in Events.
	// Sensitive input values (passwords, tokens) MUST be redacted before return.
	Read(ctx context.Context, selector string) (BrowserToolResult, error)
	// Screenshot captures the current page/active tab as an image.
	Screenshot(ctx context.Context) (BrowserScreenshotResult, error)
	// Tabs lists the open tabs, marking the active one.
	Tabs(ctx context.Context) ([]BrowserTab, error)
	Close()
}

BrowserDriver executes browser-use verbs against a browser backend: a Playwright-launched browser, or the user's real browser via the opentask extension bridge.

type BrowserScreenshotResult

type BrowserScreenshotResult struct {
	Data     string `json:"-"`
	MimeType string `json:"mime_type"`
	URL      string `json:"url,omitempty"`
	Title    string `json:"title,omitempty"`
	Width    int    `json:"width,omitempty"`
	Height   int    `json:"height,omitempty"`
}

BrowserScreenshotResult carries a screenshot of the current page. Data is base64-encoded image bytes; the BrowserScreenshot tool persists them and sets the attachment's on-disk source path.

type BrowserTab

type BrowserTab struct {
	Index  int    `json:"index"`
	URL    string `json:"url"`
	Title  string `json:"title"`
	Active bool   `json:"active"`
}

BrowserTab describes one open tab/page as reported by BrowserTabs. Active marks the tab the browser-use verbs currently act on.

type BrowserToolResult

type BrowserToolResult struct {
	Action   string   `json:"action"`
	URL      string   `json:"url,omitempty"`
	Title    string   `json:"title,omitempty"`
	Selector string   `json:"selector,omitempty"`
	Text     string   `json:"text,omitempty"`
	Content  string   `json:"content,omitempty"`
	Events   []string `json:"events,omitempty"`
}

BrowserToolResult represents the result of a browser use operation. One shared shape for all browser tools; each tool fills the fields it produces.