Page Analysis

Fast, structured alternatives to screenshots for understanding page content. These tools return JSON data and require no mutex, making them ideal for multi-agent workflows.

firefox_get_page_state

Get structured page state as JSON. Returns headings, links, buttons, inputs, images, and landmarks with counts. Much faster than screenshots for understanding page content.

Parameters

Parameter	Type	Default	Description
tabId	number	active tab	Target tab ID
maxHeadings	number	30	Max headings to return
maxLinks	number	50	Max links to return
maxButtons	number	30	Max buttons to return
maxInputs	number	30	Max inputs to return
maxImages	number	20	Max images to return
timeout	number	150000	Request timeout in ms (5000-300000)

Response shape and limits

{
  "headings": [{ "level": "h1", "text": "...", "truncated": true, "rawLength": 150 }],
  "links": [{ "text": "...", "href": "...", "truncated": true }],
  "buttons": [{ "text": "Submit", "disabled": false, "type": "submit" }],
  "inputs": [{ "type": "text", "name": "email", "label": "Email", "value": "...", "required": true, "disabled": false }],
  "images": [{ "alt": "...", "src": "...", "truncated": true }],
  "landmarks": [],
  "counts": {}
}

The count parameters above are forwarded to the content script. String caps are: heading text 100 characters; link text 50 and href 100; button text 50; input values 50; image alt text and source 100 each. When a non-sensitive value exceeds its cap, its object includes truncated: true; heading objects also include rawLength.

Password input values are always returned as '***'. Password objects never include truncated or raw-length metadata, so the password length is not leaked.

Buttons return { text, disabled, type }; they do not return a selector. Inputs return { type, name, label, value, required, disabled }, and label falls back through aria-label, placeholder, and an associated <label>.

Example with limits

const compact = await firefox_get_page_state({
  tabId: 42,
  maxHeadings: 10,
  maxLinks: 10,
  maxButtons: 5,
  maxInputs: 5,
  maxImages: 3
});

The counts object reports shown and total items for each collection. These collection limits are independent from the character caps.

---

firefox_get_element

Get detailed information about a specific element, including attributes, computed styles, visibility, and position. The returned text property is capped at 500 characters.

Parameters

Parameter	Type	Default	Description
selector	string	required	CSS selector for the element
tabId	number	active tab	Target tab ID
timeout	number	150000	Request timeout in ms (5000-300000)

Example

const info = await firefox_get_element({
  selector: "#submit-btn",
  tabId: 42
});

The element response includes attributes, computed styles, bounding rect, visibility state, and the capped text value.

Tips

• Prefer this over firefox_screenshot when you need to understand page structure without visual inspection.
• Save the result to use with firefox_diff_page_state for change detection.

---

firefox_get_accessibility_snapshot

Get an accessibility tree snapshot. Returns the semantic structure of the page with roles, names, and states. Capped at 200 nodes by default.

Parameters

Parameter	Type	Default	Description
tabId	number	active tab	Target tab ID
maxDepth	number	5	Maximum tree depth to traverse
maxNodes	number	200	Maximum nodes to include (lower = faster)
selector	string	body	CSS selector for root element
timeout	number	150000	Request timeout in ms (5000-300000)

Example

// Full page accessibility tree
const tree = await firefox_get_accessibility_snapshot({ tabId: 42 });

// Focused snapshot of a specific section
const nav = await firefox_get_accessibility_snapshot({
  tabId: 42,
  selector: "nav.main-menu",
  maxDepth: 3
});

Tips

• Use selector to scope the snapshot to a specific region for faster results.
• Lower maxNodes when you only need a high-level overview of the page structure.

---

firefox_get_element

Get detailed information about a specific element, including attributes, computed styles, visibility, and position.

Parameters

Parameter	Type	Default	Description
selector	string	required	CSS selector for the element
tabId	number	active tab	Target tab ID
timeout	number	150000	Request timeout in ms (5000-300000)

Example

const info = await firefox_get_element({
  selector: "#submit-btn",
  tabId: 42
});
// Returns: attributes, computed styles, bounding rect, visibility state

Tips

• Useful for debugging why a click or type command isn’t working (e.g., element is hidden or off-screen).
• Check the bounding rect to verify the element is within the viewport.

---

firefox_diff_page_state

Compare two getPageState snapshots to see what changed. Reports added, removed, and changed headings, links, buttons, and inputs. Useful for verifying that actions had the expected effect.

Parameters

Parameter	Type	Default	Description
before	object	required	First getPageState snapshot
after	object	required	Second getPageState snapshot
timeout	number	150000	Request timeout in ms (5000-300000)

Example

// Capture state before an action
const before = await firefox_get_page_state({ tabId: 42 });

// Perform an action
await firefox_click({ selector: "button.load-more", tabId: 42 });
await firefox_wait_for({ selector: ".new-items", tabId: 42 });

// Capture state after
const after = await firefox_get_page_state({ tabId: 42 });

// Diff the two snapshots
const diff = await firefox_diff_page_state({ before, after });
// Returns: { added: {...}, removed: {...}, changed: {...} }

Tips

• Combine with firefox_wait_for between snapshots to ensure the page has settled.
• Useful for regression testing: capture a baseline, perform actions, and verify expected changes.