One-call status: version, sessions, scope, providers, limits.

Local usage stats: tool call counts, per-day breakdown, totals.

Purely local counters (~/.navigator/usage.json) for your own insight — no URLs or page content are ever recorded. Disable with NAVIGATOR_STATS=false.

Navigate the active (or specified) tab to a URL.

Returns the new page state; includes a fresh snapshot when the page content changed (configurable via NAVIGATOR_AUTO_SNAPSHOT).

Go back one history step in the tab.

Go forward one history step in the tab.

Reload the current page.

Wait until text appears, a selector matches, or a fixed time passes.

Return the current URL and title of the tab.

Click an element by its ref from the snapshot (e.g. e12).

Uses human-like mouse movement when the session has humanization enabled.

Fallback click by visible text (use when no ref is available).

Fill a field by ref. Set submit=true to press Enter afterwards.

Fill multiple fields in one call — much faster than one fill per field.

Each item: {"ref": "e5", "value": "text"}. Optionally submits the form (presses Enter on the last field).

Select an option in a dropdown (combobox) by ref. Value = option label.

Hover over an element by ref (reveals menus/tooltips).

Press a key (e.g. Enter, Escape, Tab, ArrowDown, Control+a).

Drag element ref_a onto element ref_b.

Upload a local file. Ref may point to the file input or its trigger button.

Scroll the page. Directions: down, up, top, bottom. Optional pixel amount.

Capture the page as a compact, ref-annotated accessibility snapshot.

Elements get stable refs (e.g. [ref=e12]) that interaction tools accept. If the page is unchanged since the last snapshot, returns a short "unchanged" notice instead of re-sending the tree (major token savings). Set force=true to always re-render, deep=true to include child frames.

Search the current page for elements matching a text pattern (regex ok) and/or an ARIA role. Returns matching elements with their refs — cheaper than a full snapshot when you know what you're looking for.

By default only actionable elements (links, buttons, fields, headings, images) are searched; set include_structural=true to also match containers (banner/navigation/form/...).

Extract text — whole page (smart main-content) or a specific element.

Supports pagination via offset/limit for very long pages.

Extract HTML — whole document or a specific element (by ref).

List all links (deduplicated) with their resolved hrefs.

List forms and their fields — the map for browser_fill_form.

Extract content as row arrays (token-cheap vs HTML).

Extract page metadata: title, meta tags, OpenGraph, canonical.

Read console messages and page errors (JS errors, logs, failed asserts).

Set drain=true to clear the buffer after reading.

Read captured alert/confirm/prompt dialogs.

List the tabs of a session with their URLs.

Open a new tab (optionally navigating it to a URL) and make it active.

Make another tab active (subsequent tools act on it).

Close a tab (the active one cannot be the last tab).

List all active sessions with their tabs, fingerprints and proxies.

Create a new isolated browser session.

• stealth: apply anti-detection fingerprinting (default on)

• profile: restore a saved auth profile (from browser_session_save_auth)

• proxy: per-session proxy server (e.g. http://user:pass@host:port)

• fingerprint: specific fingerprint name (see stealth docs), random by default

• bypass_csp: disable CSP enforcement (for pages that block inline scripts)

• record_har: file path to record a HAR archive of all traffic

Close a session and all its tabs (cookies/storage are discarded unless saved with browser_session_save_auth).

Persist the session's cookies + localStorage to a named profile.

Restore later via browser_session_new(profile=profile_name) — survives restarts, perfect for staying logged in across engagements.

Take a screenshot and return it as an image (for vision-capable LLMs).

area: "viewport" (visible part), "full" (entire page), or "element" (requires ref). format: png or jpeg (jpeg is much smaller — prefer it for large pages).

Save the current page as a PDF (Chromium only).

Start recording network requests on this tab.

Optional filters: resource_types (document/xhr/fetch/script/image/...), url_contains (substring), methods (GET/POST/...). Default records everything.

Stop recording network requests (buffer is kept until next start).

List captured network requests (most recent first).

Fetch full details + response body of a captured request (by index).

Block requests matching URL globs across all sessions.

Examples: ["/*.css", "/ads.js", "doubleclick"]. Blocks ads and trackers to speed up automation dramatically.

Remove all URL blocking rules.

Discover background API endpoints (fetch/XHR) used by the page — including internal/undocumented APIs, from capture or live observation.

Set dialog handling: "auto" (accept + record — default) or "manual" (pause dialogs so browser_dialog_respond can answer them).

Respond to a pending manual-mode dialog (accept/dismiss, optional text).

Evaluate JavaScript in the page and return the result.

Prefer a function expression: "() => document.title" or "(x) => fetch(x).then(r => r.status)" with arg passed as parameter. Everything the page can do is allowed — this is raw power, use responsibly.

Detect captcha widgets on the current page (type, evidence, sitekeys).

Solve the captcha on the current page automatically.

Needs a provider API key (TWOCAPTCHA_API_KEY or CAPSOLVER_API_KEY). Supports reCAPTCHA v2, hCaptcha, Turnstile and plain image captchas (via ref pointing at the image element).

Wait for a human to solve the captcha (requires a visible browser: NAVIGATOR_HEADLESS=false or VNC). Polls until the captcha disappears.