GCU Browser Automation

All GCU browser tools drive a real Chrome instance through the Beeline extension and Chrome DevTools Protocol (CDP). That means clicks, keystrokes, and screenshots are processed by the actual browser's native hit testing, focus, and layout engines — not a synthetic event layer. Understanding this unlocks strategies that make hard sites easy.

Coordinates

Every browser tool that takes or returns coordinates operates in fractions of the viewport (0..1 for both axes). Read a target's proportional position off browser_screenshot — "this button is about 35% from the left and 20% from the top" → pass (0.35, 0.20). Rect-returning tools (browser_get_rect, browser_shadow_query, and the rect inside focused_element) also return fractions. The tools convert to CSS pixels internally before dispatching to Chrome.

browser_screenshot()                  → image + cssWidth/cssHeight in meta
browser_click_coordinate(x, y)        → x, y are fractions 0..1
browser_hover_coordinate(x, y)        → fractions
browser_press_at(x, y, key)           → fractions
browser_get_rect(selector) → rect     → rect.cx / rect.cy are fractions
browser_shadow_query(...)  → rect     → same

Why fractions: every vision model (Claude ~1.15 MP target, GPT-4o 512-px tiles, Gemini, local VLMs) resizes or tiles images differently before the model sees the pixels. Proportions survive every such transform; pixel coordinates only "work" per-model and silently break when you swap backends. Four-decimal precision (0.0001 ≈ 0.17 CSS px on a 1717-wide viewport) is more than enough for the tightest targets.

Exception for zoomed elements: pages that use zoom or transform: scale() on a container (LinkedIn's #interop-outlet, some embedded iframes) render in a scaled local coordinate space. getBoundingClientRect there may not match CDP's hit space. Prefer browser_shadow_query (which handles the math and returns fractions) or visually pick coordinates from a screenshot. Avoid raw browser_evaluate + getBoundingClientRect() for coord lookup — that returns CSS px and will be wrong when fed to click tools.

Screenshot + coordinates is shadow-agnostic — prefer it on shadow-heavy sites

Start with browser_snapshot when you need to inspect the page structure or find ordinary controls. If the snapshot does not show the thing you need, shows stale or misleading refs, or cannot prove where a visible target is, take browser_screenshot and use the screenshot + coordinate path. This is especially useful on sites that use Shadow DOM heavily

Why:

• CDP hit testing walks shadow roots natively. browser_click_coordinate(x, y) routes through Chrome's native hit tester, which traverses open shadow roots automatically. You don't need to know the shadow structure.
• Keyboard dispatch follows focus into shadow roots. After a click focuses an input (even one three shadow levels deep), browser_press(...) with no selector dispatches keys to document.activeElement's computed focus target.
• Screenshots render the real layout regardless of DOM implementation.

Whereas wait_for_selector, browser_click(selector=...), browser_type(selector=...) all use document.querySelector under the hood, which stops at shadow boundaries. They cannot see elements inside shadow roots. For shadow-DOM inputs, use browser_type_focused after focusing via click-coordinate.

Recommended workflow on shadow-heavy sites

1. browser_screenshot() → JPEG; meta includes cssWidth/cssHeight for reference.
2. Identify the target visually → estimate its proportional position (fx, fy) where each is in 0..1.
3. browser_click_coordinate(fx, fy) → tool converts to CSS px and dispatches; CDP native hit testing focuses the element. The response includes `focused_element: {tag, id, role, contenteditable, rect, inFrame?, ...}` — use it to verify you actually focused what you intended. rect is in fractions (same space as your input). When focus is inside a same-origin iframe, the descriptor reports the inner element and adds inFrame: [...] breadcrumbs.
4. browser_type_focused(text="...") → inserts text into document.activeElement (traverses into same-origin iframes automatically). Shadow roots, iframes, Lexical, Draft.js, ProseMirror all just work. Use browser_type(selector, text) instead when you have a reliable CSS selector for a light-DOM element.
5. Verify via browser_screenshot OR browser_get_attribute on a known-reachable marker (e.g. check that the Send button's aria-disabled flipped to false).

The click→type loop (canonical pattern)

1. Call browser_click_coordinate(x, y) to click the target element.
2. Check the focused_element field in the response — it tells you what actually received focus (tag, id, role, contenteditable, rect).
3. If the focused element is editable, call browser_type_focused(text="...") to insert text. Use tools to verify the text took effect — prefer checking the underlying .value / innerText via browser_evaluate or confirming the submit button enabled. A screenshot alone can mislead: narrow input boxes visually clip long text, so only a portion may appear on screen even though the full string was accepted.
4. If it is NOT editable, your click landed on the wrong thing — refine coordinates and retry. Do NOT reach for browser_evaluate + execCommand('insertText') or shadow-root traversals. The problem is the click target, not the typing method.

browser_click (selector-based) also returns focused_element, so the same check works whether you clicked by selector or coordinate.

Empirically verified (2026-04-11)

Tested against https://www.reddit.com/r/programming/ whose search input lives at:

document > reddit-search-large [shadow]
         > faceplate-search-input#search-input [shadow]
         > input[name="q"]

Shadow-piercing selectors

When you DO want a selector-based approach and know the shadow structure, browser_shadow_query and browser_get_rect support >>> shadow-piercing syntax:

browser_shadow_query("reddit-search-large >>> #search-input")
browser_get_rect("#interop-outlet >>> #ember37 >>> p")

Returns the element's rect as fractions of the viewport (feed rect.cx / rect.cy directly to click tools). Remember: browser_type and wait_for_selector do not support >>> — only shadow_query and get_rect do.

Navigation and waiting

The basics

browser_navigate(url, wait_until="load")   # "load" | "domcontentloaded" | "networkidle"
browser_wait_for_selector("h1", timeout_ms=2000)
browser_wait_for_text("Some text", timeout_ms=2000)
browser_go_back()
browser_go_forward()
browser_reload()

All return real URLs and titles. On a fast page navigate(wait_until="load") returns in sub-second. wait_for_selector and wait_for_text typically resolve in single-digit milliseconds on elements already in the DOM.

Timing expectations (measured against real sites)

For LinkedIn and other heavy SPAs, rely on sleep() after navigation to let the page hydrate.

After navigate, always let SPA hydrate

Even after wait_until="load", React/Vue SPAs often render their real chrome in a second pass. Add await sleep(2) to await sleep(3) before querying for site-specific elements. Otherwise wait_for_selector will fail on elements that do exist moments later.

Reading pages efficiently

• Prefer `browser_snapshot` over `browser_get_text("body")` — returns a compact ~1–5 KB accessibility tree vs 100+ KB of raw HTML.
• Interaction tools browser_click, browser_type, browser_type_focused, and browser_scroll wait 0.5 s for the page to settle after a successful action, then attach a fresh accessibility snapshot under the snapshot key of their result. Use it to decide your next action — do NOT call browser_snapshot separately after every action. Tune the capture via auto_snapshot_mode: "default" (full tree, the default), "simple" (trims unnamed structural nodes), "interactive" (only controls — tightest token footprint), or "off" to skip the capture entirely (useful when batching several interactions and you don't need the intermediate trees). Call browser_snapshot explicitly only when you need a newer view or a different mode than what was auto-captured.
• Complex pages (LinkedIn, Twitter/X, SPAs with virtual scrolling) can have DOMs that don't match what's visually rendered — snapshot refs may be stale, missing, or misaligned with visible layout. Try the available snapshot first; when the target is not present in that snapshot or visual position matters, switch to browser_screenshot to orient yourself.
• Only fall back to browser_get_text for extracting specific small elements by CSS selector.

Typing and keyboard input

ALWAYS click before typing into rich-text editors

The single most common "looks like it worked but send button stays disabled" failure. If you're typing into a modern editor (X/Twitter's Draft.js compose, LinkedIn's post composer, Reddit's comment box, Gmail compose, Slack, Discord, Notion, Monaco, any contenteditable), click the input area first with `browser_click_coordinate` or `browser_click(selector)` before you type.

Why this is necessary:

• React / Vue controlled components don't trust JS-sourced .focus(). React uses event delegation and watches for _native_ pointer/focus events — a click dispatched via CDP fires the real pointerdown/pointerup/click/focus sequence that React listens to, and updates its internal state. A JS-only .focus() sets document.activeElement but the framework's controlled state doesn't see it.
• Draft.js (X/Twitter compose) and Lexical (Gmail, LinkedIn DMs) use contenteditable divs with immutable editor state. They only enter "edit mode" after a real click on the editor surface. Typing at them without clicking routes keys to document.body or gets silently discarded.
• Send/submit buttons are bound to framework state, not DOM state. They're typically disabled={!hasRealContent} where hasRealContent is computed from React/Vue/Svelte state. The input field can have characters in the DOM but the button stays disabled because the framework never saw a real input event.

The symptom is always the same: you type, the characters appear visually, and the send button doesn't enable. The agent then clicks send anyway, nothing happens, and it thinks the post failed.

Safe "click-then-type-then-verify" pattern

1. Focus the real element via a real click (not JS .focus()). Use browser_get_rect(selector) (or browser_shadow_query for shadow sites) to get coordinates, then browser_click_coordinate(cx, cy). Wait ~0.5 s for the editor to open and focus to settle.

1. Type the text. Use browser_type(selector, text) for light-DOM inputs, or browser_type_focused(text=...) for shadow-DOM / already-focused inputs. Both use CDP Input.insertText by default, which is the most reliable method for rich editors (Lexical, Draft.js, ProseMirror). Wait ~500 ms for framework state to commit.

1. Verify the submit button is enabled before clicking it. Use browser_evaluate to check the button's disabled or aria-disabled attribute. Do NOT trust that typing worked — always check state.

Partial visibility is fine. Small single-line inputs, chat boxes with fixed width, and search fields commonly clip or truncate long text visually — only the tail or head may be shown on screen. Don't treat that as failure. What matters is that the framework accepted the input: the submit button enabled, or element.value / innerText read via browser_evaluate contains the full string. If the visible pixels don't match what you typed but the button is enabled and the underlying value is correct, typing succeeded — proceed.

1. Only click send if the button is enabled. If the button is still disabled, try the recovery dance: click the textarea again, press End, press a space, press Backspace — this forces React to recompute hasRealContent. Then re-check the button state.

Why browser_type uses Input.insertText by default

CDP has a dedicated method — Input.insertText — for committing text into the focused element as if IME just committed it. It bypasses the keyboard event pipeline entirely and works cleanly on every rich-text editor tested to date: Lexical (LinkedIn DMs, Gmail), Draft.js (X compose), ProseMirror (Reddit), Monaco, and plain contenteditable. Playwright uses this under the hood for keyboard.type() on rich editors.

Per-character Input.dispatchKeyEvent looks equivalent on paper, but some rich editors listen for beforeinput events with a specific shape and route insertion through their own state machine — the raw keys arrive but never get turned into text. That was the exact failure mode that left LinkedIn's message composer empty (and its Send button disabled) during the 2026-04-11 empirical run.

If you need per-keystroke dispatch (autocomplete testing, code editors, animated typing with delay_ms), pass use_insert_text=False to fall back to the old keyDown/keyUp path.

Neutralizing beforeunload draft dialogs

When a composer has unsent text and you try to navigate away or close the tab, sites like LinkedIn pop a native "You have an unsent message, leave?" confirm dialog via window.onbeforeunload. Your automation hangs waiting on the dialog — browser_close_tab and browser_navigate both time out.

Strip the handler via `browser_evaluate` before navigating:

browser_evaluate("""
    (function(){
      window.onbeforeunload = null;
      window.addEventListener('beforeunload', function(e){
        e.stopImmediatePropagation();
      }, true);
      return true;
    })()
""")
# Now browser_navigate / close_tab work without hitting a confirm

Always include an equivalent cleanup block in any script that types into a compose UI — without it, a script crash mid-type leaves the tab in an unusable state with the draft modal blocking every subsequent automation call.

Verified site-specific quirks

Plain text into a real input

For plain <input> and <textarea> elements with no framework wrapper (forms on static sites, simple search bars that pass a selector string straight through), browser_type(selector, text) is sufficient — the bridge's internal focus() call does the right thing. But when in doubt, click first. It's cheap insurance.

browser_type(selector, text)

• Sends keyDown (with key, code, text fields populated) → keyUp per character (or a single Input.insertText by default)
• Fires real keydown / keypress / input / keyup events — frameworks that branch on event.key or event.code see the right values
• Matches what Playwright and Puppeteer send

Works on real <input>, <textarea>, and contenteditable elements. For shadow-DOM inputs, see the "shadow-heavy sites" section above — browser_type(selector=) can't see past shadow boundaries; use browser_type_focused after click-coordinate focus.

Keyboard shortcuts (Ctrl+A, Shift+Tab, Cmd+Enter)

browser_press("a", modifiers=["ctrl"])         # Ctrl+A — select all
browser_press("Backspace")                      # clear selected text
browser_press("Enter", modifiers=["meta"])     # Cmd+Enter (mac) — submit
browser_press("Tab", modifiers=["shift"])      # Shift+Tab — reverse focus

Accepted modifier names (case-insensitive): "alt", "ctrl" / "control", "meta" / "cmd", "shift".

Behind the scenes this dispatches the modifier's own keyDown first, then the main key with code and windowsVirtualKeyCode populated (so Chrome's shortcut dispatcher recognises it), then releases modifiers in reverse order. Without the code + windowsVirtualKeyCode fields Chrome routes the event to the DOM without firing shortcuts — which is what plain string keys get.

Special keys

Recognized without modifiers: Enter, Tab, Escape, Backspace, Delete, ArrowUp/Down/Left/Right, Home, End, PageUp, PageDown.

Screenshots

browser_screenshot()                    # viewport, 800 px wide JPEG
browser_screenshot(full_page=True)      # full scrollable page (overview only — don't click off a full-page shot)
browser_screenshot(selector="#header")  # clip to element's rect

Returns a JPEG (quality 75, ~50–120 KB) at 800 px wide. The pixel width is purely a bandwidth choice; all tool coordinates are fractions of the viewport and are invariant to image size. Metadata includes imageWidth (800), cssWidth, cssHeight (for reference), and physicalScale. The image is annotated with a highlight rectangle/dot showing the last interaction (click, hover, type) if one happened on this tab.

The highlight overlay stays visible on the page for 10 seconds after each interaction, then fades. Before a screenshot is likely, make sure your click / hover / type happens <10 s before the screenshot.

Scrolling

• Use large scroll amounts (~2000) when loading more content — sites like Twitter and LinkedIn have lazy loading for paging.
• The scroll result includes a snapshot automatically — no need to call browser_snapshot separately.
• Never re-navigate to the same URL after scrolling — this resets your scroll position and loses loaded content.

Batching actions

• You can call multiple tools in a single turn — they execute in parallel. ALWAYS batch independent actions together. Examples: fill multiple form fields in one turn, navigate + snapshot in one turn, click + scroll if targeting different elements.
• When batching, set auto_snapshot=false on all but the last action to avoid redundant snapshots.
• Aim for 3–5 tool calls per turn minimum. One tool call per turn is wasteful.

Tab management

Close tabs as soon as you are done with them — not only at the end of the task. After reading or extracting data from a tab, close it immediately.

• Finished reading/extracting from a tab? browser_close(tab_id=...) (or no arg to close the active tab)
• Completed a multi-tab workflow? Call browser_close for each tab you opened — list with browser_tabs first if you've lost track of IDs
• More than 3 tabs open? Stop and close finished ones before opening more
• Popup appeared that you didn't need? Close it immediately

browser_tabs returns an origin field for each tab:

• "agent" — you opened it; you own it; close it when done
• "popup" — opened by a link or script; close after extracting what you need
• "startup" or "user" — leave these alone unless the task requires it

Never accumulate tabs. Treat every tab you open as a resource you must free.

The bridge automatically evicts per-tab state (_cdp_attached, _interaction_highlights) when a tab is closed, so you can't leak stale annotations or attached-debugger flags.

Site-specific selectors (verified 2026-04-11)

LinkedIn

LinkedIn enforces strict Trusted Types CSP. Any script you inject via browser_evaluate that uses innerHTML = "<...>" will be silently dropped — the wrapper element gets added but its content is empty, no console error. Always use createElement + appendChild + setAttribute for DOM injection on LinkedIn. style.cssText, textContent, and .value assignments are fine (they don't go through the Trusted Types sink).

Reddit (new reddit / shreddit)

Reddit's search input lives two shadow levels deep inside reddit-search-large > faceplate-search-input. You cannot reach it with browser_type(selector=). The working pattern:

1. browser_shadow_query("reddit-search-large >>> #search-input") → rect
2. browser_click_coordinate(rect.cx, rect.cy) → click lands on the real shadow input via native hit testing; input becomes focused
3. browser_type_focused(text="query") → dispatches to focused element via Input.insertText
4. Verify by reading .value via browser_evaluate walking the shadow path

X / Twitter

X uses Draft.js for the compose text editor, which does NOT accept synthetic input reliably. Working workaround: browser_type(selector='[data-testid="tweetTextarea_0"]', text="...", delay_ms=20). The delay gives Draft.js time to process each keystroke. The first 1–2 characters may still get eaten — accept minor truncation or prepend a throwaway character. After typing, check [data-testid="tweetButton"] has disabled: false before clicking submit.

After submitting, press Escape to close the composer.

File uploads — use browser_upload, never click the upload button

Clicking an `<input type="file">` or the button that triggers one (X's photo button, LinkedIn's attach button, Gmail's paperclip) opens Chrome's native OS file picker. That dialog is rendered by the operating system, NOT the page, so CDP cannot see it, cannot interact with it, and the automation wedges. This is the single most common way to lock up a browser session on any "compose with media" flow.

The only correct pattern: call browser_upload(selector, file_paths). It uses the CDP DOM.setFileInputFiles method, which sets the files directly on the input element's internal state as if the user had picked them — no OS dialog ever opens.

# WRONG — opens the native file picker, agent gets stuck
browser_click_coordinate(photo_button_x, photo_button_y)   # ❌

# RIGHT — sets the file programmatically, no dialog
browser_upload(
    selector="input[type='file']",          # the underlying file input
    file_paths=["/absolute/path/to/image.png"],
)

Finding the file input. On most modern SPAs the visible "Add photo" / "Attach" button is a styled <button> or <label>, and the real <input type="file"> is hidden (often display:none or opacity:0, positioned offscreen, wrapped in a <label for="...">, or injected on click). Use browser_evaluate to enumerate ALL file inputs on the page first:

browser_evaluate("""
  (function(){
    const inputs = Array.from(document.querySelectorAll('input[type="file"]'));
    return inputs.map(el => ({
      name: el.name || '',
      accept: el.accept || '',
      multiple: el.multiple,
      id: el.id || '',
      inViewport: (() => {
        const r = el.getBoundingClientRect();
        return r.width > 0 && r.height > 0;
      })(),
    }));
  })();
""")

Then pass the most specific selector that uniquely identifies the right input (e.g. input[type='file'][accept*='image'] for a photo-only upload). browser_upload doesn't care if the input is hidden or offscreen — DOM.setFileInputFiles works on any valid file input node, visible or not.

X / LinkedIn / Twitter pattern. On X (x.com/compose/post), the photo upload input is input[data-testid='fileInput'] — hidden, reachable via browser_upload. On LinkedIn feed compose, look for input[type='file'][accept*='image'] inside the post-creation modal after clicking "Add media" (clicking the Add-media button reveals the input but does NOT open the dialog; only clicking the SECOND layer — the "From computer" entry — would trigger the picker. Stop at the first layer, find the input, call browser_upload).

Verification after upload. DOM.setFileInputFiles dispatches a change event on the input but NOT the click / focus events that some sites gate their UI on. Always verify the upload actually took effect by screenshotting the composer (the uploaded image should appear as a preview) or by checking for a "preview" / "remove" element that only exists post-upload. If verification fails, the site may be reading the file via some other bridge — fall back to reading the file bytes and pasting them via the clipboard (navigator.clipboard.write with a ClipboardItem) through browser_evaluate.

If a native file picker DOES open (you clicked the wrong thing): there is no recovery via CDP. Press Escape via browser_press("Escape") immediately — this dismisses the OS dialog in Chrome on Linux/macOS. Then find the actual <input type='file'> and use browser_upload.

Common pitfalls

• Typing into a rich-text editor without clicking first → send button stays disabled. Draft.js (X), Lexical (Gmail, LinkedIn DMs), ProseMirror (Reddit), and React-controlled contenteditable elements only register input as "real" when the element received a native focus event — JS-sourced .focus() is not enough. browser_type now does this automatically via a real CDP pointer click before inserting text, but always verify the submit button's disabled state before clicking send. See the "ALWAYS click before typing" section above.
• Using per-character `keyDown` on Lexical / Draft.js editors → keys dispatch but text never appears. Those editors intercept beforeinput and route insertion through their own state machine; raw keyDown events are silently dropped. browser_type now uses Input.insertText by default (the CDP IME-commit method) which these editors accept cleanly. Only set use_insert_text=False when you explicitly need per-keystroke dispatch.
• Leaving a composer with text then trying to navigate → `beforeunload` dialog hangs the bridge. LinkedIn and several other sites pop a native "unsent message" confirm. browser_navigate and close_tab both time out against this. Always strip window.onbeforeunload = null via browser_evaluate before any navigation after typing in a composer, or wrap your logic in a try/finally that runs the cleanup block.
• Click landed in the wrong region (sidebar / header instead of target). Check focused_element in the click response — it's ground truth for what actually got focused, including the inFrame breadcrumb when focus ends up inside a same-origin iframe. If it isn't the target (e.g. className: "msg-conversation-listitem__link" when you meant to hit a composer), adjust the fraction and retry. Coordinates you pass are fractions of the viewport; the tool multiplies by cssWidth / cssHeight internally, so a wrong result means your estimated proportion was off — not that any scale went sideways.
• Accidentally passing pixels to click / hover / press_at. The tools reject any coord outside [-0.1, 1.5] with a clear error. If you see that error, you passed a pixel (like 815) instead of a fraction (like 0.475). Use browser_get_rect to get exact fractional cx/cy, or read proportions off browser_screenshot.
• Calling `wait_for_selector` on a shadow element. It'll always time out. Use browser_shadow_query or the screenshot + coordinate strategy.
• Relying on `innerHTML` in injected scripts on LinkedIn. Silently discarded. Use createElement + appendChild.
• Not waiting for SPA hydration. wait_until="load" fires before React/Vue rendering on many sites. Add a 2–3 s sleep before querying for chrome elements.
• Using `browser_type(selector)` on LinkedIn DMs or any shadow-DOM input. Won't find the element. Use browser_click_coordinate to focus, then browser_type_focused(text=...) to type.
• Clicking a "Photo" / "Attach" / "Upload" button to pick a file. This opens Chrome's NATIVE OS file picker, which is rendered outside the web page and cannot be interacted with via CDP. Your automation will hang staring at an unreachable dialog. ALWAYS use browser_upload(selector, file_paths) against the underlying <input type='file'> element — see the "File uploads" section above for the full pattern. This is the single most common way to wedge a browser session on compose-with-media flows (X/LinkedIn/Gmail).
• Keyboard shortcuts without the `code` field. Chrome's shortcut dispatcher ignores keyboard events that lack a code or windowsVirtualKeyCode. browser_press(..., modifiers=[...]) populates these automatically; raw Input.dispatchKeyEvent calls from browser_evaluate may not.
• Taking a screenshot more than 10s after the last interaction and expecting the highlight to still be visible. The overlay fades after 10s. Take the screenshot sooner, or re-trigger the interaction.
• Expecting `browser_navigate` to return when you specified `wait_until="networkidle"` on a busy site. networkidle is approximate — some sites keep a websocket or analytics beacon open forever. Use "load" or "domcontentloaded" for reliable timing.

Dead CDP sessions and auto-recovery

If Chrome detaches the debugger for its own reasons (tab closed, user opened DevTools manually, cross-origin navigation, chrome:// page loaded), the bridge detects the "target closed" / "not attached" error on the next call and automatically reattaches + retries once. You don't need to handle this yourself.

If reattach also fails, you'll get the underlying CDP error string — that's a real problem, usually the tab is gone.

browser_evaluate is a last-resort escape hatch

Before using `browser_evaluate`, try these first — in this order:

1. `browser_screenshot` + `browser_click_coordinate` — works on every site regardless of shadow DOM, iframes, obfuscated classes. This is the default path for "click a thing you can see."
2. `browser_type(use_insert_text=True, text=...)` — for typing into ANY input/contenteditable, including Lexical and Draft.js. Handles click-focus-insert with built-in retries. Do not call document.execCommand('insertText') via evaluate; this tool already does it correctly.
3. `browser_shadow_query` or `browser_get_rect(selector)` with the >>> shadow-piercing syntax — for selector-based lookups across shadow roots.
4. `browser_get_text` / `browser_get_attribute` — for reading element state by selector.
5. `browser_snapshot` — for dumping the accessibility tree of the page.

If all five of those fit your goal, do not use `browser_evaluate`. Each evaluate call is a small LLM round-trip of ~30-100 tokens of JS plus a JSON response; five of them burn more context than a single screenshot-and-coordinate does, with less reliability.

Anti-patterns — stop immediately if you catch yourself doing these

• Trying multiple `querySelectorAll` variants when the first returned `[]`. Different selectors on the same page rarely work if the first guess failed — modern SPAs obfuscate class names at build time. After one empty result, switch to browser_screenshot + browser_click_coordinate. Do not write .artdeco-list__item, then [data-test-incoming-invitation-card], then [class*="invitation"] — you are already on the wrong path.
• Writing `walk(root)` recursive shadow-DOM traversal functions. Use browser_shadow_query — it traverses at the CDP level (native C++), not by re-running a recursive JS function every call.
• Calling `document.execCommand('insertText', ...)` to type into a contenteditable. Use browser_type(use_insert_text=True, text='...'). The high-level tool handles the exact same Lexical/Draft.js case but with click-focus-retry logic built in.
• Accessing `iframe.contentDocument`. Rarely works (cross-origin, late hydration) and when it does, the code is brittle. Use browser_screenshot to see the iframe, then browser_click_coordinate to interact.
• Using `innerHTML = "<...>"` on a Trusted Types site (LinkedIn, GitHub). The assignment is silently dropped. Use createElement + appendChild if you must inject DOM — but first, ask whether you really need to.
• Triggering React/Vue state via synthetic `dispatchEvent`. Frameworks watch for real browser events. Use browser_click_coordinate, browser_press, or browser_type — all go through CDP's native event pipeline.

Legitimate uses (when nothing semantic fits)

• Reading a computed style, window.innerWidth/Height, document.scrollingElement.scrollTop, or other layout values the tools don't expose.
• Firing a one-shot site-specific API call (analytics beacon, feature-flag toggle).
• Stripping onbeforeunload before navigating away from a page with an unsent draft (LinkedIn, Gmail).
• Detecting whether a specific shadow-root host exists before a follow-up screenshot.

In all of these cases the script is SHORT (< 10 lines) and the result is CONSUMED (read, then acted on), not further probed.

Login & auth walls

• If you see a "Log in" or "Sign up" prompt, report the auth wall to user immediately — do NOT attempt to log in.
• Check for cookie consent banners and dismiss them if they block content.

Error recovery

• If a tool fails, retry once with the same approach.
• If it fails a second time, STOP retrying and switch approach.
• If browser_snapshot fails, try browser_get_text with a specific small selector as fallback.
• If browser_open fails or page seems stale, browser_stop, then browser_open(url) again to recreate a fresh context.

Verified workflows

These sequences have been empirically verified against real production sites on 2026-04-11.

Search on X and read the live dropdown

browser_navigate("https://x.com/explore", wait_until="load")
# Wait for SPA hydration
sleep(3)
browser_wait_for_selector("input[data-testid='SearchBox_Search_Input']", timeout_ms=5000)
rect = browser_get_rect("input[data-testid='SearchBox_Search_Input']")
browser_click_coordinate(rect.cx, rect.cy)
browser_type("input[data-testid='SearchBox_Search_Input']", "openai", clear_first=True)
# Screenshot now shows live search suggestions
browser_screenshot()
browser_press("Escape", selector="input[data-testid='SearchBox_Search_Input']")

Search Reddit (shadow DOM)

browser_navigate("https://www.reddit.com/r/programming/", wait_until="load")
sleep(2)
# Shadow-pierce the nested search input
sq = browser_shadow_query("reddit-search-large >>> #search-input")
browser_click_coordinate(sq.rect.cx, sq.rect.cy)
# Typing can't use selector (shadow); use browser_type_focused on the focused input
browser_type_focused(text="python")
browser_screenshot()
browser_press("Escape")

Search LinkedIn and dismiss without submitting

browser_navigate("https://www.linkedin.com/feed/", wait_until="load")
sleep(3)
browser_wait_for_selector("input[data-testid='typeahead-input']", timeout_ms=5000)
rect = browser_get_rect("input[data-testid='typeahead-input']")
browser_click_coordinate(rect.cx, rect.cy)
browser_type("input[data-testid='typeahead-input']", "anthropic", clear_first=True)
# Dropdown shows real live suggestions
browser_screenshot()
browser_press("Escape", selector="input[data-testid='typeahead-input']")

Debugging checklist when a click / type "didn't work"

1. Send button stays disabled after typing? Two possible causes. (a) You didn't click the input first, so React never saw a native focus event. browser_type now clicks automatically — but if you're using raw Input.dispatchKeyEvent, click first yourself. (b) You're using per-character keyDown on a Lexical / Draft.js editor, and those editors dropped the keys because they listen for beforeinput with a specific shape. Switch to browser_type(selector, text) (which now uses Input.insertText by default) or, at a lower level, call CDP Input.insertText directly. Always browser_evaluate the submit button's disabled / aria-disabled state before clicking send; if still disabled after those fixes, the framework never saw real input.
2. Did the selector match anything? Run browser_get_rect(selector) — if it returns visible=False or zero rect, the element isn't laid out yet. Wait longer or use a different selector.
3. Is the element inside a shadow root? Try browser_shadow_query(path). If your selector is light-DOM only, switch to the screenshot + coordinate strategy.
4. Did the click hit something on top of the element? Register a temporary event listener via browser_evaluate on the target element, click, then read window.__hits to see what actually received the click. If something else is intercepting (overlay, modal, floating button), dismiss it first.
5. Did `type_text` find the element but fail to insert text? Some editors (Draft.js on X, ProseMirror on some sites, Monaco) require a small delay_ms between keystrokes. Try delay_ms=20.
6. Is this a keyboard shortcut that doesn't fire? Make sure you're using browser_press(key, modifiers=[...]) — not raw browser_evaluate with dispatchEvent. Chrome ignores shortcut key events that lack code and windowsVirtualKeyCode.
7. Did the navigation actually complete? Check the return value of browser_navigate — it now returns a real url and title. An empty title usually means a blank page or a hung load.
8. Is your screenshot stale? The highlight overlay stays for 10 s; if the screenshot was taken later, the annotation is gone but the click was real. Check the logs of browser_click_coordinate to see the coordinates that were actually sent.