chrome-bridge-mcp 1.17.0 → 1.19.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -180,7 +180,11 @@ paste into a chat.
180
180
  | `Port N is held by a process that is not chrome-bridge` | Something else owns the port. Free it or change `CHROME_BRIDGE_PORT`. |
181
181
  | `execute_js` fails | Enable **Allow user scripts** in `chrome://extensions` → Chrome Bridge → Details (Chrome 138+; on 135-137 enable Developer Mode). |
182
182
  | `read_console` returns `note=Instrumentation not loaded` | The page was opened before the extension, "Capture console & metrics" is off, or the page isn't injectable (`chrome://`). Reload it. |
183
- | Screenshot times out | On ChromeOS a fully occluded window stops producing frames; captures fail after 10s. Bring the window forward. |
183
+ | Screenshot times out, or `image readback failed` | A minimized or fully covered window stops producing frames; captures fail after 10 s. Bring the window forward. |
184
+ | `wait_for`/`scroll` return `page_hidden: true`, pages stop updating | Same cause: Chrome does not render a hidden page and slows its timers (after a few minutes, to one wake-up per minute). `get_page_info` reports `visibility`. Bring the window on screen, or `create_tab` with `new_window` and bounds. |
185
+ | `screenshot presets` says `NOT APPLIED` | Presets resize the window; they do not emulate a phone (no device pixel ratio, UA or touch). The window manager enforces a minimum width, and no window exceeds the screen (e.g. at most a 1536×686 viewport on a 1536×864 ChromeOS screen). For phone emulation or larger viewports use a headless browser. |
186
+ | `type_text` returns `mismatch: true` | The field rejected the value. Events from an extension have `isTrusted: false`, and some widgets (date pickers, search boxes with tokenizers) discard them. Try `mode: 'keys'`, then the site's own controls (the calendar buttons), or `handoff`. |
187
+ | `[media removed: request limit]` instead of a screenshot | Written by the MCP client, not by the bridge: too many images in one request. Use `save_to`, or `element_screenshot` with `region` and a small `scale`. |
184
188
  | Commands work, then stop | The MV3 service worker restarted and in-memory state (network log, diff baselines, HTTP auth) was reset. Re-run the monitoring call. |
185
189
  | Extension dropped on every ChromeOS reboot | Install from the Web Store instead of Load unpacked. |
186
190
  | Tool missing from the list | It's in an opt-in group. Check `get_status` → `caps_available`, then set `CHROME_BRIDGE_CAPS=all`. |
@@ -0,0 +1,97 @@
1
+ /**
2
+ * findTextInPage gira nella pagina: chrome.scripting ne serializza il
3
+ * sorgente, quindi è autocontenuta (niente import, niente chiusure).
4
+ *
5
+ * Cerca sul testo degli elementi, non dei singoli nodi di testo: un'etichetta
6
+ * spezzata in più nodi («<span>Prestazioni</span> e <b>clic</b>», o con
7
+ * &nbsp;) prima non veniva mai trovata, ed è la forma normale dei menu di
8
+ * Meta. Fra le parole dell'ago vale qualunque sequenza di spazi o &nbsp;.
9
+ * Ogni occorrenza è riportata una volta, dall'elemento più interno che la
10
+ * contiene tutta; si entra anche negli shadow root aperti.
11
+ *
12
+ * Il costo resta basso sulle pagine grandi: si scende solo nei rami il cui
13
+ * testo contiene l'ago.
14
+ */
15
+ export function findTextInPage(needleRaw, caseSensitive, maxResults, doc = document, win = window) {
16
+ const words = String(needleRaw || '').split(/[\s ]+/).filter(Boolean);
17
+ if (!words.length) return { count: 0, matches: [] };
18
+ const pattern = words.map((w) => w.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')).join('[\\s\\u00a0]+');
19
+ const flags = caseSensitive ? 'g' : 'gi';
20
+ const SKIP = new Set(['SCRIPT', 'STYLE', 'NOSCRIPT', 'TEMPLATE']);
21
+
22
+ const textOf = (node) => {
23
+ if (node.nodeType === 3) return node.textContent || '';
24
+ if (node.nodeType !== 1 || SKIP.has(node.tagName)) return '';
25
+ let out = '';
26
+ for (const c of node.childNodes) out += textOf(c);
27
+ return out;
28
+ };
29
+ const ranges = (text) => {
30
+ const re = new RegExp(pattern, flags);
31
+ const out = [];
32
+ let m;
33
+ while ((m = re.exec(text))) out.push([m.index, m.index + m[0].length]);
34
+ return out;
35
+ };
36
+ const selectorFor = (el) => {
37
+ if (el.id) return `#${el.id}`;
38
+ const tag = el.tagName.toLowerCase();
39
+ const cls = el.classList && el.classList.length ? `.${[...el.classList].slice(0, 2).join('.')}` : '';
40
+ return `${tag}${cls}`;
41
+ };
42
+
43
+ const matches = [];
44
+ const visit = (el, inShadow) => {
45
+ if (matches.length >= maxResults || SKIP.has(el.tagName)) return;
46
+ const text = textOf(el);
47
+ const found = ranges(text);
48
+ if (!found.length) return;
49
+ // Intervalli dei figli elemento nel testo di el: textOf concatena i figli
50
+ // in ordine, quindi gli offset coincidono.
51
+ const kids = [];
52
+ let off = 0;
53
+ for (const c of el.childNodes) {
54
+ const len = textOf(c).length;
55
+ if (c.nodeType === 1) kids.push({ el: c, from: off, to: off + len });
56
+ off += len;
57
+ }
58
+ const owned = found.filter(([a, b]) => !kids.some((k) => a >= k.from && b <= k.to));
59
+ if (owned.length) {
60
+ const rect = el.getBoundingClientRect();
61
+ const flat = text.replace(/[\s ]/g, ' ');
62
+ for (const [a, b] of owned) {
63
+ if (matches.length >= maxResults) break;
64
+ matches.push({
65
+ selector: selectorFor(el),
66
+ context: flat.substring(Math.max(0, a - 40), b + 40).replace(/ +/g, ' ').trim(),
67
+ visible: rect.width > 0 && rect.height > 0,
68
+ position: { x: Math.round(rect.x + win.scrollX), y: Math.round(rect.y + win.scrollY) },
69
+ ...(inShadow ? { in_shadow: true } : {}),
70
+ });
71
+ }
72
+ }
73
+ for (const k of kids) {
74
+ if (found.some(([a, b]) => a >= k.from && b <= k.to)) visit(k.el, inShadow);
75
+ }
76
+ };
77
+
78
+ // Gli shadow root non compaiono nel testo dell'ospite: si cercano a parte,
79
+ // anche annidati, dopo il documento.
80
+ const roots = [];
81
+ const collectShadows = (root) => {
82
+ for (const el of root.querySelectorAll('*')) {
83
+ if (el.shadowRoot) {
84
+ roots.push(el.shadowRoot);
85
+ collectShadows(el.shadowRoot);
86
+ }
87
+ }
88
+ };
89
+ if (doc.body) {
90
+ visit(doc.body, false);
91
+ collectShadows(doc.body);
92
+ }
93
+ for (const root of roots) {
94
+ for (const c of root.children) visit(c, true);
95
+ }
96
+ return { count: matches.length, matches };
97
+ }
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "manifest_version": 3,
3
3
  "name": "Chrome Bridge for Claude Code",
4
- "version": "1.17.0",
4
+ "version": "1.19.0",
5
5
  "description": "Bridge your browser to Claude Code: 59 token-efficient web-dev automation tools over a local WebSocket. ChromeOS included.",
6
6
  "homepage_url": "https://github.com/frsorrentino/chrome-bridge",
7
7
  "minimum_chrome_version": "135",
@@ -11,6 +11,7 @@ import { pageFingerprint } from './lib/page-fingerprint.js';
11
11
  import { createPacer } from './lib/capture-pacing.js';
12
12
  import { computeTiles } from './lib/tile-layout.js';
13
13
  import { classifyDownload } from './lib/download-state.js';
14
+ import { findTextInPage } from './lib/find-text.js';
14
15
  const { pushError } = globalThis.__cbTelemetry;
15
16
 
16
17
  const DEFAULT_PORT = 8765;
@@ -603,11 +604,60 @@ async function captureVisible(windowId, options = { format: 'png' }) {
603
604
  }, 10000);
604
605
  chrome.tabs.captureVisibleTab(windowId, options).then(
605
606
  (dataUrl) => { clearTimeout(timer); resolve(dataUrl); },
606
- (err) => { clearTimeout(timer); reject(err); },
607
+ (err) => {
608
+ clearTimeout(timer);
609
+ // «image readback failed» è lo stesso guasto della finestra che non
610
+ // disegna, detto dal compositor: stesso messaggio, stesso rimedio.
611
+ if (/readback failed/i.test(err?.message || '')) {
612
+ reject(new Error(`Capture failed (${err.message}): the window is not rendering frames (fully occluded, minimized, or frozen). Bring the Chrome window at least partially on screen and retry.`));
613
+ } else {
614
+ reject(err);
615
+ }
616
+ },
607
617
  );
608
618
  });
609
619
  }
610
620
 
621
+ // --- Utility: scadenze tenute dal service worker ---
622
+ // Le attese che girano nella pagina usano setTimeout, e Chrome raggruppa i
623
+ // timer delle pagine nascoste fino a un risveglio al minuto: il controllo
624
+ // della scadenza scattava tardi e vinceva il timeout di trasporto (60 s per
625
+ // un'attesa da 25, osservato su Meta Ads Manager il 23/09/2026). Il service
626
+ // worker non è rallentato: la scadenza la tiene lui.
627
+ const HIDDEN_HINT = 'The page is hidden (window minimized, covered or in the background): Chrome slows its timers and stops rendering. Bring the window on screen, or create_tab new_window with bounds, then retry.';
628
+
629
+ async function pageHidden(tabId) {
630
+ try {
631
+ const q = chrome.scripting.executeScript({ target: { tabId }, func: () => document.visibilityState === 'hidden' });
632
+ const r = await Promise.race([q, new Promise((res) => setTimeout(() => res(null), 1000))]);
633
+ return r?.[0]?.result === true;
634
+ } catch {
635
+ return false;
636
+ }
637
+ }
638
+
639
+ // Esegue run(); se non risponde entro deadlineMs restituisce onTimeout. In
640
+ // entrambi i casi un esito negativo su pagina nascosta porta page_hidden e il
641
+ // suggerimento, così chi chiama non ritenta tre volte la stessa attesa.
642
+ async function withDeadline(tabId, deadlineMs, run, onTimeout, isNegative) {
643
+ const DEADLINE = Symbol('deadline');
644
+ let timer;
645
+ let result;
646
+ try {
647
+ result = await Promise.race([
648
+ run(),
649
+ new Promise((res) => { timer = setTimeout(() => res(DEADLINE), deadlineMs); }),
650
+ ]);
651
+ } finally {
652
+ clearTimeout(timer);
653
+ }
654
+ if (result === DEADLINE) result = { ...onTimeout };
655
+ if (result && isNegative(result) && await pageHidden(tabId)) {
656
+ return { ...result, page_hidden: true, hint: HIDDEN_HINT };
657
+ }
658
+ return result;
659
+ }
660
+
611
661
  // --- Utility: rendi visibile un tab per la cattura SENZA rubare il focus OS ---
612
662
  // captureVisibleTab richiede solo che il tab sia attivo nella sua finestra,
613
663
  // non che la finestra sia in primo piano. Al termine ripristina il tab che
@@ -683,7 +733,7 @@ async function bitmapToBase64Capped(bitmap, maxSide = MAX_IMAGE_SIDE) {
683
733
 
684
734
  // --- Implementazione comandi ---
685
735
 
686
- async function cmdGetTabs({ include_windows = false } = {}) {
736
+ async function cmdGetTabs({ include_windows = false, ended } = {}) {
687
737
  const tabs = await chrome.tabs.query({});
688
738
  const list = tabs.map((t) => ({
689
739
  id: t.id,
@@ -691,8 +741,17 @@ async function cmdGetTabs({ include_windows = false } = {}) {
691
741
  title: t.title,
692
742
  active: t.active,
693
743
  windowId: t.windowId,
744
+ // Una scheda scartata da Chrome (memory saver) esiste ancora ma non ha un
745
+ // renderer: execute_js e compagni falliscono finché non torna in primo piano.
746
+ ...(t.discarded ? { discarded: true } : {}),
694
747
  }));
695
- if (!include_windows) return list;
748
+ // ended = id che il server sa suoi e non trova più: la forma ad array resta
749
+ // quella di sempre per chi non lo chiede.
750
+ const open = new Set(list.map((t) => t.id));
751
+ const endings = Array.isArray(ended) && ended.length
752
+ ? Object.fromEntries(ended.filter((id) => !open.has(id)).map((id) => [id, tabEndings.get(id) ?? { reason: 'unknown' }]))
753
+ : null;
754
+ if (!include_windows) return endings ? { tabs: list, ended: endings } : list;
696
755
 
697
756
  // Senza geometria e stato non si può decidere dove mettere una finestra: si
698
757
  // può solo indovinare. `left` in particolare è ciò che identifica il monitor
@@ -712,7 +771,7 @@ async function cmdGetTabs({ include_windows = false } = {}) {
712
771
  // provarci.
713
772
  scriptable: (w.tabs || []).some((t) => /^(https?|file):/i.test(t.url || '')),
714
773
  }));
715
- return { tabs: list, windows };
774
+ return endings ? { tabs: list, windows, ended: endings } : { tabs: list, windows };
716
775
  }
717
776
 
718
777
  async function cmdNavigate({ url, tab_id }) {
@@ -1161,7 +1220,14 @@ async function cmdMoveTab({ tab_id, window_id, new_window = false, window_type,
1161
1220
  if (typeof v === 'number') winOpts[k] = v;
1162
1221
  }
1163
1222
  const win = await chrome.windows.create(winOpts);
1223
+ // La finestra nuova può nascere dietro un'altra (lo decide il window
1224
+ // manager): allora la pagina resta hidden e non disegna. Lo si dice qui,
1225
+ // non dopo tre attese scadute.
1226
+ await new Promise((r) => setTimeout(r, 300));
1227
+ const hidden = await pageHidden(tab_id);
1164
1228
  return {
1229
+ visibility: hidden ? 'hidden' : 'visible',
1230
+ ...(hidden ? { hint: HIDDEN_HINT } : {}),
1165
1231
  moved: tab_id,
1166
1232
  url: before.url,
1167
1233
  from_window: before.windowId,
@@ -1189,12 +1255,26 @@ async function cmdMoveTab({ tab_id, window_id, new_window = false, window_type,
1189
1255
  };
1190
1256
  }
1191
1257
 
1258
+ // Perché una scheda non c'è più: «No tab with id» da solo non distingue una
1259
+ // chiusura a mano da una sostituzione. In memoria e limitato: se il service
1260
+ // worker riparte il registro si svuota e il motivo torna 'unknown'.
1261
+ const tabEndings = new Map();
1262
+ const TAB_ENDINGS_MAX = 200;
1263
+ const bridgeClosing = new Set();
1264
+
1265
+ function recordTabEnding(tabId, ending) {
1266
+ if (bridgeClosing.delete(tabId) && ending.reason === 'closed') ending = { reason: 'closed_by_bridge' };
1267
+ tabEndings.set(tabId, { ...ending, at: new Date().toISOString() });
1268
+ if (tabEndings.size > TAB_ENDINGS_MAX) tabEndings.delete(tabEndings.keys().next().value);
1269
+ }
1270
+
1192
1271
  async function cmdTabAction({ action, tab_id, bypass_cache = false }) {
1193
1272
  if (!action) throw new Error('Missing required parameter: action');
1194
1273
  const tabId = await resolveTabId(tab_id);
1195
1274
 
1196
1275
  if (action === 'close') {
1197
- await chrome.tabs.remove(tabId);
1276
+ bridgeClosing.add(tabId);
1277
+ try { await chrome.tabs.remove(tabId); } catch (e) { bridgeClosing.delete(tabId); throw e; }
1198
1278
  return { action, closed: tabId };
1199
1279
  }
1200
1280
  if (action === 'discard') {
@@ -1321,6 +1401,9 @@ async function cmdGetPageInfo({ tab_id, frame_id }) {
1321
1401
  url: location.href,
1322
1402
  doctype: document.doctype ? document.doctype.name : null,
1323
1403
  charset: document.characterSet,
1404
+ // hidden = finestra minimizzata, coperta o in secondo piano: la pagina
1405
+ // non disegna e i suoi timer rallentano. Va saputo prima di attendere.
1406
+ visibility: document.visibilityState,
1324
1407
  metas,
1325
1408
  scripts,
1326
1409
  stylesheets,
@@ -1731,7 +1814,7 @@ async function cmdWaitForElement({ selector, timeout = 10000, interval = 200, vi
1731
1814
  const tabId = await resolveTabId(tab_id);
1732
1815
  const clampedInterval = Math.max(interval, 50);
1733
1816
 
1734
- const results = await chrome.scripting.executeScript({
1817
+ const run = () => chrome.scripting.executeScript({
1735
1818
  target: scriptTarget(tabId, frame_id),
1736
1819
  func: (sel, tout, intv, vis) => {
1737
1820
  function deepQuery(sel) {
@@ -1783,8 +1866,9 @@ async function cmdWaitForElement({ selector, timeout = 10000, interval = 200, vi
1783
1866
  },
1784
1867
  args: [selector, timeout, clampedInterval, visible],
1785
1868
  world: 'MAIN',
1786
- });
1787
- return results?.[0]?.result ?? { found: false, error: 'No result' };
1869
+ }).then((results) => results?.[0]?.result ?? { found: false, error: 'No result' });
1870
+ return withDeadline(tabId, timeout + 1000, run,
1871
+ { found: false, error: `Element not found within ${timeout}ms: ${selector}` }, (r) => r.found === false);
1788
1872
  }
1789
1873
 
1790
1874
  // --- scroll_to ---
@@ -2035,11 +2119,16 @@ async function cmdViewportResize({ preset, width, height, left, top, state, read
2035
2119
  }
2036
2120
  const winNow = await chrome.windows.get(tab.windowId);
2037
2121
  const geometry = { left: winNow.left, top: winNow.top, width: winNow.width, height: winNow.height, state: winNow.state };
2038
- if (read_only) return { actual, window: geometry, viewport_error };
2122
+ // Uno zoom sotto il 100% allarga il viewport in CSS px oltre lo schermo:
2123
+ // senza questo numero un 2226 px su uno schermo da 1536 non si spiega.
2124
+ let zoom = null;
2125
+ try { zoom = await chrome.tabs.getZoom(tabId); } catch { /* scheda senza zoom (chrome://) */ }
2126
+ if (read_only) return { actual, window: geometry, zoom, viewport_error };
2039
2127
  const win = winNow;
2040
2128
  return {
2041
2129
  requested: { width: targetW, height: targetH, preset: preset || null, left, top, state },
2042
2130
  actual,
2131
+ zoom,
2043
2132
  viewport_error,
2044
2133
  // Il window manager può accettare e ignorare: senza il confronto una
2045
2134
  // finestra rimasta ferma passerebbe per spostata.
@@ -3513,41 +3602,7 @@ async function cmdFindText({ text, case_sensitive = false, max_results = 20, tab
3513
3602
  const tabId = await resolveTabId(tab_id);
3514
3603
  const results = await chrome.scripting.executeScript({
3515
3604
  target: { tabId },
3516
- func: (needleRaw, caseSensitive, maxResults) => {
3517
- const needle = caseSensitive ? needleRaw : needleRaw.toLowerCase();
3518
- const matches = [];
3519
- const walker = document.createTreeWalker(document.body, NodeFilter.SHOW_TEXT, {
3520
- acceptNode: (n) => {
3521
- const p = n.parentElement;
3522
- if (!p || ['SCRIPT', 'STYLE', 'NOSCRIPT'].includes(p.tagName)) return NodeFilter.FILTER_REJECT;
3523
- return NodeFilter.FILTER_ACCEPT;
3524
- },
3525
- });
3526
- const selectorFor = (el) => {
3527
- if (el.id) return `#${el.id}`;
3528
- const tag = el.tagName.toLowerCase();
3529
- const cls = el.classList.length ? `.${[...el.classList].slice(0, 2).join('.')}` : '';
3530
- return `${tag}${cls}`;
3531
- };
3532
- let node;
3533
- while ((node = walker.nextNode()) && matches.length < maxResults) {
3534
- const hay = caseSensitive ? node.textContent : node.textContent.toLowerCase();
3535
- let idx = hay.indexOf(needle);
3536
- while (idx !== -1 && matches.length < maxResults) {
3537
- const parent = node.parentElement;
3538
- const rect = parent.getBoundingClientRect();
3539
- const ctx = node.textContent.substring(Math.max(0, idx - 40), idx + needleRaw.length + 40);
3540
- matches.push({
3541
- selector: selectorFor(parent),
3542
- context: ctx.trim(),
3543
- visible: rect.width > 0 && rect.height > 0,
3544
- position: { x: Math.round(rect.x + window.scrollX), y: Math.round(rect.y + window.scrollY) },
3545
- });
3546
- idx = hay.indexOf(needle, idx + 1);
3547
- }
3548
- }
3549
- return { count: matches.length, matches };
3550
- },
3605
+ func: findTextInPage,
3551
3606
  args: [text, case_sensitive, max_results],
3552
3607
  world: 'MAIN',
3553
3608
  });
@@ -4425,7 +4480,7 @@ async function cmdWaitForText({ text, selector, timeout = 10000, interval = 200,
4425
4480
  if (!text) throw new Error('Missing required parameter: text');
4426
4481
  const tabId = await resolveTabId(tab_id);
4427
4482
 
4428
- const results = await chrome.scripting.executeScript({
4483
+ const run = () => chrome.scripting.executeScript({
4429
4484
  target: scriptTarget(tabId, frame_id),
4430
4485
  func: (needle, scopeSel, tout, intv) => new Promise((resolve) => {
4431
4486
  const started = Date.now();
@@ -4454,9 +4509,10 @@ async function cmdWaitForText({ text, selector, timeout = 10000, interval = 200,
4454
4509
  }),
4455
4510
  args: [text, selector || null, timeout, Math.max(50, interval)],
4456
4511
  world: 'MAIN',
4457
- });
4512
+ }).then((results) => results?.[0]?.result ?? { found: false, error: 'No result' });
4458
4513
 
4459
- return results?.[0]?.result ?? { found: false, error: 'No result' };
4514
+ return withDeadline(tabId, timeout + 1000, run,
4515
+ { found: false, error: `Text not found within ${timeout}ms: ${text}` }, (r) => r.found === false);
4460
4516
  }
4461
4517
 
4462
4518
  async function cmdSavePage({ tab_id }) {
@@ -4561,9 +4617,16 @@ async function cmdHttpAuth({ action, username, password }) {
4561
4617
 
4562
4618
  // --- Tab lifecycle: cleanup injection state ---
4563
4619
 
4564
- chrome.tabs.onRemoved.addListener((tabId) => {
4620
+ chrome.tabs.onRemoved.addListener((tabId, info) => {
4565
4621
  browserNetLog.delete(tabId);
4566
4622
  mainFrameHeaders.delete(tabId);
4623
+ recordTabEnding(tabId, { reason: info?.isWindowClosing ? 'window_closed' : 'closed' });
4624
+ });
4625
+
4626
+ // Chrome sostituisce l'id quando rimpiazza la scheda (prerender, alcuni
4627
+ // discard): per chi teneva il vecchio id è una scheda sparita senza chiusura.
4628
+ chrome.tabs.onReplaced.addListener((addedTabId, removedTabId) => {
4629
+ recordTabEnding(removedTabId, { reason: 'replaced', replaced_by: addedTabId });
4567
4630
  });
4568
4631
 
4569
4632
  // --- get_interactives ---
@@ -4677,52 +4740,92 @@ async function cmdWaitForFunction({ expression, timeout = 10000, polling_ms = 10
4677
4740
 
4678
4741
  // --- scroll_until ---
4679
4742
 
4680
- async function cmdScrollUntil({ until = 'no_new_content', selector, max_scrolls = 20, step_px, settle_ms = 400, tab_id }) {
4743
+ async function cmdScrollUntil({ until = 'no_new_content', selector, container, max_scrolls = 20, step_px, settle_ms = 400, tab_id }) {
4681
4744
  const tabId = await resolveTabId(tab_id);
4682
4745
  if (until === 'network_idle') await ensureNetworkHook(tabId);
4683
- const results = await chrome.scripting.executeScript({
4746
+ const run = () => chrome.scripting.executeScript({
4684
4747
  target: { tabId },
4685
- func: (mode, sel, maxScrolls, stepPx, settleMs) => new Promise((resolve) => {
4748
+ func: (mode, sel, containerSel, maxScrolls, stepPx, settleMs) => new Promise((resolve) => {
4686
4749
  const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
4750
+ const docEl = document.scrollingElement || document.documentElement;
4751
+ const scrollable = (el) => {
4752
+ if (el.scrollHeight <= el.clientHeight + 2) return false;
4753
+ const oy = getComputedStyle(el).overflowY;
4754
+ return oy === 'auto' || oy === 'scroll' || oy === 'overlay';
4755
+ };
4756
+ const describe = (el) => {
4757
+ if (el.id) return `#${el.id}`;
4758
+ const cls = typeof el.className === 'string' && el.className.trim() ? `.${el.className.trim().split(/\s+/).slice(0, 2).join('.')}` : '';
4759
+ return el.tagName.toLowerCase() + cls;
4760
+ };
4761
+ // Le app come Ads Manager bloccano il documento e scorrono in un div
4762
+ // interno: scorrendo solo window la condizione «bottom» era vera al
4763
+ // primo giro. Senza container esplicito si prende il contenitore
4764
+ // scorrevole più grande, ma solo se il documento non scorre.
4765
+ let box = null;
4766
+ if (containerSel) {
4767
+ box = document.querySelector(containerSel);
4768
+ if (!box) { resolve({ stopped_reason: 'error', error: `Container not found: ${containerSel}`, scrolls: 0 }); return; }
4769
+ } else if (docEl.scrollHeight <= innerHeight + 2
4770
+ // overflow hidden su html o body: l'utente non può scorrere il
4771
+ // documento anche se è più alto della finestra (e scrollBy sì).
4772
+ || [document.documentElement, document.body].some((e) => e && getComputedStyle(e).overflowY === 'hidden')) {
4773
+ let best = null;
4774
+ let bestArea = 0;
4775
+ for (const el of document.querySelectorAll('body *')) {
4776
+ if (el.clientHeight < 50 || !scrollable(el)) continue;
4777
+ const area = el.clientWidth * el.clientHeight;
4778
+ if (area > bestArea) { best = el; bestArea = area; }
4779
+ }
4780
+ box = best;
4781
+ }
4782
+ const s = box
4783
+ ? { by: (d) => box.scrollBy(0, d), pos: () => box.scrollTop, height: () => box.scrollHeight, view: () => box.clientHeight, name: describe(box) }
4784
+ : { by: (d) => window.scrollBy(0, d), pos: () => window.scrollY, height: () => docEl.scrollHeight, view: () => innerHeight, name: 'document' };
4785
+ const done = (stopped_reason, scrolls) => resolve({ stopped_reason, scrolls, finalScrollY: Math.round(s.pos()), container: s.name });
4687
4786
  (async () => {
4688
4787
  let scrolls = 0;
4689
- let lastHeight = document.documentElement.scrollHeight;
4788
+ let lastHeight = s.height();
4690
4789
  let stableCount = 0;
4691
- const step = stepPx || window.innerHeight;
4790
+ const step = stepPx || s.view();
4692
4791
  for (scrolls = 0; scrolls < maxScrolls; scrolls++) {
4693
4792
  if (mode === 'element' && sel) {
4694
4793
  const el = document.querySelector(sel);
4695
4794
  if (el) {
4696
4795
  const r = el.getBoundingClientRect();
4697
4796
  const vis = r.width > 0 && r.height > 0 && getComputedStyle(el).visibility !== 'hidden';
4698
- if (vis) { resolve({ stopped_reason: 'element_found', scrolls, finalScrollY: Math.round(window.scrollY) }); return; }
4797
+ if (vis) { done('element_found', scrolls); return; }
4699
4798
  }
4700
4799
  }
4701
- window.scrollBy(0, step);
4800
+ s.by(step);
4702
4801
  await sleep(settleMs);
4703
4802
 
4704
4803
  if (mode === 'network_idle') {
4705
4804
  const inflight = window.__chromeBridge_inflight || 0;
4706
4805
  const last = window.__chromeBridge_lastNetActivity || Date.now();
4707
- if (inflight === 0 && Date.now() - last >= settleMs) { resolve({ stopped_reason: 'network_idle', scrolls: scrolls + 1, finalScrollY: Math.round(window.scrollY) }); return; }
4806
+ if (inflight === 0 && Date.now() - last >= settleMs) { done('network_idle', scrolls + 1); return; }
4708
4807
  }
4709
4808
 
4710
- const h = document.documentElement.scrollHeight;
4809
+ const h = s.height();
4711
4810
  if (h === lastHeight) {
4712
4811
  stableCount++;
4713
- if (mode === 'no_new_content' && stableCount >= 2) { resolve({ stopped_reason: 'no_new_content', scrolls: scrolls + 1, finalScrollY: Math.round(window.scrollY) }); return; }
4812
+ if (mode === 'no_new_content' && stableCount >= 2) { done('no_new_content', scrolls + 1); return; }
4714
4813
  } else { stableCount = 0; lastHeight = h; }
4715
4814
 
4716
4815
  // bottom reached
4717
- if (window.innerHeight + window.scrollY >= h - 2) { resolve({ stopped_reason: 'bottom', scrolls: scrolls + 1, finalScrollY: Math.round(window.scrollY) }); return; }
4816
+ if (s.view() + s.pos() >= h - 2) { done('bottom', scrolls + 1); return; }
4718
4817
  }
4719
- resolve({ stopped_reason: 'max_scrolls', scrolls, finalScrollY: Math.round(window.scrollY) });
4818
+ done('max_scrolls', scrolls);
4720
4819
  })();
4721
4820
  }),
4722
- args: [until, selector || null, max_scrolls, step_px || null, settle_ms],
4821
+ args: [until, selector || null, container || null, max_scrolls, step_px || null, settle_ms],
4723
4822
  world: 'MAIN',
4724
- });
4725
- return results?.[0]?.result ?? { stopped_reason: 'error', scrolls: 0 };
4823
+ }).then((results) => results?.[0]?.result ?? { stopped_reason: 'error', scrolls: 0 });
4824
+ // Stesse sleep in pagina delle attese: su una scheda nascosta rallentano.
4825
+ const deadline = Math.min(55000, max_scrolls * (settle_ms + 250) + 5000);
4826
+ return withDeadline(tabId, deadline, run,
4827
+ { stopped_reason: 'deadline', error: `Scrolling did not finish within ${deadline}ms` },
4828
+ (r) => r.stopped_reason === 'deadline');
4726
4829
  }
4727
4830
 
4728
4831
  // --- Avvia la connessione + registra l'instrumentation ---
package/package.json CHANGED
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "name": "chrome-bridge-mcp",
3
3
  "mcpName": "io.github.frsorrentino/chrome-bridge",
4
- "version": "1.17.0",
4
+ "version": "1.19.0",
5
5
  "type": "module",
6
6
  "description": "MCP server with 59 tools for browser automation via Chrome extension. Works on ChromeOS/Crostini.",
7
7
  "main": "server/index.js",
package/server/cli.js CHANGED
@@ -119,10 +119,13 @@ function connect(port) {
119
119
  sendCommand(type, params = {}) {
120
120
  return new Promise((res, rej) => {
121
121
  const command = createCommand(type, params);
122
+ // Come ws-manager: un timeout chiesto dal comando alza quello di trasporto.
123
+ const asked = Number(params?.timeout);
124
+ const timeout = Number.isFinite(asked) && asked > 0 ? Math.max(getTimeout(type), asked + 5000) : getTimeout(type);
122
125
  const timer = setTimeout(() => {
123
126
  pending.delete(command.id);
124
- rej(new Error(`Command ${type} timed out after ${getTimeout(type)}ms`));
125
- }, getTimeout(type));
127
+ rej(new Error(`Command ${type} timed out after ${timeout}ms`));
128
+ }, timeout);
126
129
  pending.set(command.id, { res, rej, timer });
127
130
  ws.send(JSON.stringify(command));
128
131
  });
package/server/index.js CHANGED
@@ -64,6 +64,10 @@ async function main() {
64
64
  'For more than one field use fill_form once (with submit_selector to submit in the same call) instead of repeated type_text: one turn instead of N.',
65
65
  'For tables use extract_table (server-side where/columns filtering) or extract, never read_page: read_page on a big table costs tens of thousands of tokens for data you filter anyway.',
66
66
  'To read detail in a screenshot, crop and zoom with element_screenshot (selector or region, scale) instead of taking another full one; to check an outcome, assert or wait_for poll for you — one call, no screenshot.',
67
+ // Il blocco più frequente sul campo (Meta Ads Manager, 23/09/2026): tre
68
+ // attese scadute e due screenshot falliti prima di capire che la
69
+ // finestra era dietro un'altra.
70
+ 'A tab in a minimized, covered or background window does not render: screenshots fail and page timers slow down. get_page_info reports visibility; page_hidden in a result means the same. Bring the window on screen, or create_tab new_window with bounds.',
67
71
  ].join(' '),
68
72
  });
69
73
 
@@ -92,12 +92,15 @@ export const MessageType = Object.freeze({
92
92
  });
93
93
 
94
94
  // Versione
95
- export const VERSION = '1.17.0';
95
+ export const VERSION = '1.19.0';
96
96
 
97
97
  // Configurazione
98
98
  export const DEFAULT_PORT = 8765;
99
99
  export const COMMAND_TIMEOUT_MS = 30000; // 30s per comandi normali
100
- export const SCREENSHOT_TIMEOUT_MS = 10000; // 10s per screenshot
100
+ // Più del timer interno di captureVisible (10 s, dopo il pacer da 520 ms):
101
+ // a parità vinceva il trasporto col messaggio generico «the tab may be busy»,
102
+ // e quello utile («the window is not rendering frames») non arrivava mai.
103
+ export const SCREENSHOT_TIMEOUT_MS = 15000;
101
104
  export const PING_INTERVAL_MS = 15000; // 15s heartbeat
102
105
  export const IDENT_TIMEOUT_MS = 5000; // tempo max per identificarsi
103
106
  export const PENDING_RELAY_TTL_MS = 150000; // deve superare il timeout comando più lungo (120s full_page_screenshot)
package/server/tools.js CHANGED
@@ -113,6 +113,34 @@ function fitArray(items, max, serialize) {
113
113
  return items.slice(0, lo);
114
114
  }
115
115
 
116
+ /**
117
+ * Output di execute_js sotto `max` senza rompere il JSON: tagliare la
118
+ * serializzazione a metà stringa dava «Invalid control character» a chi la
119
+ * parsava. Una stringa si accorcia dentro il suo valore, un array per elementi
120
+ * (jsonText), il resto diventa un prefisso dichiarato come tale.
121
+ */
122
+ export function jsResultText(data, max) {
123
+ const text = JSON.stringify(data);
124
+ if (text == null || text.length <= max) return text;
125
+ const hint = 'raise max_length, or save_to for the whole result';
126
+ const value = data?.result;
127
+ const wrap = (field, cut, total) => JSON.stringify({ [field]: cut, truncated: true, total_chars: total, hint });
128
+ const fitString = (field, full) => {
129
+ // Gli escape JSON allungano la stringa: il prefisso più lungo che sta nel
130
+ // tetto si cerca sulla serializzazione, non sui caratteri grezzi.
131
+ let lo = 0;
132
+ let hi = Math.min(full.length, max);
133
+ while (lo < hi) {
134
+ const mid = Math.ceil((lo + hi) / 2);
135
+ if (wrap(field, full.slice(0, mid), full.length).length <= max) lo = mid; else hi = mid - 1;
136
+ }
137
+ return wrap(field, full.slice(0, lo), full.length);
138
+ };
139
+ if (typeof value === 'string') return fitString('result', value);
140
+ if (Array.isArray(value)) return jsonText(data, max, hint);
141
+ return fitString('result_json_prefix', JSON.stringify(value));
142
+ }
143
+
116
144
  /** Una riga (oggetto colonna→valore o array di celle) soddisfa il filtro where. */
117
145
  function tableRowMatches(row, where) {
118
146
  return Object.entries(where).every(([col, needle]) => {
@@ -444,17 +472,73 @@ export function registerTools(server, wsManager, caps = 'all', options = {}) {
444
472
  }
445
473
  };
446
474
 
447
- // Mappa ref → selector per tab, popolata da get_interactives.
448
- // Permette click/type_text/hover per ref (n1, n2…) senza ripetere selettori lunghi.
475
+ // Registro ref → selector per tab, popolato da get_interactives, dalla
476
+ // preview di navigate e dai vicini di find_text. Permette click/type_text/
477
+ // hover per ref (n1, n2…) senza ripetere selettori lunghi.
478
+ // Un selettore tiene lo stesso ref finché la pagina non cambia, e un numero
479
+ // non viene mai riassegnato: prima ogni chiamata di scoperta ripartiva da n1
480
+ // sulla stessa mappa, e dopo un find_text n3 indicava un altro elemento —
481
+ // il click andava a segno sull'elemento sbagliato senza errori.
449
482
  const interactivesRefs = new Map();
483
+ const REFS_MAX = 2000;
450
484
 
451
485
  const refsKey = (tab_id) => tab_id ?? sessionTabId ?? 'active';
452
486
 
487
+ function refRegistry(tab_id) {
488
+ const key = refsKey(tab_id);
489
+ let reg = interactivesRefs.get(key);
490
+ if (!reg) {
491
+ reg = { next: 1, floor: 1, bySel: new Map(), byRef: new Map() };
492
+ interactivesRefs.set(key, reg);
493
+ }
494
+ return reg;
495
+ }
496
+
497
+ // Assegna i ref agli elementi (in place). Solo gli elementi con selector
498
+ // ricevono un ref: senza, non ci sarebbe nulla da risolvere.
499
+ function assignRefs(tab_id, elements) {
500
+ const reg = refRegistry(tab_id);
501
+ for (const e of elements) {
502
+ if (!e.selector) continue;
503
+ let ref = reg.bySel.get(e.selector);
504
+ if (!ref) {
505
+ ref = `n${reg.next++}`;
506
+ reg.bySel.set(e.selector, ref);
507
+ reg.byRef.set(ref, e.selector);
508
+ }
509
+ e.ref = ref;
510
+ }
511
+ while (reg.byRef.size > REFS_MAX) {
512
+ const [oldRef, oldSel] = reg.byRef.entries().next().value;
513
+ reg.byRef.delete(oldRef);
514
+ reg.bySel.delete(oldSel);
515
+ }
516
+ return reg.byRef.size;
517
+ }
518
+
519
+ // Una navigazione cambia pagina: i selettori vecchi potrebbero trovare altro.
520
+ // La mappa si svuota ma il contatore no, così un ref vecchio dà errore
521
+ // invece di indicare un elemento nuovo con lo stesso numero.
522
+ function resetRefs(tab_id) {
523
+ const reg = refRegistry(tab_id);
524
+ reg.bySel.clear();
525
+ reg.byRef.clear();
526
+ reg.floor = reg.next;
527
+ }
528
+
453
529
  function resolveTarget(selector, ref, tab_id) {
454
530
  if (selector) return selector;
455
531
  if (ref) {
456
- const sel = interactivesRefs.get(refsKey(tab_id))?.get(ref);
457
- if (!sel) throw new Error(`Unknown ref ${ref} — run get_interactives first`);
532
+ const reg = interactivesRefs.get(refsKey(tab_id));
533
+ const sel = reg?.byRef.get(ref);
534
+ if (!sel) {
535
+ const num = Number(String(ref).replace(/^n/, ''));
536
+ const why = !reg || reg.next === 1 ? 'no refs issued for this tab yet'
537
+ : num >= reg.next ? `refs issued so far go up to n${reg.next - 1}`
538
+ : num < reg.floor ? 'it was issued before the last navigate, which resets refs'
539
+ : 'it was evicted from the ref cache';
540
+ throw new Error(`Unknown ref ${ref} (${why}) — run get_interactives or find_text again`);
541
+ }
458
542
  return sel;
459
543
  }
460
544
  throw new Error('Either selector or ref is required');
@@ -489,13 +573,9 @@ export function registerTools(server, wsManager, caps = 'all', options = {}) {
489
573
  async function interactivesPreview(tab_id, limit = 12) {
490
574
  try {
491
575
  const data = await send(MessageType.GET_INTERACTIVES, { limit, visible_only: true, tab_id });
492
- const refMap = new Map();
493
- (data?.elements ?? []).forEach((e, i) => {
494
- e.ref = `n${i + 1}`;
495
- if (e.selector) refMap.set(e.ref, e.selector);
496
- });
497
- interactivesRefs.set(refsKey(tab_id), refMap);
498
- return refMap.size ? truncateText(interactivesLines(data), 1500) : null;
576
+ const elements = data?.elements ?? [];
577
+ assignRefs(tab_id, elements);
578
+ return elements.some((e) => e.ref) ? truncateText(interactivesLines(data), 1500) : null;
499
579
  } catch { return null; }
500
580
  }
501
581
 
@@ -534,21 +614,38 @@ export function registerTools(server, wsManager, caps = 'all', options = {}) {
534
614
  server.tool(
535
615
  'get_tabs',
536
616
  'List every open tab with id, url, title, active flag and mine (created by this session). Read-only. Find a tab_id when the implicit '
537
- + 'target is not the one you mean. include_windows adds position, size, state and type of each window — needed before moving or tiling.',
617
+ + 'target is not the one you mean. include_windows adds position, size, state and type of each window — needed before moving or tiling. '
618
+ + 'Session tabs gone since the last call come back once in closed_session_tabs with the reason (closed, window_closed, replaced, closed_by_bridge).',
538
619
  {
539
620
  include_windows: z.boolean().optional().default(false)
540
621
  .describe('Also return the windows with bounds, state, type and tab count'),
541
622
  },
542
623
  async ({ include_windows }) => {
543
- const data = await send(MessageType.GET_TABS, { include_windows: include_windows === true });
624
+ // `ended` = le schede che il server sa sue: l'estensione dà il motivo di
625
+ // quelle che non trova più. Senza schede di sessione la risposta resta
626
+ // quella di sempre.
627
+ const data = await send(MessageType.GET_TABS, {
628
+ include_windows: include_windows === true,
629
+ ...(ownedTabs.size ? { ended: [...ownedTabs] } : {}),
630
+ });
544
631
  const list = Array.isArray(data) ? data : (data?.tabs ?? []);
632
+ const windows = Array.isArray(data) ? null : data?.windows;
545
633
  for (const t of list) if (ownedTabs.has(t.id)) t.mine = true;
546
- return {
547
- content: [{
548
- type: 'text',
549
- text: jsonText(data),
550
- }],
551
- };
634
+ const missing = [...ownedTabs].filter((id) => !list.some((t) => t.id === id));
635
+ const shaped = windows ? { tabs: list, windows } : list;
636
+ if (!missing.length) return { content: [{ type: 'text', text: jsonText(shaped) }] };
637
+
638
+ // Un'estensione precedente ignora `ended` e risponde con l'array: motivo ignoto.
639
+ const ended = (!Array.isArray(data) && data?.ended) || {};
640
+ const closed = missing.map((id) => ({ id, ...(ended[id] ?? { reason: 'unknown' }) }));
641
+ for (const c of closed) {
642
+ ownedTabs.delete(c.id);
643
+ const successor = c.replaced_by != null ? list.find((t) => t.id === c.replaced_by) : null;
644
+ // Scheda sostituita da Chrome: resta della sessione con il nuovo id.
645
+ if (successor) { ownedTabs.add(successor.id); successor.mine = true; }
646
+ if (sessionTabId === c.id) sessionTabId = successor?.id ?? null;
647
+ }
648
+ return { content: [{ type: 'text', text: jsonText({ tabs: list, ...(windows ? { windows } : {}), closed_session_tabs: closed }) }] };
552
649
  }
553
650
  );
554
651
 
@@ -567,6 +664,7 @@ export function registerTools(server, wsManager, caps = 'all', options = {}) {
567
664
  const data = await send(MessageType.NAVIGATE, { url, tab_id });
568
665
  // Il tab navigato diventa il default di sessione per i comandi successivi
569
666
  if (data?.tabId != null) sessionTabId = data.tabId;
667
+ resetRefs(data?.tabId ?? tab_id);
570
668
  const preview = await interactivesPreview(data?.tabId ?? tab_id);
571
669
  return {
572
670
  content: [{
@@ -578,6 +676,8 @@ export function registerTools(server, wsManager, caps = 'all', options = {}) {
578
676
  );
579
677
 
580
678
  // --- screenshot ---
679
+ // Larghezze dei preset, le stesse di viewport_resize nell'estensione.
680
+ const PRESET_WIDTHS = { mobile: 375, tablet: 768, desktop: 1440 };
581
681
  server.tool(
582
682
  'screenshot',
583
683
  'Screenshot of the visible viewport only (PNG), at the current scroll position, or one per viewport preset with presets. Read-only. '
@@ -586,7 +686,7 @@ export function registerTools(server, wsManager, caps = 'all', options = {}) {
586
686
  {
587
687
  tab_id: tabId,
588
688
  save_to: saveToField('the PNG'),
589
- presets: z.array(z.enum(['mobile', 'tablet', 'desktop'])).optional().describe('One capture per preset, window restored; save_to = directory, one file each'),
689
+ presets: z.array(z.enum(['mobile', 'tablet', 'desktop'])).optional().describe('Resizes the window per preset, then restores it; not phone emulation. A refused width is reported, not shot. save_to = directory'),
590
690
  },
591
691
  async ({ tab_id, save_to, presets }) => {
592
692
  guardWrite(save_to);
@@ -598,10 +698,24 @@ export function registerTools(server, wsManager, caps = 'all', options = {}) {
598
698
  const notes = [];
599
699
  try {
600
700
  for (const preset of presets) {
601
- await send(MessageType.VIEWPORT_RESIZE, { preset, tab_id });
701
+ const resized = await send(MessageType.VIEWPORT_RESIZE, { preset, tab_id });
602
702
  await new Promise((r) => setTimeout(r, 400));
603
703
  const shot = await send(MessageType.SCREENSHOT, { tab_id });
604
704
  if (!shot?.image) { notes.push(`${preset}: no image`); continue; }
705
+ // presets ridimensiona la finestra, non emula un dispositivo: il
706
+ // window manager può imporre una larghezza minima o ignorare la
707
+ // richiesta. Un'immagine desktop etichettata «mobile» porta a
708
+ // conclusioni sbagliate (2226 px su med-systems.it, 23/09/2026):
709
+ // meglio nessuna immagine e una nota esplicita.
710
+ const got = shot.viewport?.width;
711
+ const want = PRESET_WIDTHS[preset];
712
+ if (got && want && Math.abs(got - want) > Math.max(40, want * 0.15)) {
713
+ const zoom = resized?.zoom && resized.zoom !== 1 ? `, page zoom ${Math.round(resized.zoom * 100)}%` : '';
714
+ notes.push(`${preset}: NOT APPLIED — viewport stayed ${got}×${shot.viewport.height} CSS px (asked ~${want}${zoom}); `
715
+ + 'the window manager refused the width. No capture saved: this would not be a ' + preset + ' rendering. '
716
+ + 'Device emulation needs a headless browser.');
717
+ continue;
718
+ }
605
719
  if (save_to) {
606
720
  await mkdir(save_to, { recursive: true });
607
721
  const path = join(save_to, `${preset}.png`);
@@ -645,21 +759,25 @@ export function registerTools(server, wsManager, caps = 'all', options = {}) {
645
759
  // --- execute_js ---
646
760
  server.tool(
647
761
  'execute_js',
648
- 'Run JavaScript in the page (MAIN world). Requires the extension\'s "Allow user scripts" toggle; errors explain setup if disabled.',
762
+ 'Run JavaScript in the page (MAIN world). Requires the extension\'s "Allow user scripts" toggle; errors explain setup if disabled. '
763
+ + 'A promise is awaited up to timeout. A cut result stays valid JSON and says truncated; save_to writes the whole result to disk.',
649
764
  {
650
765
  code: z.string().describe('JS evaluated in the page; the value of the last expression is returned'),
651
766
  tab_id: tabId,
652
767
  frame_id: frameId,
653
768
  max_length: z.number().optional().default(20000).describe('Max output chars'),
769
+ timeout: z.number().optional().describe('Max ms to wait for the result (default 30000); raise it for long async work'),
770
+ save_to: saveToField('the result (a string as is, anything else as JSON)'),
654
771
  },
655
- async ({ code, tab_id, frame_id, max_length }) => {
656
- const data = await send(MessageType.EXECUTE_JS, { code, tab_id, frame_id });
657
- return {
658
- content: [{
659
- type: 'text',
660
- text: truncateText(JSON.stringify(data), max_length ?? DEFAULT_MAX_OUTPUT, 'max_length'),
661
- }],
662
- };
772
+ async ({ code, tab_id, frame_id, max_length, timeout, save_to }) => {
773
+ guardWrite(save_to);
774
+ const data = await send(MessageType.EXECUTE_JS, { code, tab_id, frame_id, ...(timeout ? { timeout } : {}) });
775
+ if (save_to) {
776
+ const value = data?.result;
777
+ const isString = typeof value === 'string';
778
+ return savedSummary(save_to, Buffer.from(isString ? value : JSON.stringify(value ?? null), 'utf8'), { format: isString ? 'text' : 'json' });
779
+ }
780
+ return { content: [{ type: 'text', text: jsResultText(data, max_length ?? DEFAULT_MAX_OUTPUT) }] };
663
781
  }
664
782
  );
665
783
 
@@ -1034,12 +1152,13 @@ export function registerTools(server, wsManager, caps = 'all', options = {}) {
1034
1152
  max_scrolls: z.number().optional().default(20).describe('Cap on scroll steps, so an infinite feed terminates'),
1035
1153
  step_px: z.number().optional().describe('px per step, default viewport height'),
1036
1154
  settle_ms: z.number().optional().default(400).describe('Pause ms after each step'),
1155
+ container: z.string().optional().describe('until: selector that scrolls; default the document, or the largest scrollable element if the document does not scroll'),
1037
1156
  tab_id: tabId,
1038
1157
  frame_id: frameId,
1039
1158
  },
1040
- async ({ action, selector, x, y, behavior, offset_y, until, max_scrolls, step_px, settle_ms, tab_id, frame_id }) => {
1159
+ async ({ action, selector, x, y, behavior, offset_y, until, max_scrolls, step_px, settle_ms, container, tab_id, frame_id }) => {
1041
1160
  const data = (action ?? 'to') === 'until'
1042
- ? await send(MessageType.SCROLL_UNTIL, { until, selector, max_scrolls, step_px, settle_ms, tab_id })
1161
+ ? await send(MessageType.SCROLL_UNTIL, { until, selector, container, max_scrolls, step_px, settle_ms, tab_id })
1043
1162
  : await send(MessageType.SCROLL_TO, { selector, x, y, behavior, offset_y, tab_id, frame_id });
1044
1163
  return { content: [{ type: 'text', text: jsonText(data) }] };
1045
1164
  }
@@ -1367,6 +1486,7 @@ export function registerTools(server, wsManager, caps = 'all', options = {}) {
1367
1486
  return { content: [{ type: 'text', text: jsonText({ closed }) }] };
1368
1487
  }
1369
1488
  const data = await send(MessageType.TAB_ACTION, { action, bypass_cache, tab_id });
1489
+ if (action === 'close') ownedTabs.delete(data?.closed ?? tab_id ?? sessionTabId);
1370
1490
  if (action === 'close' && (tab_id == null || tab_id === sessionTabId)) sessionTabId = null;
1371
1491
  if (action === 'activate' && tab_id != null) sessionTabId = tab_id;
1372
1492
  return { content: [{ type: 'text', text: jsonText(data) }] };
@@ -1452,12 +1572,7 @@ export function registerTools(server, wsManager, caps = 'all', options = {}) {
1452
1572
  .slice(0, 5)
1453
1573
  .map((c) => c.e);
1454
1574
  if (els.length) {
1455
- const refMap = new Map();
1456
- els.forEach((e, i) => {
1457
- e.ref = `n${i + 1}`;
1458
- if (e.selector) refMap.set(e.ref, e.selector);
1459
- });
1460
- interactivesRefs.set(refsKey(tab_id), refMap);
1575
+ assignRefs(tab_id, els);
1461
1576
  near = truncateText(interactivesLines({ count: els.length, elements: els, note: 'near first match' }), 1200);
1462
1577
  }
1463
1578
  } catch {}
@@ -2150,13 +2265,8 @@ export function registerTools(server, wsManager, caps = 'all', options = {}) {
2150
2265
  },
2151
2266
  async ({ scope, limit, visible_only, format, frame_id, tab_id }) => {
2152
2267
  const data = await send(MessageType.GET_INTERACTIVES, { scope, limit, visible_only, frame_id, tab_id });
2153
- // Assegna ref n1..nN e memorizza la mappa ref → selector per click/type_text/hover
2154
- const refMap = new Map();
2155
- (data?.elements ?? []).forEach((e, i) => {
2156
- e.ref = `n${i + 1}`;
2157
- if (e.selector) refMap.set(e.ref, e.selector);
2158
- });
2159
- interactivesRefs.set(refsKey(tab_id), refMap);
2268
+ // Ref stabili per selettore, memorizzati per click/type_text/hover
2269
+ assignRefs(tab_id, data?.elements ?? []);
2160
2270
  if ((format ?? 'lines') === 'json') {
2161
2271
  return { content: [{ type: 'text', text: jsonText(data) }] };
2162
2272
  }
@@ -150,8 +150,10 @@ highlighted diff; zoom into a region with `element_screenshot({region, scale:2})
150
150
  ### Three viewports
151
151
  Triggers: "check it on mobile/tablet/desktop", «com'è su telefono?».
152
152
  `screenshot({presets:['mobile','tablet','desktop'], save_to:'./shots'})`: one
153
- call, one file per viewport, window restored. Report overflow, overlapping
154
- elements, hidden CTAs. Dark mode: `emulate_media({colorScheme:'dark'})`.
153
+ call, one file per viewport, window restored. Presets resize the window and
154
+ do not emulate a phone: a width the window manager refuses comes back as
155
+ `NOT APPLIED`, with no image — say so instead of judging the mobile layout.
156
+ Report overflow, overlapping elements, hidden CTAs. Dark mode: `emulate_media({colorScheme:'dark'})`.
155
157
  Print stylesheet: `emulate_media({printMode:true})` then `screenshot`.
156
158
 
157
159
  ### Accessibility and keyboard navigation
@@ -264,6 +266,21 @@ Triggers: "download the orders export and tell me…".
264
266
  `click` the export button → `manage_downloads({action:'wait_for_complete'})` → read the file
265
267
  with Claude Code. Spam in signups: `extract_table({where:{email:'/\\.ru$|xn--/'}})`.
266
268
 
269
+ ### Heavy web apps (Meta Ads Manager, Google Ads, big back offices)
270
+ Triggers: "change the campaign budget", "read the ad set breakdown", «modifica
271
+ la campagna su Ads Manager», «leggi i dati dell'inserzione».
272
+ Keep the tab visible first: `get_page_info` must say `visibility: 'visible'`,
273
+ else `create_tab({url, new_window:true, left, top, width, height})` on screen.
274
+ Menus without ARIA roles: `find_text({text})` on the label, not
275
+ `get_interactives` scoped by role. The app scrolls inside a panel:
276
+ `scroll({action:'until', container})` (it picks the largest scrollable panel
277
+ when the page itself does not scroll). Virtualized lists render the next rows
278
+ only after a frame: scroll in one call, click in the next. Fields that save on
279
+ Enter: `type_text` then `press_key({key:'Enter'})`. Several clicks on a React
280
+ UI: one call per click, or `execute_js` with about 350 ms between them. A
281
+ button that opens `window.open('')` navigates the tab away: override
282
+ `window.open` with `execute_js` before clicking.
283
+
267
284
  ### Email deliverability
268
285
  Triggers: "do the site emails land in spam?". `navigate('https://www.mail-tester.com')`
269
286
  → `clipboard({action:'read'})` after copying the address → send from the site