@remcp/runtime 0.2.177 → 0.2.178

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@remcp/runtime",
-  "version": "0.2.177",
+  "version": "0.2.178",
   "description": "First-party ReMCP local device runtime: filesystem, terminal, computer-use, browser, diagnostics and document tools over MCP.",
   "type": "module",
   "license": "MIT",

package/src/extended/browser-devtools.mjs

@@ -3,6 +3,7 @@
 import { jsonResult, clamp, requireEnum } from './common.mjs';
 import { withTarget } from './browser-session.mjs';
 import { inspectNativeBrowserAudit } from './browser-native-audit.mjs';
+import { inspectDomGeometry } from './browser-geometry.mjs';
 
 const METRIC_NAMES = new Set([
   'Timestamp','Documents','Frames','JSEventListeners','Nodes','LayoutCount',
@@ -41,6 +42,7 @@ export async function sampleBrowserDevtools(session, action, durationMs, maxEven
     if (events.length < maxEvents) events.push(event);
   }
   if (action === 'layout' || action === 'accessibility') return inspectNativeBrowserAudit(session, action);
+  if (action === 'geometry') return inspectDomGeometry(session);
   if (action === 'performance') {
     await session.send('Performance.enable');
     try {
@@ -99,11 +101,16 @@ export async function sampleBrowserDevtools(session, action, durationMs, maxEven
   return { action, sampled_ms:durationMs, counts, events, truncated:Object.values(counts).reduce((a,b) => a+b,0) > maxEvents,
     note:'Live events only; DevTools does not backfill events recorded before sampling. Request/response contents and headers are excluded.' };
 }
+export function buildDevtoolsReport(target, action, result) {
+  return { target_id:target.id,
+    ...(action === 'geometry' ? {} : {url:safeDevtoolsUrl(target.url)}),
+    ...result };
+}
 export async function browserDevtools(args = {}) {
-  const action = requireEnum(args.action, 'action', ['console','network','performance','layout','accessibility']);
+  const action = requireEnum(args.action, 'action', ['console','network','performance','layout','accessibility','geometry']);
   const durationMs = clamp(args.sample_ms, 1500, 250, 5000);
   const maxEvents = clamp(args.max_events, 50, 1, 100);
   return withTarget(args, async (session,target) =>
-    jsonResult({ target_id:target.id, url:safeDevtoolsUrl(target.url),
-      ...(await sampleBrowserDevtools(session,action,durationMs,maxEvents)) }));
+    jsonResult(buildDevtoolsReport(target, action,
+      await sampleBrowserDevtools(session,action,durationMs,maxEvents))));
 }

package/src/extended/browser-geometry.mjs

@@ -0,0 +1,95 @@
+// Local-only Chrome DOMSnapshot geometry aggregation.
+// DOMSnapshot can contain page text and attributes: NEVER serialize, log or return it.
+// The result contains counts and bounded geometry of tag types, not DOM content.
+const ALLOWED_TAGS = new Set([
+  'A','ARTICLE','ASIDE','BLOCKQUOTE','BUTTON','CANVAS','CODE','DD','DETAILS','DIV',
+  'DL','DT','FIELDSET','FIGCAPTION','FIGURE','FOOTER','FORM','H1','H2','H3','H4',
+  'H5','H6','HEADER','IFRAME','IMG','INPUT','LABEL','LI','MAIN','NAV','OL','OPTION',
+  'P','PICTURE','PRE','SECTION','SELECT','SPAN','SVG','TABLE','TBODY','TD',
+  'TEXTAREA','TH','THEAD','TR','UL','VIDEO',
+]);
+const MAX_LAYOUT_NODES = 12000;
+const MAX_CANDIDATES = 8;
+const EPSILON = 1;
+function decimal(value) {
+  return typeof value === 'number' && Number.isFinite(value)
+    ? Math.max(-10000000, Math.min(10000000, Math.round(value * 100) / 100))
+    : null;
+}
+function typeForNode(doc, strings, nodeIndex) {
+  const idx = doc?.nodes?.nodeName?.[nodeIndex];
+  if (!Number.isSafeInteger(idx) || !Array.isArray(strings)) return '';
+  const str = strings[idx];
+  if (typeof str !== 'string' || str.length > 20) return '';
+  const upper = str.toUpperCase();
+  return ALLOWED_TAGS.has(upper) ? upper.toLowerCase() : '';
+}
+export function summarizeDomGeometry(snapshot, pageMetrics, { maxNodes = MAX_LAYOUT_NODES } = {}) {
+  const rawDocs = Array.isArray(snapshot?.documents) ? snapshot.documents : [];
+  const doc = rawDocs[0] || {};
+  const indices = Array.isArray(doc?.layout?.nodeIndex) ? doc.layout.nodeIndex : [];
+  const bounds = Array.isArray(doc?.layout?.bounds) ? doc.layout.bounds : [];
+  const width = decimal(pageMetrics?.cssLayoutViewport?.clientWidth);
+  const scrollX = decimal(pageMetrics?.cssLayoutViewport?.pageX) ?? 0;
+  if (width == null || width <= 0) throw new Error('Chrome did not report a valid layout viewport width');
+  const left = scrollX, right = scrollX + width;
+  const count = Math.min(MAX_LAYOUT_NODES, Math.max(0,Math.floor(Number(maxNodes) || MAX_LAYOUT_NODES)),indices.length,bounds.length);
+  let eligible = 0, candidateCount = 0, leftOverflowCount = 0, rightOverflowCount = 0;
+  let biggestOverflow = 0;
+  const byTag = Object.create(null);
+  const top = [];
+  for (let i = 0; i < count; i++) {
+    const tag = typeForNode(doc,snapshot?.strings,indices[i]);
+    if (!tag) continue;
+    const box = bounds[i];
+    if (!Array.isArray(box) || box.length < 4) continue;
+    const [x,y,w,h] = box.map(decimal);
+    if ([x,y,w,h].some(v => v == null) || w <= 0 || h <= 0) continue;
+    eligible++;
+    const extraLeft = Math.max(0,left-x);
+    const extraRight = Math.max(0,x+w-right);
+    if (extraLeft <= EPSILON && extraRight <= EPSILON) continue;
+    candidateCount++;
+    if (extraLeft > EPSILON) leftOverflowCount++;
+    if (extraRight > EPSILON) rightOverflowCount++;
+    const overflow = Math.ceil(Math.max(extraLeft,extraRight));
+    biggestOverflow = Math.max(biggestOverflow,overflow);
+    byTag[tag] = (byTag[tag] || 0) + 1;
+    // Never send selectors, attributes, text, sources, titles or URLs.
+    const candidate = { tag, side:extraLeft > EPSILON && extraRight > EPSILON ? 'both' :
+      extraLeft > EPSILON ? 'left' : 'right', overflow_px:overflow,
+      bounds: { x, y, width:w, height:h } };
+    if (top.length < MAX_CANDIDATES) top.push(candidate);
+    else {
+      let smallest = 0;
+      for (let j=1; j<top.length; j++) if (top[j].overflow_px < top[smallest].overflow_px) smallest = j;
+      if (candidate.overflow_px > top[smallest].overflow_px) top[smallest] = candidate;
+    }
+  }
+  top.sort((a,b)=>b.overflow_px-a.overflow_px);
+  return {
+    viewport_width_px:width,
+    sampled_layout_nodes:count,
+    eligible_elements:eligible,
+    candidate_count:candidateCount,
+    left_overflow_count:leftOverflowCount,
+    right_overflow_count:rightOverflowCount,
+    maximum_overflow_px:biggestOverflow,
+    by_tag:Object.fromEntries(Object.entries(byTag).sort((a,b)=>b[1]-a[1]).slice(0,20)),
+    candidates:top,
+    frames_omitted:Math.max(0,rawDocs.length-1),
+    truncated:count<indices.length || count<bounds.length,
+    note:'Only numeric geometry and allowlisted HTML tag types leave the device. Candidate bounds may be intentional (for example carousels or offscreen animations); this is not proof of a layout bug. Only the main frame is inspected.',
+  };
+}
+export async function inspectDomGeometry(session) {
+  // Fetch just enough geometry for local assessment. The CDP transport enforces
+  // its own 16 MiB message bound; snapshot text never leaves this process.
+  const [metrics,snapshot] = await Promise.all([
+    session.send('Page.getLayoutMetrics'),
+    session.send('DOMSnapshot.captureSnapshot',{
+      computedStyles:[],includePaintOrder:false,includeDOMRects:false,
+    }),
+  ]);
+  return {action:'geometry',geometry:summarizeDomGeometry(snapshot,metrics)};
+}

package/src/extended/catalog.mjs

@@ -167,7 +167,7 @@ export const extendedToolDefinitions = [
   define('browser_find', 'Find browser element', 'Find visible DOM elements by CSS selector, text, or ARIA role and return reusable selectors plus text, value, and bounding boxes.', o({ ...browserTarget, selector:s('CSS selector.'), text:s('Visible text substring.'), role:s('ARIA role.'), limit:n() }), readOnlyOpen, browserHandlers.browser_find, ['browser_cdp']),
   define('browser_action', 'Browser element action', 'Click, focus, type/set a value, fill up to 20 form fields in one guarded call, select, scroll a page or scrollable element, scroll an element into view, upload files, press a key, or set an exact responsive-test viewport in a debuggable browser page using DOM/CDP semantics.', o({ ...browserTarget, action:e(['click','focus','type','set_value','select','scroll','scroll_into_view','upload','press','set_viewport','fill_form','workflow']), selector:s('CSS selector for the target element; prefer the reusable selector returned by browser_find.'), steps:{type:'array',minItems:1,maxItems:12,description:'Workflow on one page: explicit CSS actions and bounded waits, max 12 seconds. Never silently submits; click must be explicit.',items:{type:'object',additionalProperties:false,properties:{action:e(['fill_form','set_value','select','focus','click','wait_for']),selector:s('CSS selector for action or wait_for.'),value:{type:['string','number','boolean']},fields:{type:'array',description:'CSS-selected field edits within the workflow step; no implicit form submission.',minItems:1,maxItems:20,items:{type:'object',additionalProperties:false,properties:{selector:s(),value:{type:['string','number','boolean']}},required:['selector','value']}},condition:e(['selector','text','url_contains','load']),text:s('Text or URL fragment for wait condition.'),timeout_ms:n('Wait timeout 100..4000 ms; default 1500.')},required:['action']}}, fields:{type:'array',minItems:1,maxItems:20,items:{type:'object',additionalProperties:false,properties:{selector:s('CSS selector.'),value:{type:['string','boolean','number'],description:'New field value; not returned in results.'}},required:['selector','value']},description:'fill_form: 1–20 CSS form fields in one CDP call. Does not submit.'}, browser_text:s('Visible text used only to locate a DOM element; use this instead of overloading text for action=type.'), text:s('For action=type, Unicode text to insert. For non-type element actions without selector/browser_text, visible text may identify the target.'), clear:b('For action=type, clear the current value before inserting text.'), value:s('Value for action=set_value; also accepted as a legacy type payload when text is used as the locator.'), text_value:s('Compatibility alias for the type/set_value text value.'), option:s(), delta_x:n('Horizontal scroll delta; positive means right.'), delta_y:n('Vertical scroll delta; positive means down.'), delta:n('Alias for delta_y.'), direction:e(['up','down','left','right']), wheel_times:n('Number of 120-unit wheel steps for direction; 1..50.'), path:s(), paths:{type:'array',items:{type:'string'}}, key:s(), width:n('CSS viewport width for set_viewport.'), height:n('CSS viewport height for set_viewport.'), device_scale_factor:n('Device scale factor for set_viewport; default 1.'), mobile:b('Enable mobile emulation for set_viewport.') }, ['action']), openMutating, browserHandlers.browser_action, ['browser_cdp']),
   define('browser_wait', 'Wait for browser', 'Wait for DOM state, page text, URL, completed loading, a navigation away from the current URL, or a short network-idle period in a debuggable browser page. Use browser_evaluate separately for JavaScript predicates or page-specific inspection.', o({ ...browserTarget, condition:e(['selector','text','url_contains','load','navigation','network_idle']), selector:s(), text:s(), value:s(), poll_ms:n(), idle_ms:n('Required zero-in-flight network quiet window for network_idle; default 500 ms.') }), readOnlyOpen, browserHandlers.browser_wait, ['browser_cdp']),
-  define('browser_devtools', 'Browser DevTools diagnostics', 'Use this to read a bounded live sample of Chrome Console or Network events, or native Performance metrics through CDP; excludes cookies, request bodies, response bodies and authentication headers.', o({ ...browserTarget, action:e(['console','network','performance','layout','accessibility'],'Diagnostics mode. Layout reports CSS-pixel viewport/overflow; accessibility reports anonymized role counts and unlabeled controls.'), sample_ms:n('Live event capture period 250–5000 ms; default 1500. Ignored for performance, layout and accessibility.'), max_events:n('Maximum 1–100 events retained; default 50.') }, ['action']), readOnlyOpen, browserHandlers.browser_devtools, ['browser_cdp']),
+  define('browser_devtools', 'Browser DevTools diagnostics', 'Use this to inspect Chrome Console/Network live events, Performance, Layout, Accessibility or screenshot-free DOMSnapshot Geometry via native CDP. Network headers, bodies and cookies are excluded; geometry returns bounded numeric summaries without page text.', o({ ...browserTarget, action:e(['console','network','performance','layout','accessibility','geometry'],'Diagnostics mode. Layout measures CSS viewport, accessibility returns anonymized roles and unnamed controls, geometry audits DOMSnapshot bounds without page contents.'), sample_ms:n('Live event capture period 250–5000 ms; default 1500. Ignored for performance, layout, accessibility and geometry.'), max_events:n('Maximum 1–100 events retained; default 50.') }, ['action']), readOnlyOpen, browserHandlers.browser_devtools, ['browser_cdp']),
   define('browser_evaluate', 'Evaluate browser JavaScript', 'Evaluate JavaScript in a selected debuggable browser page and return its serializable value. This is a powerful escape hatch for page-specific automation.', o({ ...browserTarget, expression:s('JavaScript expression.'), await_promise:b('Await a returned Promise; default true.') }, ['expression']), openMutating, browserHandlers.browser_evaluate, ['browser_evaluate']),
 
   define('service', 'Service control', 'List, inspect, start, stop, or restart operating-system services using Service Control Manager, launchctl, or systemd. Mutating actions pass through ReMCP command policy.', o({ action:e(['list','status','start','stop','restart']), name:s('Service/unit/launchd label.'), scope:e(['user','system']) }, ['action']), mutating, diagnosticHandlers.service, ['services']),