browsentic 0.4.15 → 0.5.0

package/dist/cli.js

@@ -28284,6 +28284,54 @@ var focusInput = defineAction({
   }
 });
 
+// ../lib/actions/page/switch-frame.ts
+var switchFrame = defineAction({
+  name: "page.switchFrame",
+  description: 'Step into an <iframe> so that every later page action \u2014 reading, clicking, typing, waiting, site tools, injected code \u2014 runs inside it as if it were the page, or step back out. page.getPageInfo lists the frames the current document embeds under "frames"; pass one as "frame" to enter it, and call again from inside to go deeper. Call it with no arguments to return to the top document, or with to: "parent" to step out one level. The focus is per tab, clears when the tab navigates, and page.getPageInfo reports it under "frame" whenever it is not the top. A frame that forbids scripts (a sandbox without allow-scripts) or a browser-internal frame cannot be entered; page.screenshot and page.trustedClick by coordinates still reach those. page.screenshot and page.navigate always act on the whole tab.',
+  input: external_exports.object({
+    frame: targetSchema.optional().describe(
+      'The <iframe> or <frame> element to enter, matched inside the frame currently in focus \u2014 use a selector from the "frames" list of page.getPageInfo. Omit it to leave instead.'
+    ),
+    to: external_exports.enum(["top", "parent"]).default("top").describe('Where to go when no "frame" is given: "top" returns to the main document, "parent" steps out one level.')
+  }),
+  execute({ frame }) {
+    if (!frame) {
+      throw new ActionError("Leaving a frame is resolved by the Browsentic extension, not in the page", "UNSUPPORTED");
+    }
+    const el = resolveTarget(frame);
+    if (!isFrameElement(el)) {
+      throw new ActionError(
+        `${cssPath(el)} is a <${el.tagName.toLowerCase()}>, not an <iframe> \u2014 pick one of the frames page.getPageInfo lists`,
+        "INVALID_TARGET"
+      );
+    }
+    return {
+      frameId: frameIdOf(el),
+      src: el.src || void 0,
+      name: el.name || void 0,
+      sandbox: sandboxOf(el),
+      ...describeElement(el)
+    };
+  }
+});
+function isFrameElement(el) {
+  return el instanceof HTMLIFrameElement || el instanceof HTMLFrameElement;
+}
+function sandboxOf(el) {
+  return el instanceof HTMLIFrameElement && el.hasAttribute("sandbox") ? el.sandbox.value || "all restrictions" : void 0;
+}
+function frameIdOf(target) {
+  const scope = globalThis;
+  const id = (scope.browser?.runtime?.getFrameId ?? scope.chrome?.runtime?.getFrameId)?.(target);
+  if (typeof id !== "number" || id < 0) {
+    throw new ActionError(
+      "This browser cannot identify frames from the page \u2014 entering a frame needs Chrome 116 or Firefox 96 or newer",
+      "UNSUPPORTED"
+    );
+  }
+  return id;
+}
+
 // ../lib/actions/page/get-page-info.ts
 var IMPLICIT_ROLES = {
   header: "banner",
@@ -28308,7 +28356,7 @@ var LANDMARK_ROLES = /* @__PURE__ */ new Set([
 ]);
 var getPageInfo = defineAction({
   name: "page.getPageInfo",
-  description: "Snapshot the current page: document metadata, viewport and scroll state, a semantic layout tree with a text diagram, the heading outline, and an inventory of interactive elements \u2014 each carrying its ARIA role, its live state (disabled, checked, expanded, filled, aria-current) and the landmark region it sits in. When the site registers WebMCP tools, the result also carries a siteTools list \u2014 prefer page.callSiteTool over clicking wherever a listed tool covers the step.",
+  description: 'Snapshot the current page: document metadata, viewport and scroll state, a semantic layout tree with a text diagram, the heading outline, and an inventory of interactive elements \u2014 each carrying its ARIA role, its live state (disabled, checked, expanded, filled, aria-current) and the landmark region it sits in. When the site registers WebMCP tools, the result also carries a siteTools list \u2014 prefer page.callSiteTool over clicking wherever a listed tool covers the step. Visible iframes come back under "frames"; nothing inside one is in this snapshot until page.switchFrame enters it, and "frame" then says which one is in focus.',
   input: external_exports.object({
     maxPerKind: external_exports.number().int().positive().default(30).describe("Cap on links, buttons, fields, and forms listed per kind")
   }),
@@ -28337,10 +28385,23 @@ var getPageInfo = defineAction({
         level: Number(heading.tagName[1]),
         text: accessibleText(heading).slice(0, 120)
       })),
-      interactive: inventory(found, owners, maxPerKind)
+      interactive: inventory(found, owners, maxPerKind),
+      frames: embeddedFrames()
     };
   }
 });
+var MAX_FRAMES = 20;
+function embeddedFrames() {
+  const frames = [...document.querySelectorAll("iframe,frame")].filter(isFrameElement).filter(isExposed).slice(0, MAX_FRAMES).map((frame) => ({
+    selector: cssPath(frame),
+    src: frame.src || void 0,
+    name: frame.name || void 0,
+    title: frame.title || void 0,
+    sandbox: sandboxOf(frame),
+    bounds: documentBounds(frame)
+  }));
+  return frames.length ? frames : void 0;
+}
 function regionRole(el) {
   const explicit = el.getAttribute("role");
   if (explicit) return LANDMARK_ROLES.has(explicit) ? explicit : void 0;
@@ -29979,6 +30040,7 @@ var actions = new Map(
     navigate,
     openTab,
     switchTab,
+    switchFrame,
     closeTab,
     screenshot,
     listFiles,
@@ -34184,7 +34246,7 @@ function purgeNpxCache(entries) {
 // package.json
 var package_default = {
   name: "browsentic",
-  version: "0.4.15",
+  version: "0.5.0",
   description: "Hand your real, logged-in browser to the AI agent you already run. Installs the browser extension, runs the local daemon, and speaks MCP.",
   type: "module",
   license: "MIT",

package/dist/daemon-main.js

@@ -19876,6 +19876,54 @@ var focusInput = defineAction({
   }
 });
 
+// ../lib/actions/page/switch-frame.ts
+var switchFrame = defineAction({
+  name: "page.switchFrame",
+  description: 'Step into an <iframe> so that every later page action \u2014 reading, clicking, typing, waiting, site tools, injected code \u2014 runs inside it as if it were the page, or step back out. page.getPageInfo lists the frames the current document embeds under "frames"; pass one as "frame" to enter it, and call again from inside to go deeper. Call it with no arguments to return to the top document, or with to: "parent" to step out one level. The focus is per tab, clears when the tab navigates, and page.getPageInfo reports it under "frame" whenever it is not the top. A frame that forbids scripts (a sandbox without allow-scripts) or a browser-internal frame cannot be entered; page.screenshot and page.trustedClick by coordinates still reach those. page.screenshot and page.navigate always act on the whole tab.',
+  input: external_exports.object({
+    frame: targetSchema.optional().describe(
+      'The <iframe> or <frame> element to enter, matched inside the frame currently in focus \u2014 use a selector from the "frames" list of page.getPageInfo. Omit it to leave instead.'
+    ),
+    to: external_exports.enum(["top", "parent"]).default("top").describe('Where to go when no "frame" is given: "top" returns to the main document, "parent" steps out one level.')
+  }),
+  execute({ frame }) {
+    if (!frame) {
+      throw new ActionError("Leaving a frame is resolved by the Browsentic extension, not in the page", "UNSUPPORTED");
+    }
+    const el = resolveTarget(frame);
+    if (!isFrameElement(el)) {
+      throw new ActionError(
+        `${cssPath(el)} is a <${el.tagName.toLowerCase()}>, not an <iframe> \u2014 pick one of the frames page.getPageInfo lists`,
+        "INVALID_TARGET"
+      );
+    }
+    return {
+      frameId: frameIdOf(el),
+      src: el.src || void 0,
+      name: el.name || void 0,
+      sandbox: sandboxOf(el),
+      ...describeElement(el)
+    };
+  }
+});
+function isFrameElement(el) {
+  return el instanceof HTMLIFrameElement || el instanceof HTMLFrameElement;
+}
+function sandboxOf(el) {
+  return el instanceof HTMLIFrameElement && el.hasAttribute("sandbox") ? el.sandbox.value || "all restrictions" : void 0;
+}
+function frameIdOf(target) {
+  const scope = globalThis;
+  const id = (scope.browser?.runtime?.getFrameId ?? scope.chrome?.runtime?.getFrameId)?.(target);
+  if (typeof id !== "number" || id < 0) {
+    throw new ActionError(
+      "This browser cannot identify frames from the page \u2014 entering a frame needs Chrome 116 or Firefox 96 or newer",
+      "UNSUPPORTED"
+    );
+  }
+  return id;
+}
+
 // ../lib/actions/page/get-page-info.ts
 var IMPLICIT_ROLES = {
   header: "banner",
@@ -19900,7 +19948,7 @@ var LANDMARK_ROLES = /* @__PURE__ */ new Set([
 ]);
 var getPageInfo = defineAction({
   name: "page.getPageInfo",
-  description: "Snapshot the current page: document metadata, viewport and scroll state, a semantic layout tree with a text diagram, the heading outline, and an inventory of interactive elements \u2014 each carrying its ARIA role, its live state (disabled, checked, expanded, filled, aria-current) and the landmark region it sits in. When the site registers WebMCP tools, the result also carries a siteTools list \u2014 prefer page.callSiteTool over clicking wherever a listed tool covers the step.",
+  description: 'Snapshot the current page: document metadata, viewport and scroll state, a semantic layout tree with a text diagram, the heading outline, and an inventory of interactive elements \u2014 each carrying its ARIA role, its live state (disabled, checked, expanded, filled, aria-current) and the landmark region it sits in. When the site registers WebMCP tools, the result also carries a siteTools list \u2014 prefer page.callSiteTool over clicking wherever a listed tool covers the step. Visible iframes come back under "frames"; nothing inside one is in this snapshot until page.switchFrame enters it, and "frame" then says which one is in focus.',
   input: external_exports.object({
     maxPerKind: external_exports.number().int().positive().default(30).describe("Cap on links, buttons, fields, and forms listed per kind")
   }),
@@ -19929,10 +19977,23 @@ var getPageInfo = defineAction({
         level: Number(heading.tagName[1]),
         text: accessibleText(heading).slice(0, 120)
       })),
-      interactive: inventory(found, owners, maxPerKind)
+      interactive: inventory(found, owners, maxPerKind),
+      frames: embeddedFrames()
     };
   }
 });
+var MAX_FRAMES = 20;
+function embeddedFrames() {
+  const frames = [...document.querySelectorAll("iframe,frame")].filter(isFrameElement).filter(isExposed).slice(0, MAX_FRAMES).map((frame) => ({
+    selector: cssPath(frame),
+    src: frame.src || void 0,
+    name: frame.name || void 0,
+    title: frame.title || void 0,
+    sandbox: sandboxOf(frame),
+    bounds: documentBounds(frame)
+  }));
+  return frames.length ? frames : void 0;
+}
 function regionRole(el) {
   const explicit = el.getAttribute("role");
   if (explicit) return LANDMARK_ROLES.has(explicit) ? explicit : void 0;
@@ -21577,6 +21638,7 @@ var actions = new Map(
     navigate,
     openTab,
     switchTab,
+    switchFrame,
     closeTab,
     screenshot,
     listFiles,
@@ -24554,6 +24616,7 @@ var READ_ONLY_ACTIONS = /* @__PURE__ */ new Set([
   "page.findSearch",
   "page.findCaptcha",
   "page.listSiteTools",
+  "page.switchFrame",
   "page.monitorStatus",
   "page.timerStatus",
   "page.readTheme",
@@ -27181,7 +27244,7 @@ X-Browsentic-Reason: ${reason}\r
 // package.json
 var package_default = {
   name: "browsentic",
-  version: "0.4.15",
+  version: "0.5.0",
   description: "Hand your real, logged-in browser to the AI agent you already run. Installs the browser extension, runs the local daemon, and speaks MCP.",
   type: "module",
   license: "MIT",