preflite 1.0.1 → 1.1.0

package/dist/mcp/exploration/index.js

@@ -53,7 +53,7 @@ export function registerExplorationTools(server, ctx) {
     // ---------------------------------------------------------------------------
     server.registerTool("exploration_get_page_summary", {
         title: "Get Page Summary",
-        description: "Get a natural language summary of the current screen, including layout type (fixed single-screen / scrollable long page / multi-tab / list). Call this first when entering a new page to understand its structure before acting. Use this INSTEAD of ai_act if you just want to observe — ai_act is for changing state, not for looking around.",
+        description: "Get a natural language summary of the current screen, including layout type (fixed single-screen / scrollable long page / multi-tab / list). Call this first when entering a new page to understand its structure before acting. The summary is saved internally; if you then call ai_act, it will be used as the before-state for change comparison. Use this INSTEAD of ai_act if you just want to observe — ai_act is for changing state, not for looking around.",
         inputSchema: {
             sessionId: z.string().describe("Session ID from exploration_start"),
         },
@@ -76,7 +76,7 @@ export function registerExplorationTools(server, ctx) {
     // ---------------------------------------------------------------------------
     server.registerTool("exploration_ai_act", {
         title: "AI Act",
-        description: "Perform a high-level UI interaction described in natural language. Examples: 'Go back', 'Tap the settings icon', 'Type text in the search box'. After execution, returns a summary of the new page state including whether the action actually changed anything.\n\nIMPORTANT: Use get_page_summary FIRST to check if the page is a fixed single-screen layout. If it is, do NOT request scroll actions — there is nothing to scroll to. Use ai_act only for meaningful interactions (tap, type, swipe between tabs), not for 'look around' or 'scroll to see more'. If the post-action summary reports that the page did not change, stop acting on this page and move on.",
+        description: "Perform a high-level UI interaction described in natural language. Examples: 'Go back', 'Tap the settings icon', 'Type text in the search box'. After execution, returns a summary of the new page state including whether the action actually changed anything.\n\nIMPORTANT: Use get_page_summary FIRST to check if the page is a fixed single-screen layout. If it is, do NOT request scroll actions — there is nothing to scroll to. Use ai_act only for meaningful interactions (tap, type, swipe between tabs), not for 'look around' or 'scroll to see more'. If the post-action summary reports that the page did not change, stop acting on this page and move on.\n\nNote on before/after comparison: The AI model is stateless — each summary only sees the current screenshot. To detect changes, ai_act internally captures a before-state description (either from your prior get_page_summary call, or by grabbing a quick snapshot before acting) and passes it to the after-summary for comparison. For richer change detection, call get_page_summary before ai_act.",
         inputSchema: {
             sessionId: z.string().describe("Session ID from exploration_start"),
             intent: z.string().describe("Description of what to do"),

package/dist/mcp/exploration/sessionManager.js

@@ -24,6 +24,7 @@ async function persistMeta(state, env) {
         resourceId: state.resourceId,
         platform: state.platform,
         env,
+        appRef: state.appRef,
         createdAt: state.createdAt,
         lastActivityAt: state.lastActivityAt,
     };
@@ -50,12 +51,13 @@ async function removeMeta(id) {
 // Session lifecycle
 // ---------------------------------------------------------------------------
 /** Create a new exploration session and persist metadata to disk. */
-export function createSession(id, resourceId, platform, session, env) {
+export function createSession(id, resourceId, platform, session, env, appRef) {
     const state = {
         id,
         resourceId,
         platform,
         session,
+        appRef,
         createdAt: Date.now(),
         lastActivityAt: Date.now(),
     };
@@ -88,6 +90,7 @@ export function storeSession(id, session, meta) {
         resourceId: meta.resourceId,
         platform: meta.platform,
         session,
+        appRef: meta.appRef,
         createdAt: meta.createdAt,
         lastActivityAt: Date.now(),
     };

package/dist/mcp/exploration/tools-intelligent.js

@@ -1,4 +1,65 @@
+import http from "node:http";
+import { execSync } from "node:child_process";
 import { resolveSession } from "./tools-session.js";
+import { getSession } from "./sessionManager.js";
+async function detectIosForegroundApp(session) {
+    if (session.platform !== "ios")
+        return null;
+    const { wdaHost, wdaPort } = session.target;
+    return new Promise((resolve) => {
+        const req = http.get(`http://${wdaHost}:${wdaPort}/wda/activeAppInfo`, { timeout: 3000 }, (res) => {
+            let body = "";
+            res.on("data", (chunk) => { body += chunk; });
+            res.on("end", () => {
+                try {
+                    const data = JSON.parse(body);
+                    if (data?.value?.bundleId) {
+                        resolve({ bundleId: data.value.bundleId, name: data.value.name });
+                    }
+                    else {
+                        resolve(null);
+                    }
+                }
+                catch {
+                    resolve(null);
+                }
+            });
+        });
+        req.on("error", () => resolve(null));
+        req.on("timeout", () => { req.destroy(); resolve(null); });
+    });
+}
+function detectAndroidForegroundApp(session) {
+    if (session.platform !== "android")
+        return null;
+    const { serial, adbHost, adbPort } = session.target;
+    try {
+        const output = execSync(`adb -H ${adbHost} -P ${adbPort} -s ${serial} shell dumpsys window 2>/dev/null | grep -E 'mCurrentFocus|mFocusedApp'`, { encoding: "utf8", timeout: 5000 });
+        // Match patterns like:
+        //   mCurrentFocus=Window{... com.example.app/...}
+        //   mFocusedApp=AppWindowToken{... token=... appPackageName=com.example.app}
+        const focusMatch = output.match(/mCurrentFocus[=:].*?\s+([^\s/}]+)/);
+        const packageMatch = output.match(/appPackageName=([^\s}]+)/);
+        if (focusMatch)
+            return { bundleId: focusMatch[1] };
+        if (packageMatch)
+            return { bundleId: packageMatch[1] };
+    }
+    catch {
+        // adb not reachable or no foreground window
+    }
+    return null;
+}
+async function detectForegroundApp(session) {
+    if (session.platform === "ios")
+        return detectIosForegroundApp(session);
+    if (session.platform === "android")
+        return detectAndroidForegroundApp(session);
+    return null;
+}
+// ---------------------------------------------------------------------------
+// Tool handlers
+// ---------------------------------------------------------------------------
 export function getPageSummaryHandler(ctx) {
     return async (input) => {
         const session = await resolveSession(input.sessionId, ctx);
@@ -8,7 +69,20 @@ export function getPageSummaryHandler(ctx) {
             "2) 是否有弹窗、广告或遮挡物？\n" +
             "3) 整体布局类型：固定单屏 / 可滚动长页面 / 多Tab / 列表\n" +
             "先判断布局类型，再逐一描述每个区域的内容。");
-        return { summary };
+        const state = getSession(input.sessionId);
+        const foregroundApp = await detectForegroundApp(session);
+        const appRef = state.appRef;
+        // Save for aiActHandler's before/after comparison
+        state.lastPageSummary = summary;
+        return {
+            summary,
+            app: {
+                platform: session.platform,
+                resourceId: state.resourceId,
+                ...(appRef ? { appRef } : {}),
+                ...(foregroundApp ? { foregroundApp } : {}),
+            },
+        };
     };
 }
 export function askAboutScreenHandler(ctx) {
@@ -21,13 +95,30 @@ export function askAboutScreenHandler(ctx) {
 export function aiActHandler(ctx) {
     return async (input) => {
         const session = await resolveSession(input.sessionId, ctx);
+        const state = getSession(input.sessionId);
+        // Before-state: reuse from get_page_summary if available, otherwise grab a quick one
+        const beforeSummary = state.lastPageSummary ?? await session.agent.aiAsk("用一句话描述当前页面的最关键特征：什么类型的页面（列表/表单/弹窗/首页等），最显著的内容是什么。");
         await session.agent.aiAct(input.intent);
-        const afterSummary = await session.agent.aiAsk("刚刚的操作已完成。请判断：\n" +
-            "1) 操作是否改变了页面内容？（新页面、弹窗、滚动到底部、输入框获得焦点等）\n" +
-            "2) 如果操作是滑动页面，是否滑到了底部或页面内容没有变化？\n" +
+        const afterSummary = await session.agent.aiAsk(`操作前的页面：${beforeSummary}\n` +
+            `执行的操作：${input.intent}\n` +
+            "请判断操作结果：\n" +
+            "1) 页面内容是否发生了变化？（进入了新页面、弹出弹窗、滚动到底部、输入框获得焦点等）\n" +
+            "2) 如果操作是滑动，是否已到达底部或页面没有变化？\n" +
             "3) 当前页面布局类型是固定单屏还是可滚动长页面？\n" +
-            "4) 当前页面出现的最关键变化是什么？\n" +
-            "如果你发现操作没有产生任何实际变化（比如反复滑动但没有新内容），请明确指出\"页面没有变化\"。");
-        return { ok: true, summary: afterSummary };
+            "4) 当前页面最关键的变化是什么？\n" +
+            "如果操作没有产生任何实际变化（比如反复滑动但没有新内容），请明确指出\"页面没有变化\"。");
+        // Update for next aiAct call
+        state.lastPageSummary = afterSummary;
+        const foregroundApp = await detectForegroundApp(session);
+        return {
+            ok: true,
+            summary: afterSummary,
+            app: {
+                platform: session.platform,
+                resourceId: state.resourceId,
+                ...(state.appRef ? { appRef: state.appRef } : {}),
+                ...(foregroundApp ? { foregroundApp } : {}),
+            },
+        };
     };
 }

package/dist/mcp/exploration/tools-session.js

@@ -241,25 +241,26 @@ export function getExplorationStartHandler(ctx) {
         }
         const session = await createMidsceneSessionFromResourceId(resourceId, runtimeEnv);
         const sessionId = generateSessionId();
-        createSession(sessionId, resourceId, platform, session, runtimeEnv);
+        const appRef = input.appRef;
+        createSession(sessionId, resourceId, platform, session, runtimeEnv, appRef);
         try {
-            if (input.appRef) {
-                if (isInstallableRef(input.appRef)) {
-                    await ctx.client.installApp(resourceId, input.appRef);
+            if (appRef) {
+                if (isInstallableRef(appRef)) {
+                    await ctx.client.installApp(resourceId, appRef);
                     return {
                         sessionId,
                         device: { platform, resourceId },
-                        note: `App installed from ${input.appRef}. Use exploration_ai_act to launch it.`,
+                        note: `App installed from ${appRef}. Use exploration_ai_act to launch it.`,
                     };
                 }
-                await session.agent.launch(input.appRef);
+                await session.agent.launch(appRef);
             }
         }
         catch (err) {
             await destroySessionById(sessionId).catch(() => { });
             throw err;
         }
-        return { sessionId, device: { platform, resourceId } };
+        return { sessionId, device: { platform, resourceId }, appRef: appRef ?? undefined };
     };
 }
 export function getExplorationEndHandler() {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "preflite",
-  "version": "1.0.1",
+  "version": "1.1.0",
   "description": "Preflight — Local mobile AI testing via MCP. AI-assisted testing on real Android/iOS/Harmony devices.",
   "license": "Apache-2.0",
   "type": "module",