screenpipe-mcp 0.12.0 → 0.14.0

package/dist/index.js

@@ -53,10 +53,13 @@ for (let i = 0; i < args.length; i++) {
     }
 }
 const SCREENPIPE_API = `http://localhost:${port}`;
+// Read version from package.json (single source of truth)
+// eslint-disable-next-line @typescript-eslint/no-var-requires
+const PKG_VERSION = require("../package.json").version;
 // Initialize server
 const server = new index_js_1.Server({
     name: "screenpipe",
-    version: "0.9.0",
+    version: PKG_VERSION,
 }, {
     capabilities: {
         tools: {},
@@ -133,9 +136,10 @@ const TOOLS = [
     },
     {
         name: "activity-summary",
-        description: "Lightweight activity overview (~200-500 tokens): app usage with active minutes, audio speakers, recent texts. " +
+        description: "Rich activity overview: app usage, window/tab titles with URLs and time spent, key text per context, audio transcriptions. " +
             "USE THIS FIRST for broad questions: 'what was I doing?', 'how long on X?', 'which apps?'. " +
-            "Only escalate to search-content if you need specific text content.",
+            "The 'windows' field shows exactly what the user worked on (e.g. 'Debug crash issue — 20 min', 'Stripe pricing page — 5 min'). " +
+            "Usually sufficient without further searches.",
         annotations: { title: "Activity Summary", readOnlyHint: true, openWorldHint: false, idempotentHint: true },
         inputSchema: {
             type: "object",
@@ -658,8 +662,19 @@ server.setRequestHandler(types_js_1.CallToolRequestSchema, async (request) => {
                         : "";
                     return `  ${a.name}: ${a.minutes} min (${a.frame_count} frames${timeSpan})`;
                 });
+                // Window/tab activity — what pages/documents were open
+                const windowLines = (data.windows || []).map((w) => {
+                    const url = w.browser_url ? ` (${w.browser_url})` : "";
+                    return `  [${w.app_name}] ${w.window_name}${url} — ${w.minutes} min`;
+                });
                 const speakerLines = (data.audio_summary?.speakers || []).map((s) => `  ${s.name}: ${s.segment_count} segments`);
-                const textLines = (data.recent_texts || []).map((t) => `  [${t.app_name}] ${t.text}`);
+                // Actual audio transcriptions (not just counts)
+                const transcriptLines = (data.audio_summary?.top_transcriptions || []).map((t) => `  [${t.speaker}, ${t.timestamp.slice(11, 19)}] ${t.transcription}`);
+                // Key text content sampled across the time range
+                const textLines = (data.key_texts || data.recent_texts || []).map((t) => {
+                    const win = t.window_name ? ` | ${t.window_name}` : "";
+                    return `  [${t.app_name}${win}, ${t.timestamp.slice(11, 19)}] ${t.text}`;
+                });
                 const summary = [
                     `Activity Summary (${data.time_range?.start} → ${data.time_range?.end})`,
                     `Total frames: ${data.total_frames}`,
@@ -667,11 +682,15 @@ server.setRequestHandler(types_js_1.CallToolRequestSchema, async (request) => {
                     "Apps:",
                     ...(appsLines.length ? appsLines : ["  (none)"]),
                     "",
+                    "Windows & Tabs:",
+                    ...(windowLines.length ? windowLines.slice(0, 20) : ["  (none)"]),
+                    "",
                     `Audio: ${data.audio_summary?.segment_count || 0} segments`,
                     ...(speakerLines.length ? speakerLines : []),
+                    ...(transcriptLines.length ? ["", "Audio transcriptions:", ...transcriptLines.slice(0, 15)] : []),
                     "",
-                    "Recent texts:",
-                    ...(textLines.length ? textLines.slice(0, 10) : ["  (none)"]),
+                    "Key content (sampled across time range):",
+                    ...(textLines.length ? textLines.slice(0, 20) : ["  (none)"]),
                 ].join("\n");
                 return { content: [{ type: "text", text: summary }] };
             }

package/manifest.json

@@ -2,7 +2,7 @@
   "manifest_version": "0.3",
   "name": "screenpipe",
   "display_name": "Screenpipe",
-  "version": "0.8.4",
+  "version": "0.0.0-injected-from-package-json",
   "description": "Search your screen recordings and audio transcriptions with AI",
   "long_description": "Screenpipe is a 24/7 screen and audio recorder that lets you search everything you've seen or heard. This extension connects Claude to your local screenpipe instance, enabling AI-powered search through your digital memory.",
   "author": {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "screenpipe-mcp",
-  "version": "0.12.0",
+  "version": "0.14.0",
   "description": "MCP server for screenpipe - search your screen recordings and audio transcriptions",
   "main": "dist/index.js",
   "bin": {

package/src/index.ts

@@ -28,11 +28,15 @@ for (let i = 0; i < args.length; i++) {
 
 const SCREENPIPE_API = `http://localhost:${port}`;
 
+// Read version from package.json (single source of truth)
+// eslint-disable-next-line @typescript-eslint/no-var-requires
+const PKG_VERSION: string = require("../package.json").version;
+
 // Initialize server
 const server = new Server(
   {
     name: "screenpipe",
-    version: "0.9.0",
+    version: PKG_VERSION,
   },
   {
     capabilities: {
@@ -115,9 +119,10 @@ const TOOLS: Tool[] = [
   {
     name: "activity-summary",
     description:
-      "Lightweight activity overview (~200-500 tokens): app usage with active minutes, audio speakers, recent texts. " +
+      "Rich activity overview: app usage, window/tab titles with URLs and time spent, key text per context, audio transcriptions. " +
       "USE THIS FIRST for broad questions: 'what was I doing?', 'how long on X?', 'which apps?'. " +
-      "Only escalate to search-content if you need specific text content.",
+      "The 'windows' field shows exactly what the user worked on (e.g. 'Debug crash issue — 20 min', 'Stripe pricing page — 5 min'). " +
+      "Usually sufficient without further searches.",
     annotations: { title: "Activity Summary", readOnlyHint: true, openWorldHint: false, idempotentHint: true },
     inputSchema: {
       type: "object",
@@ -705,14 +710,37 @@ server.setRequestHandler(CallToolRequestSchema, async (request) => {
           }
         );
 
+        // Window/tab activity — what pages/documents were open
+        const windowLines = (data.windows || []).map(
+          (w: {
+            app_name: string;
+            window_name: string;
+            browser_url: string;
+            minutes: number;
+            frame_count: number;
+          }) => {
+            const url = w.browser_url ? ` (${w.browser_url})` : "";
+            return `  [${w.app_name}] ${w.window_name}${url} — ${w.minutes} min`;
+          }
+        );
+
         const speakerLines = (data.audio_summary?.speakers || []).map(
           (s: { name: string; segment_count: number }) =>
             `  ${s.name}: ${s.segment_count} segments`
         );
 
-        const textLines = (data.recent_texts || []).map(
-          (t: { text: string; app_name: string; timestamp: string }) =>
-            `  [${t.app_name}] ${t.text}`
+        // Actual audio transcriptions (not just counts)
+        const transcriptLines = (data.audio_summary?.top_transcriptions || []).map(
+          (t: { transcription: string; speaker: string; device: string; timestamp: string }) =>
+            `  [${t.speaker}, ${t.timestamp.slice(11, 19)}] ${t.transcription}`
+        );
+
+        // Key text content sampled across the time range
+        const textLines = (data.key_texts || data.recent_texts || []).map(
+          (t: { text: string; app_name: string; window_name?: string; timestamp: string }) => {
+            const win = t.window_name ? ` | ${t.window_name}` : "";
+            return `  [${t.app_name}${win}, ${t.timestamp.slice(11, 19)}] ${t.text}`;
+          }
         );
 
         const summary = [
@@ -722,11 +750,15 @@ server.setRequestHandler(CallToolRequestSchema, async (request) => {
           "Apps:",
           ...(appsLines.length ? appsLines : ["  (none)"]),
           "",
+          "Windows & Tabs:",
+          ...(windowLines.length ? windowLines.slice(0, 20) : ["  (none)"]),
+          "",
           `Audio: ${data.audio_summary?.segment_count || 0} segments`,
           ...(speakerLines.length ? speakerLines : []),
+          ...(transcriptLines.length ? ["", "Audio transcriptions:", ...transcriptLines.slice(0, 15)] : []),
           "",
-          "Recent texts:",
-          ...(textLines.length ? textLines.slice(0, 10) : ["  (none)"]),
+          "Key content (sampled across time range):",
+          ...(textLines.length ? textLines.slice(0, 20) : ["  (none)"]),
         ].join("\n");
 
         return { content: [{ type: "text", text: summary }] };