open-agents-ai 0.18.0 → 0.19.0

package/README.md

@@ -32,6 +32,7 @@ An autonomous multi-turn tool-calling agent that reads your code, makes changes,
 - **35 autonomous tools** — file I/O, shell, grep, web search/fetch, memory, sub-agents, background tasks, image/OCR, git, diagnostics, vision, desktop automation, structured files, code sandbox
 - **Moondream vision** — see and interact with the desktop via Moondream VLM (caption, query, detect, point-and-click)
 - **Desktop automation** — vision-guided clicking: describe a UI element in natural language, the agent finds and clicks it
+- **Auto-install desktop deps** — screenshot, mouse, OCR, and image tools auto-install missing system packages (scrot, xdotool, tesseract, imagemagick) on first use
 - **Parallel tool execution** — read-only tools run concurrently via `Promise.allSettled`
 - **Sub-agent delegation** — spawn independent agents for parallel workstreams
 - **Ralph Loop** — iterative task execution that keeps retrying until completion criteria are met
@@ -226,10 +227,27 @@ python3 -m venv .moondream-venv
 
 The vision tools auto-detect a running Moondream Station on `localhost:2020`. For cloud inference, set `MOONDREAM_API_KEY` instead.
 
-**System requirements for desktop interaction:**
-- Screenshot: `scrot`, `gnome-screenshot`, or Python PIL (auto-detected)
-- Mouse control: `xdotool` or `python-xlib` (auto-detected)
-- Vision: Moondream Station (local) or Moondream Cloud API
+**System dependencies (auto-installed on first use):**
+
+Desktop tools automatically install missing system packages when first needed. No manual setup required — just use the tool and it handles the rest:
+
+| Tool | Linux Package | What It Does |
+|------|--------------|-------------|
+| `scrot` | `apt install scrot` | Screenshot capture |
+| `xdotool` | `apt install xdotool` | Mouse/keyboard automation |
+| `tesseract` | `apt install tesseract-ocr` | OCR text extraction |
+| `identify` | `apt install imagemagick` | Image dimensions/conversion |
+
+Supports `apt` (Debian/Ubuntu), `dnf` (Fedora), `pacman` (Arch), and `brew` (macOS). You can also pre-install everything at once:
+
+```bash
+./scripts/setup-desktop.sh          # Install all desktop deps
+./scripts/setup-desktop.sh --check-only  # Just check what's missing
+```
+
+**Vision backend:**
+- Moondream Station (local) — runs entirely on your machine, no API keys needed
+- Moondream Cloud API — set `MOONDREAM_API_KEY` for cloud inference
 
 ## Interactive TUI
 

package/dist/index.js

@@ -1304,7 +1304,17 @@ ${stdinInput ?? ""}`);
           let stdout = "";
           let stderr = "";
           let killed = false;
+          let resolved = false;
           const maxBuf = 1024 * 1024;
+          const doResolve = (result) => {
+            if (resolved)
+              return;
+            resolved = true;
+            clearTimeout(timer);
+            if (exitFlushTimer)
+              clearTimeout(exitFlushTimer);
+            resolve19(result);
+          };
           const timer = setTimeout(() => {
             killed = true;
             child.kill("SIGTERM");
@@ -1315,6 +1325,7 @@ ${stdinInput ?? ""}`);
               }
             }, 5e3);
           }, timeout);
+          let exitFlushTimer = null;
           child.stdout.on("data", (data) => {
             stdout += data.toString();
             if (stdout.length > maxBuf) {
@@ -1331,14 +1342,39 @@ ${stdinInput ?? ""}`);
             child.stdin.write(stdinInput);
           }
           child.stdin.end();
+          child.on("exit", (code) => {
+            exitFlushTimer = setTimeout(() => {
+              const durationMs = performance.now() - start;
+              if (killed) {
+                const combined = stdout + stderr;
+                const looksInteractive = /\? .+[›>]|y\/n|yes\/no|\(Y\/n\)|\[y\/N\]/i.test(combined);
+                const hint = looksInteractive ? " The command appears to be waiting for interactive input. Use non-interactive flags (e.g., --yes, --no-input) or provide input via the stdin parameter." : "";
+                doResolve({
+                  success: false,
+                  output: stdout,
+                  error: `Command timed out after ${timeout}ms.${hint}`,
+                  durationMs
+                });
+                return;
+              }
+              const success = code === 0;
+              doResolve({
+                success,
+                output: stdout + (stderr && success ? `
+STDERR:
+${stderr}` : ""),
+                error: success ? void 0 : stderr || `Exit code ${code}`,
+                durationMs
+              });
+            }, 1e3);
+          });
           child.on("close", (code) => {
-            clearTimeout(timer);
             const durationMs = performance.now() - start;
             if (killed) {
               const combined = stdout + stderr;
               const looksInteractive = /\? .+[›>]|y\/n|yes\/no|\(Y\/n\)|\[y\/N\]/i.test(combined);
               const hint = looksInteractive ? " The command appears to be waiting for interactive input. Use non-interactive flags (e.g., --yes, --no-input) or provide input via the stdin parameter." : "";
-              resolve19({
+              doResolve({
                 success: false,
                 output: stdout,
                 error: `Command timed out after ${timeout}ms.${hint}`,
@@ -1347,7 +1383,7 @@ ${stdinInput ?? ""}`);
               return;
             }
             const success = code === 0;
-            resolve19({
+            doResolve({
               success,
               output: stdout + (stderr && success ? `
 STDERR:
@@ -1357,8 +1393,7 @@ ${stderr}` : ""),
             });
           });
           child.on("error", (err) => {
-            clearTimeout(timer);
-            resolve19({
+            doResolve({
               success: false,
               output: stdout,
               error: err.message,
@@ -3893,10 +3928,173 @@ Exit code: ${task.exitCode ?? "N/A"}`,
   }
 });
 
+// packages/execution/dist/system-deps.js
+import { execSync as execSync6 } from "node:child_process";
+function detectPackageManager() {
+  if (_detectedPkgManager !== void 0)
+    return _detectedPkgManager;
+  const plat = process.platform;
+  if (plat === "darwin") {
+    _detectedPkgManager = hasCommand("brew") ? "brew" : null;
+    return _detectedPkgManager;
+  }
+  if (plat === "linux") {
+    if (hasCommand("apt-get"))
+      _detectedPkgManager = "apt";
+    else if (hasCommand("dnf"))
+      _detectedPkgManager = "dnf";
+    else if (hasCommand("pacman"))
+      _detectedPkgManager = "pacman";
+    else
+      _detectedPkgManager = null;
+    return _detectedPkgManager;
+  }
+  _detectedPkgManager = null;
+  return null;
+}
+function hasCommand(cmd) {
+  try {
+    execSync6(`which ${cmd}`, { stdio: "pipe", timeout: 3e3 });
+    return true;
+  } catch {
+    return false;
+  }
+}
+function runInstall(installCmd) {
+  const needsSudo = process.platform === "linux";
+  let cmd = installCmd;
+  if (needsSudo) {
+    if (_sudoPassword) {
+      cmd = `echo ${JSON.stringify(_sudoPassword)} | sudo -S ${installCmd}`;
+    } else {
+      cmd = `sudo -n ${installCmd}`;
+    }
+  }
+  execSync6(cmd, {
+    stdio: "pipe",
+    timeout: 12e4,
+    // 2 min for package install
+    env: { ...process.env, DEBIAN_FRONTEND: "noninteractive" }
+  });
+}
+function ensureCommand(command) {
+  const cached = _cache.get(command);
+  if (cached)
+    return cached;
+  if (hasCommand(command)) {
+    const result2 = { available: true, installed: false };
+    _cache.set(command, result2);
+    return result2;
+  }
+  const dep = DESKTOP_DEPS.find((d) => d.command === command);
+  if (!dep) {
+    const result2 = {
+      available: false,
+      installed: false,
+      error: `Unknown dependency: ${command}. Cannot auto-install.`
+    };
+    _cache.set(command, result2);
+    return result2;
+  }
+  const pm = detectPackageManager();
+  if (!pm) {
+    const result2 = {
+      available: false,
+      installed: false,
+      error: `No supported package manager found. Install ${command} manually.`
+    };
+    _cache.set(command, result2);
+    return result2;
+  }
+  const pkg = dep.packages[pm];
+  if (!pkg) {
+    const result2 = {
+      available: false,
+      installed: false,
+      error: `No package for ${command} on ${pm}. Install manually.`
+    };
+    _cache.set(command, result2);
+    return result2;
+  }
+  let installCmd;
+  switch (pm) {
+    case "apt":
+      installCmd = `apt-get install -y ${pkg}`;
+      break;
+    case "dnf":
+      installCmd = `dnf install -y ${pkg}`;
+      break;
+    case "pacman":
+      installCmd = `pacman -S --noconfirm ${pkg}`;
+      break;
+    case "brew":
+      installCmd = `brew install ${pkg}`;
+      break;
+  }
+  try {
+    runInstall(installCmd);
+  } catch (err) {
+    const msg = err instanceof Error ? err.message : String(err);
+    const result2 = {
+      available: false,
+      installed: false,
+      error: `Failed to install ${pkg} via ${pm}: ${msg.slice(0, 200)}`
+    };
+    _cache.set(command, result2);
+    return result2;
+  }
+  if (hasCommand(command)) {
+    const result2 = { available: true, installed: true };
+    _cache.set(command, result2);
+    return result2;
+  }
+  const result = {
+    available: false,
+    installed: false,
+    error: `Installed ${pkg} but ${command} still not found on PATH.`
+  };
+  _cache.set(command, result);
+  return result;
+}
+var DESKTOP_DEPS, _detectedPkgManager, _cache, _sudoPassword;
+var init_system_deps = __esm({
+  "packages/execution/dist/system-deps.js"() {
+    "use strict";
+    DESKTOP_DEPS = [
+      {
+        command: "scrot",
+        packages: { apt: "scrot", dnf: "scrot", pacman: "scrot", brew: "scrot" },
+        description: "Screenshot capture tool",
+        group: "screenshot"
+      },
+      {
+        command: "xdotool",
+        packages: { apt: "xdotool", dnf: "xdotool", pacman: "xdotool" },
+        description: "X11 desktop automation (mouse/keyboard)",
+        group: "click"
+      },
+      {
+        command: "tesseract",
+        packages: { apt: "tesseract-ocr", dnf: "tesseract", pacman: "tesseract", brew: "tesseract" },
+        description: "OCR text extraction engine",
+        group: "ocr"
+      },
+      {
+        command: "identify",
+        packages: { apt: "imagemagick", dnf: "ImageMagick", pacman: "imagemagick", brew: "imagemagick" },
+        description: "Image metadata and conversion (ImageMagick)",
+        group: "image"
+      }
+    ];
+    _cache = /* @__PURE__ */ new Map();
+    _sudoPassword = null;
+  }
+});
+
 // packages/execution/dist/tools/image.js
 import { existsSync as existsSync7, readFileSync as readFileSync6, statSync as statSync4 } from "node:fs";
 import { resolve as resolve11, extname as extname2, basename } from "node:path";
-import { execSync as execSync6 } from "node:child_process";
+import { execSync as execSync7 } from "node:child_process";
 import { tmpdir } from "node:os";
 import { join as join10 } from "node:path";
 function isImagePath(path) {
@@ -3923,7 +4121,7 @@ function getMimeType(filePath) {
 }
 function getImageDimensions(filePath) {
   try {
-    const out = execSync6(`identify -format "%w %h" ${JSON.stringify(filePath)}`, {
+    const out = execSync7(`identify -format "%w %h" ${JSON.stringify(filePath)}`, {
       encoding: "utf8",
       stdio: ["pipe", "pipe", "pipe"],
       timeout: 5e3
@@ -3932,9 +4130,23 @@ function getImageDimensions(filePath) {
     if (w && h)
       return { width: w, height: h };
   } catch {
+    const dep = ensureCommand("identify");
+    if (dep.installed) {
+      try {
+        const out = execSync7(`identify -format "%w %h" ${JSON.stringify(filePath)}`, {
+          encoding: "utf8",
+          stdio: ["pipe", "pipe", "pipe"],
+          timeout: 5e3
+        }).trim();
+        const [w, h] = out.split(" ").map(Number);
+        if (w && h)
+          return { width: w, height: h };
+      } catch {
+      }
+    }
   }
   try {
-    const out = execSync6(`file ${JSON.stringify(filePath)}`, {
+    const out = execSync7(`file ${JSON.stringify(filePath)}`, {
       encoding: "utf8",
       stdio: ["pipe", "pipe", "pipe"],
       timeout: 5e3
@@ -3948,15 +4160,16 @@ function getImageDimensions(filePath) {
 }
 function hasTesseract() {
   try {
-    execSync6("tesseract --version", { stdio: "pipe", timeout: 5e3 });
+    execSync7("tesseract --version", { stdio: "pipe", timeout: 5e3 });
     return true;
   } catch {
-    return false;
+    const dep = ensureCommand("tesseract");
+    return dep.available;
   }
 }
 function runOCR(filePath) {
   try {
-    return execSync6(`tesseract ${JSON.stringify(filePath)} stdout 2>/dev/null`, {
+    return execSync7(`tesseract ${JSON.stringify(filePath)} stdout 2>/dev/null`, {
       encoding: "utf8",
       stdio: ["pipe", "pipe", "pipe"],
       timeout: 3e4
@@ -3969,6 +4182,7 @@ var IMAGE_EXTENSIONS, ImageReadTool, ScreenshotTool, OCRTool;
 var init_image = __esm({
   "packages/execution/dist/tools/image.js"() {
     "use strict";
+    init_system_deps();
     IMAGE_EXTENSIONS = /* @__PURE__ */ new Set([
       ".png",
       ".jpg",
@@ -4089,7 +4303,7 @@ ${ocrText}`);
           };
         }
         try {
-          execSync6(cmd, { stdio: "pipe", timeout: 1e4 });
+          execSync7(cmd, { stdio: "pipe", timeout: 1e4 });
         } catch (err) {
           return {
             success: false,
@@ -4137,7 +4351,7 @@ ${ocrText}`);
         if (plat === "linux") {
           for (const tool of ["scrot", "gnome-screenshot", "import"]) {
             try {
-              execSync6(`which ${tool}`, { stdio: "pipe" });
+              execSync7(`which ${tool}`, { stdio: "pipe" });
               if (tool === "scrot") {
                 if (region === "active")
                   return `scrot -u ${out}`;
@@ -4160,6 +4374,16 @@ ${ocrText}`);
             } catch {
             }
           }
+          const dep = ensureCommand("scrot");
+          if (dep.available) {
+            if (region === "active")
+              return `scrot -u ${out}`;
+            if (region !== "full" && region.includes(",")) {
+              const [x, y, w, h] = region.split(",").map(Number);
+              return `scrot -a ${x},${y},${w},${h} ${out}`;
+            }
+            return `scrot ${out}`;
+          }
         }
         return null;
       }
@@ -4204,7 +4428,7 @@ ${ocrText}`);
           return {
             success: false,
             output: "",
-            error: "Tesseract not installed. Install with: sudo apt install tesseract-ocr",
+            error: "Tesseract not available. Auto-install failed. Try manually: sudo apt install tesseract-ocr",
             durationMs: Date.now() - start
           };
         }
@@ -4214,14 +4438,14 @@ ${ocrText}`);
           if (x != null && y != null && w != null && h != null) {
             const croppedPath = join10(tmpdir(), `oa-ocr-crop-${Date.now()}.png`);
             try {
-              execSync6(`convert ${JSON.stringify(fullPath)} -crop ${w}x${h}+${x}+${y} +repage ${JSON.stringify(croppedPath)}`, { stdio: "pipe", timeout: 1e4 });
+              execSync7(`convert ${JSON.stringify(fullPath)} -crop ${w}x${h}+${x}+${y} +repage ${JSON.stringify(croppedPath)}`, { stdio: "pipe", timeout: 1e4 });
               inputPath = croppedPath;
             } catch {
             }
           }
         }
         try {
-          const text = execSync6(`tesseract ${JSON.stringify(inputPath)} stdout -l ${language} --psm ${psm} 2>/dev/null`, { encoding: "utf8", stdio: ["pipe", "pipe", "pipe"], timeout: 3e4 }).trim();
+          const text = execSync7(`tesseract ${JSON.stringify(inputPath)} stdout -l ${language} --psm ${psm} 2>/dev/null`, { encoding: "utf8", stdio: ["pipe", "pipe", "pipe"], timeout: 3e4 }).trim();
           if (!text) {
             return { success: true, output: "(no text detected in image)", durationMs: Date.now() - start };
           }
@@ -4756,7 +4980,7 @@ var init_tool_creator = __esm({
 import { existsSync as existsSync9, readdirSync as readdirSync5, readFileSync as readFileSync8 } from "node:fs";
 import { join as join12, basename as basename2, dirname as dirname2 } from "node:path";
 import { homedir as homedir4 } from "node:os";
-import { execSync as execSync7 } from "node:child_process";
+import { execSync as execSync8 } from "node:child_process";
 function getAiwgPaths() {
   const dataDir = join12(homedir4(), ".local", "share", "ai-writing-guide");
   return {
@@ -4769,7 +4993,7 @@ function findAiwgPackageRoot() {
   if (_cachedAiwgPkgRoot !== void 0)
     return _cachedAiwgPkgRoot;
   try {
-    const globalRoot = execSync7("npm root -g", {
+    const globalRoot = execSync8("npm root -g", {
       encoding: "utf-8",
       timeout: 5e3,
       stdio: ["pipe", "pipe", "pipe"]
@@ -5156,7 +5380,7 @@ ${content}`,
 import { existsSync as existsSync10, mkdirSync as mkdirSync4, writeFileSync as writeFileSync4, readFileSync as readFileSync9, unlinkSync } from "node:fs";
 import { join as join13, basename as basename3, extname as extname3, resolve as resolve12 } from "node:path";
 import { homedir as homedir5 } from "node:os";
-import { execSync as execSync8, spawn as spawn4 } from "node:child_process";
+import { execSync as execSync9, spawn as spawn4 } from "node:child_process";
 function isTranscribable(path) {
   const ext = extname3(path).toLowerCase();
   return AUDIO_EXTS.has(ext) || VIDEO_EXTS.has(ext);
@@ -5166,7 +5390,7 @@ async function loadTranscribeCli() {
     return _tcModule;
   _tcChecked = true;
   try {
-    const globalRoot = execSync8("npm root -g", {
+    const globalRoot = execSync9("npm root -g", {
       encoding: "utf-8",
       timeout: 5e3,
       stdio: ["pipe", "pipe", "pipe"]
@@ -5331,7 +5555,7 @@ var init_transcribe_tool = __esm({
           const args = [filePath, "-m", model, "-f", "txt"];
           if (diarize)
             args.push("--diarize");
-          const output = execSync8(`transcribe-cli ${args.join(" ")}`, {
+          const output = execSync9(`transcribe-cli ${args.join(" ")}`, {
             encoding: "utf-8",
             timeout: 3e5,
             // 5 min max
@@ -5398,12 +5622,12 @@ var init_transcribe_tool = __esm({
         const tmpFile = join13(tmpDir, `download-${Date.now()}${ext}`);
         try {
           try {
-            execSync8(`curl -sL -o "${tmpFile}" "${url}"`, {
+            execSync9(`curl -sL -o "${tmpFile}" "${url}"`, {
               timeout: 12e4,
               stdio: ["pipe", "pipe", "pipe"]
             });
           } catch {
-            execSync8(`wget -q -O "${tmpFile}" "${url}"`, {
+            execSync9(`wget -q -O "${tmpFile}" "${url}"`, {
               timeout: 12e4,
               stdio: ["pipe", "pipe", "pipe"]
             });
@@ -6216,7 +6440,7 @@ ${parts.join("\n\n")}`,
 
 // packages/execution/dist/tools/vision.js
 import { readFileSync as readFileSync10, existsSync as existsSync11, statSync as statSync5 } from "node:fs";
-import { execSync as execSync9, spawn as spawn6 } from "node:child_process";
+import { execSync as execSync10, spawn as spawn6 } from "node:child_process";
 import { resolve as resolve15, extname as extname6, basename as basename4, dirname as dirname4, join as join15 } from "node:path";
 import { fileURLToPath } from "node:url";
 async function probeStation(endpoint) {
@@ -6240,14 +6464,14 @@ function findStationBinary() {
   for (const p of localVenvPaths) {
     if (existsSync11(p)) {
       try {
-        execSync9(`${JSON.stringify(p)} -c "import moondream_station"`, { stdio: "pipe", timeout: 5e3 });
+        execSync10(`${JSON.stringify(p)} -c "import moondream_station"`, { stdio: "pipe", timeout: 5e3 });
         return p;
       } catch {
       }
     }
   }
   try {
-    const path = execSync9("which moondream-station", { encoding: "utf8", stdio: ["pipe", "pipe", "pipe"], timeout: 3e3 }).trim();
+    const path = execSync10("which moondream-station", { encoding: "utf8", stdio: ["pipe", "pipe", "pipe"], timeout: 3e3 }).trim();
     if (path)
       return path;
   } catch {
@@ -6518,13 +6742,13 @@ Coordinates are normalized (0-1). Multiply by image width/height for pixel value
 
 // packages/execution/dist/tools/desktop-click.js
 import { readFileSync as readFileSync11, existsSync as existsSync12 } from "node:fs";
-import { execSync as execSync10 } from "node:child_process";
+import { execSync as execSync11 } from "node:child_process";
 import { tmpdir as tmpdir3 } from "node:os";
 import { join as join16, dirname as dirname5 } from "node:path";
 import { fileURLToPath as fileURLToPath2 } from "node:url";
-function hasCommand(cmd) {
+function hasCommand2(cmd) {
   try {
-    execSync10(`which ${cmd}`, { stdio: "pipe", timeout: 3e3 });
+    execSync11(`which ${cmd}`, { stdio: "pipe", timeout: 3e3 });
     return true;
   } catch {
     return false;
@@ -6532,7 +6756,7 @@ function hasCommand(cmd) {
 }
 function getImageDimensions2(filePath) {
   try {
-    const out = execSync10(`identify -format "%w %h" ${JSON.stringify(filePath)}`, {
+    const out = execSync11(`identify -format "%w %h" ${JSON.stringify(filePath)}`, {
       encoding: "utf8",
       stdio: ["pipe", "pipe", "pipe"],
       timeout: 5e3
@@ -6543,14 +6767,14 @@ function getImageDimensions2(filePath) {
   } catch {
   }
   try {
-    const out = execSync10(`python3 -c "from PIL import Image; i=Image.open(${JSON.stringify(filePath)}); print(i.size[0], i.size[1])"`, { encoding: "utf8", stdio: ["pipe", "pipe", "pipe"], timeout: 5e3 }).trim();
+    const out = execSync11(`python3 -c "from PIL import Image; i=Image.open(${JSON.stringify(filePath)}); print(i.size[0], i.size[1])"`, { encoding: "utf8", stdio: ["pipe", "pipe", "pipe"], timeout: 5e3 }).trim();
     const [w, h] = out.split(" ").map(Number);
     if (w && h)
       return { width: w, height: h };
   } catch {
   }
   try {
-    const out = execSync10(`file ${JSON.stringify(filePath)}`, {
+    const out = execSync11(`file ${JSON.stringify(filePath)}`, {
       encoding: "utf8",
       stdio: ["pipe", "pipe", "pipe"],
       timeout: 5e3
@@ -6569,25 +6793,30 @@ function captureScreenshot(outputPath) {
   if (plat === "darwin") {
     cmd = `screencapture ${out}`;
   } else if (plat === "linux") {
-    if (hasCommand("scrot")) {
+    if (hasCommand2("scrot")) {
       cmd = `scrot ${out}`;
-    } else if (hasCommand("gnome-screenshot")) {
+    } else if (hasCommand2("gnome-screenshot")) {
       cmd = `gnome-screenshot -f ${out}`;
-    } else if (hasCommand("import")) {
+    } else if (hasCommand2("import")) {
       cmd = `import -window root ${out}`;
     } else {
-      try {
-        execSync10(`DISPLAY=:0 python3 -c "from PIL import ImageGrab; ImageGrab.grab().save(${JSON.stringify(outputPath)})"`, { stdio: "pipe", timeout: 1e4 });
-        if (existsSync12(outputPath))
-          return;
-      } catch {
+      const dep = ensureCommand("scrot");
+      if (dep.available) {
+        cmd = `scrot ${out}`;
+      } else {
+        try {
+          execSync11(`DISPLAY=:0 python3 -c "from PIL import ImageGrab; ImageGrab.grab().save(${JSON.stringify(outputPath)})"`, { stdio: "pipe", timeout: 1e4 });
+          if (existsSync12(outputPath))
+            return;
+        } catch {
+        }
       }
     }
   }
   if (!cmd) {
-    throw new Error("No screenshot tool found. Install one of: scrot, gnome-screenshot, imagemagick, or python3-pil");
+    throw new Error("No screenshot tool found. Auto-install failed. Try manually: sudo apt install scrot");
   }
-  execSync10(cmd, { stdio: "pipe", timeout: 1e4 });
+  execSync11(cmd, { stdio: "pipe", timeout: 1e4 });
   if (!existsSync12(outputPath)) {
     throw new Error("Screenshot file was not created");
   }
@@ -6599,12 +6828,17 @@ function clickAt(x, y, button, clickType) {
   const buttonNum = button === "right" ? 3 : button === "middle" ? 2 : 1;
   const clicks = clickType === "double" ? 2 : 1;
   if (plat === "linux") {
-    if (hasCommand("xdotool")) {
-      execSync10(`xdotool mousemove --sync ${rx} ${ry}`, { stdio: "pipe", timeout: 5e3 });
+    let hasXdotool = hasCommand2("xdotool");
+    if (!hasXdotool) {
+      const dep = ensureCommand("xdotool");
+      hasXdotool = dep.available;
+    }
+    if (hasXdotool) {
+      execSync11(`xdotool mousemove --sync ${rx} ${ry}`, { stdio: "pipe", timeout: 5e3 });
       if (clickType === "double") {
-        execSync10(`xdotool click --repeat 2 --delay 50 ${buttonNum}`, { stdio: "pipe", timeout: 5e3 });
+        execSync11(`xdotool click --repeat 2 --delay 50 ${buttonNum}`, { stdio: "pipe", timeout: 5e3 });
       } else {
-        execSync10(`xdotool click ${buttonNum}`, { stdio: "pipe", timeout: 5e3 });
+        execSync11(`xdotool click ${buttonNum}`, { stdio: "pipe", timeout: 5e3 });
       }
       return;
     }
@@ -6628,31 +6862,30 @@ for i in range(${clicks}):
     if i < ${clicks - 1}:
         time.sleep(0.05)
 `.trim().replace(/\n/g, "; ");
-    const pythonPaths = ["python3"];
     try {
-      execSync10(`DISPLAY=:0 python3 -c "${pyScript}"`, { stdio: "pipe", timeout: 5e3 });
+      execSync11(`DISPLAY=:0 python3 -c "${pyScript}"`, { stdio: "pipe", timeout: 5e3 });
       return;
     } catch {
     }
     try {
       const venvPy = join16(__dirname, "../../../../.moondream-venv/bin/python");
-      execSync10(`DISPLAY=:0 ${JSON.stringify(venvPy)} -c "${pyScript}"`, { stdio: "pipe", timeout: 5e3 });
+      execSync11(`DISPLAY=:0 ${JSON.stringify(venvPy)} -c "${pyScript}"`, { stdio: "pipe", timeout: 5e3 });
       return;
     } catch {
     }
-    throw new Error("No mouse control available. Install one of:\n  - xdotool: sudo apt install xdotool\n  - python-xlib: pip install python-xlib");
+    throw new Error("No mouse control available. Auto-install of xdotool failed. Try manually:\n  sudo apt install xdotool");
   } else if (plat === "darwin") {
-    if (!hasCommand("cliclick")) {
+    if (!hasCommand2("cliclick")) {
       throw new Error("cliclick not found. Install with: brew install cliclick");
     }
     const rx2 = Math.round(x);
     const ry2 = Math.round(y);
     if (clickType === "double") {
-      execSync10(`cliclick dc:${rx2},${ry2}`, { stdio: "pipe", timeout: 5e3 });
+      execSync11(`cliclick dc:${rx2},${ry2}`, { stdio: "pipe", timeout: 5e3 });
     } else if (button === "right") {
-      execSync10(`cliclick rc:${rx2},${ry2}`, { stdio: "pipe", timeout: 5e3 });
+      execSync11(`cliclick rc:${rx2},${ry2}`, { stdio: "pipe", timeout: 5e3 });
     } else {
-      execSync10(`cliclick c:${rx2},${ry2}`, { stdio: "pipe", timeout: 5e3 });
+      execSync11(`cliclick c:${rx2},${ry2}`, { stdio: "pipe", timeout: 5e3 });
     }
   } else {
     throw new Error(`Desktop click not supported on platform: ${plat}`);
@@ -6662,6 +6895,7 @@ var __dirname, DesktopClickTool, DesktopDescribeTool;
 var init_desktop_click = __esm({
   "packages/execution/dist/tools/desktop-click.js"() {
     "use strict";
+    init_system_deps();
     __dirname = dirname5(fileURLToPath2(import.meta.url));
     DesktopClickTool = class {
       workingDir;
@@ -6991,6 +7225,7 @@ var init_dist2 = __esm({
     init_structured_read();
     init_vision();
     init_desktop_click();
+    init_system_deps();
     init_shellRunner();
     init_gitWorktree();
     init_patchApplier();
@@ -11476,7 +11711,7 @@ var init_dist5 = __esm({
 });
 
 // packages/cli/dist/tui/listen.js
-import { spawn as spawn7, execSync as execSync11 } from "node:child_process";
+import { spawn as spawn7, execSync as execSync12 } from "node:child_process";
 import { existsSync as existsSync13, mkdirSync as mkdirSync5, writeFileSync as writeFileSync5 } from "node:fs";
 import { join as join19 } from "node:path";
 import { homedir as homedir6 } from "node:os";
@@ -11496,7 +11731,7 @@ function findMicCaptureCommand() {
   const platform3 = process.platform;
   if (platform3 === "linux") {
     try {
-      execSync11("which arecord", { stdio: "pipe" });
+      execSync12("which arecord", { stdio: "pipe" });
       return {
         cmd: "arecord",
         args: ["-f", "S16_LE", "-r", "16000", "-c", "1", "-t", "raw", "-q", "-"]
@@ -11506,7 +11741,7 @@ function findMicCaptureCommand() {
   }
   if (platform3 === "darwin") {
     try {
-      execSync11("which sox", { stdio: "pipe" });
+      execSync12("which sox", { stdio: "pipe" });
       return {
         cmd: "sox",
         args: ["-d", "-t", "raw", "-r", "16000", "-c", "1", "-b", "16", "-e", "signed-integer", "-"]
@@ -11515,7 +11750,7 @@ function findMicCaptureCommand() {
     }
   }
   try {
-    execSync11("which ffmpeg", { stdio: "pipe" });
+    execSync12("which ffmpeg", { stdio: "pipe" });
     if (platform3 === "linux") {
       return {
         cmd: "ffmpeg",
@@ -11564,7 +11799,7 @@ function ensureTranscribeCliBackground() {
     return;
   _bgInstallPromise = (async () => {
     try {
-      const globalRoot = execSync11("npm root -g", {
+      const globalRoot = execSync12("npm root -g", {
         encoding: "utf-8",
         timeout: 5e3,
         stdio: ["pipe", "pipe", "pipe"]
@@ -11671,7 +11906,7 @@ var init_listen = __esm({
         }
         if (!this.transcribeCliAvailable) {
           try {
-            execSync11("which transcribe-cli", { stdio: "pipe" });
+            execSync12("which transcribe-cli", { stdio: "pipe" });
             this.transcribeCliAvailable = true;
           } catch {
             this.transcribeCliAvailable = false;
@@ -11688,7 +11923,7 @@ var init_listen = __esm({
         } catch {
         }
         try {
-          const globalRoot = execSync11("npm root -g", {
+          const globalRoot = execSync12("npm root -g", {
             encoding: "utf-8",
             timeout: 5e3,
             stdio: ["pipe", "pipe", "pipe"]
@@ -11739,7 +11974,7 @@ var init_listen = __esm({
           if (!tc) {
             this.emit("info", "Installing transcribe-cli...");
             try {
-              execSync11("npm i -g transcribe-cli", { stdio: "pipe", timeout: 18e4 });
+              execSync12("npm i -g transcribe-cli", { stdio: "pipe", timeout: 18e4 });
               this.transcribeCliAvailable = null;
               tc = await this.loadTranscribeCli();
             } catch {
@@ -11750,14 +11985,23 @@ var init_listen = __esm({
           }
         }
         const TranscribeLive = tc.TranscribeLive;
-        this.liveTranscriber = new TranscribeLive({
-          model: this.config.model,
-          sampleRate: 16e3,
-          channels: 1,
-          sampleWidth: 2,
-          chunkDuration: 3
-          // 3s chunks for responsive transcription
-        });
+        if (!TranscribeLive) {
+          return "transcribe-cli does not export TranscribeLive. Try updating: npm i -g transcribe-cli@latest";
+        }
+        try {
+          this.liveTranscriber = new TranscribeLive({
+            model: this.config.model,
+            sampleRate: 16e3,
+            channels: 1,
+            sampleWidth: 2,
+            chunkDuration: 3
+            // 3s chunks for responsive transcription
+          });
+        } catch (err) {
+          const arch = process.arch;
+          const armHint = arch === "arm64" || arch === "arm" ? ` Live transcription may not be supported on ${process.platform}-${arch}.` : "";
+          return `Failed to create live transcriber.${armHint} Error: ${err instanceof Error ? err.message : String(err)}`;
+        }
         this.liveTranscriber.on("transcript", (evt) => {
           if (!evt.text.trim())
             return;
@@ -11771,17 +12015,28 @@ var init_listen = __esm({
         this.liveTranscriber.on("error", (err) => {
           this.emit("error", err);
         });
-        await new Promise((resolve19, reject) => {
-          const timeout = setTimeout(() => reject(new Error("Model load timeout (60s)")), 6e4);
-          this.liveTranscriber.on("ready", () => {
-            clearTimeout(timeout);
-            resolve19();
-          });
-          this.liveTranscriber.on("error", (err) => {
-            clearTimeout(timeout);
-            reject(err);
+        try {
+          await new Promise((resolve19, reject) => {
+            const timeout = setTimeout(() => reject(new Error("Model load timeout (60s)")), 6e4);
+            this.liveTranscriber.on("ready", () => {
+              clearTimeout(timeout);
+              resolve19();
+            });
+            this.liveTranscriber.on("error", (err) => {
+              clearTimeout(timeout);
+              reject(err);
+            });
           });
-        });
+        } catch (err) {
+          try {
+            this.liveTranscriber.stop();
+          } catch {
+          }
+          this.liveTranscriber = null;
+          const arch = process.arch;
+          const armHint = arch === "arm64" || arch === "arm" ? ` The whisper model or phonemizer WASM may not support ${process.platform}-${arch}. File transcription (/transcribe <file>) may still work.` : "";
+          return `Failed to start live transcription: ${err instanceof Error ? err.message : String(err)}${armHint}`;
+        }
         this.micProcess = spawn7(micCmd.cmd, micCmd.args, {
           stdio: ["pipe", "pipe", "pipe"],
           env: { ...process.env }
@@ -11876,7 +12131,7 @@ var init_listen = __esm({
         }
         if (!tc) {
           try {
-            execSync11("npm i -g transcribe-cli", { stdio: "pipe", timeout: 18e4 });
+            execSync12("npm i -g transcribe-cli", { stdio: "pipe", timeout: 18e4 });
             this.transcribeCliAvailable = null;
             tc = await this.loadTranscribeCli();
           } catch {
@@ -13428,7 +13683,7 @@ var init_oa_directory = __esm({
 
 // packages/cli/dist/tui/setup.js
 import * as readline from "node:readline";
-import { execSync as execSync12 } from "node:child_process";
+import { execSync as execSync13 } from "node:child_process";
 import { existsSync as existsSync15, writeFileSync as writeFileSync7, mkdirSync as mkdirSync7 } from "node:fs";
 import { join as join22 } from "node:path";
 import { homedir as homedir8 } from "node:os";
@@ -13438,7 +13693,7 @@ function detectSystemSpecs() {
   let gpuVramGB = 0;
   let gpuName = "";
   try {
-    const memInfo = execSync12("free -b 2>/dev/null || sysctl -n hw.memsize 2>/dev/null", {
+    const memInfo = execSync13("free -b 2>/dev/null || sysctl -n hw.memsize 2>/dev/null", {
       encoding: "utf8",
       timeout: 5e3
     });
@@ -13458,7 +13713,7 @@ function detectSystemSpecs() {
   } catch {
   }
   try {
-    const nvidiaSmi = execSync12("nvidia-smi --query-gpu=memory.total,name --format=csv,noheader,nounits 2>/dev/null", { encoding: "utf8", timeout: 5e3 });
+    const nvidiaSmi = execSync13("nvidia-smi --query-gpu=memory.total,name --format=csv,noheader,nounits 2>/dev/null", { encoding: "utf8", timeout: 5e3 });
     const lines = nvidiaSmi.trim().split("\n");
     if (lines.length > 0) {
       for (const line of lines) {
@@ -13520,7 +13775,7 @@ function ask(rl, question) {
 }
 function pullModelWithAutoUpdate(tag) {
   try {
-    execSync12(`ollama pull ${tag}`, {
+    execSync13(`ollama pull ${tag}`, {
       stdio: "inherit",
       timeout: 36e5
       // 1 hour max
@@ -13537,7 +13792,7 @@ function pullModelWithAutoUpdate(tag) {
 
 `);
       try {
-        execSync12("curl -fsSL https://ollama.com/install.sh | sh", {
+        execSync13("curl -fsSL https://ollama.com/install.sh | sh", {
           stdio: "inherit",
           timeout: 3e5
           // 5 min max for install
@@ -13548,7 +13803,7 @@ function pullModelWithAutoUpdate(tag) {
         process.stdout.write(`  ${c2.cyan("\u25CF")} Retrying pull of ${c2.bold(tag)}...
 
 `);
-        execSync12(`ollama pull ${tag}`, {
+        execSync13(`ollama pull ${tag}`, {
           stdio: "inherit",
           timeout: 36e5
         });
@@ -13869,7 +14124,7 @@ async function doSetup(config, rl) {
         const modelfilePath = join22(modelDir2, `Modelfile.${customName}`);
         writeFileSync7(modelfilePath, modelfileContent + "\n", "utf8");
         process.stdout.write(`  ${c2.dim("Creating model...")} `);
-        execSync12(`ollama create ${customName} -f ${modelfilePath}`, {
+        execSync13(`ollama create ${customName} -f ${modelfilePath}`, {
           stdio: "pipe",
           timeout: 12e4
         });
@@ -13954,7 +14209,7 @@ function createExpandedVariant(baseModel, specs, sizeGB) {
     mkdirSync7(modelDir2, { recursive: true });
     const modelfilePath = join22(modelDir2, `Modelfile.${customName}`);
     writeFileSync7(modelfilePath, modelfileContent + "\n", "utf8");
-    execSync12(`ollama create ${customName} -f ${modelfilePath}`, {
+    execSync13(`ollama create ${customName} -f ${modelfilePath}`, {
       stdio: "pipe",
       timeout: 12e4
     });
@@ -14624,9 +14879,9 @@ async function handleUpdate(subcommand, ctx) {
   process.stdout.write(`  ${c2.cyan("\u25CF")} Installing update...
 
 `);
-  const { execSync: execSync16 } = await import("node:child_process");
+  const { execSync: execSync17 } = await import("node:child_process");
   try {
-    execSync16(`npm cache clean --force open-agents-ai 2>/dev/null; npm install -g open-agents-ai@latest --force`, { stdio: "pipe", timeout: 18e4 });
+    execSync17(`npm cache clean --force open-agents-ai 2>/dev/null; npm install -g open-agents-ai@latest --force`, { stdio: "pipe", timeout: 18e4 });
   } catch {
     renderWarning("Update install failed. Try manually: npm i -g open-agents-ai");
     return;
@@ -14706,7 +14961,7 @@ var init_commands = __esm({
 // packages/cli/dist/tui/project-context.js
 import { existsSync as existsSync16, readFileSync as readFileSync13, readdirSync as readdirSync7 } from "node:fs";
 import { join as join23, basename as basename6 } from "node:path";
-import { execSync as execSync13 } from "node:child_process";
+import { execSync as execSync14 } from "node:child_process";
 import { homedir as homedir9, platform, release } from "node:os";
 function loadProjectFiles(repoRoot) {
   const discovered = discoverContextFiles(repoRoot);
@@ -14737,19 +14992,19 @@ function loadProjectMap(repoRoot) {
 }
 function getGitInfo(repoRoot) {
   try {
-    execSync13("git rev-parse --is-inside-work-tree", { cwd: repoRoot, stdio: "pipe" });
+    execSync14("git rev-parse --is-inside-work-tree", { cwd: repoRoot, stdio: "pipe" });
   } catch {
     return "";
   }
   const lines = [];
   try {
-    const branch = execSync13("git branch --show-current", { cwd: repoRoot, encoding: "utf-8", stdio: "pipe" }).trim();
+    const branch = execSync14("git branch --show-current", { cwd: repoRoot, encoding: "utf-8", stdio: "pipe" }).trim();
     if (branch)
       lines.push(`Branch: ${branch}`);
   } catch {
   }
   try {
-    const status = execSync13("git status --porcelain", { cwd: repoRoot, encoding: "utf-8", stdio: "pipe" }).trim();
+    const status = execSync14("git status --porcelain", { cwd: repoRoot, encoding: "utf-8", stdio: "pipe" }).trim();
     if (status) {
       const changed = status.split("\n").length;
       lines.push(`Working tree: ${changed} changed file(s)`);
@@ -14759,7 +15014,7 @@ function getGitInfo(repoRoot) {
   } catch {
   }
   try {
-    const log = execSync13("git log --oneline -5 --no-decorate", { cwd: repoRoot, encoding: "utf-8", stdio: "pipe" }).trim();
+    const log = execSync14("git log --oneline -5 --no-decorate", { cwd: repoRoot, encoding: "utf-8", stdio: "pipe" }).trim();
     if (log)
       lines.push(`Recent commits:
 ${log}`);
@@ -15807,7 +16062,7 @@ var init_carousel = __esm({
 import { existsSync as existsSync17, mkdirSync as mkdirSync8, writeFileSync as writeFileSync8, readFileSync as readFileSync14, unlinkSync as unlinkSync3 } from "node:fs";
 import { join as join24 } from "node:path";
 import { homedir as homedir10, tmpdir as tmpdir4, platform as platform2 } from "node:os";
-import { execSync as execSync14, spawn as nodeSpawn } from "node:child_process";
+import { execSync as execSync15, spawn as nodeSpawn } from "node:child_process";
 import { createRequire } from "node:module";
 function voiceDir() {
   return join24(homedir10(), ".open-agents", "voice");
@@ -16225,7 +16480,7 @@ var init_voice = __esm({
         }
         for (const player of ["paplay", "pw-play", "aplay"]) {
           try {
-            execSync14(`which ${player}`, { stdio: "pipe" });
+            execSync15(`which ${player}`, { stdio: "pipe" });
             return [player, path];
           } catch {
           }
@@ -16247,6 +16502,8 @@ var init_voice = __esm({
       async ensureRuntime() {
         if (this.ort)
           return;
+        const arch = process.arch;
+        const isArmLinux = (arch === "arm64" || arch === "arm") && process.platform === "linux";
         mkdirSync8(voiceDir(), { recursive: true });
         const pkgPath = join24(voiceDir(), "package.json");
         const expectedDeps = {
@@ -16274,16 +16531,20 @@ var init_voice = __esm({
         try {
           this.ort = voiceRequire("onnxruntime-node");
         } catch {
+          if (isArmLinux) {
+            throw new Error(`Voice synthesis (onnxruntime-node) is not available on ARM Linux (${arch}). Voice feedback is disabled on this architecture.`);
+          }
           renderInfo("Installing ONNX runtime for voice synthesis...");
           try {
-            execSync14("npm install --no-audit --no-fund", {
+            execSync15("npm install --no-audit --no-fund", {
               cwd: voiceDir(),
               stdio: "pipe",
               timeout: 12e4
             });
             this.ort = voiceRequire("onnxruntime-node");
           } catch (err) {
-            throw new Error(`Failed to install voice dependencies. Try manually: cd ${voiceDir()} && npm install
+            const armHint = arch !== "x64" ? ` onnxruntime-node may not support ${process.platform}-${arch}.` : "";
+            throw new Error(`Failed to install voice dependencies.${armHint} Try manually: cd ${voiceDir()} && npm install
 Error: ${err instanceof Error ? err.message : String(err)}`);
           }
         }
@@ -16291,9 +16552,12 @@ Error: ${err instanceof Error ? err.message : String(err)}`);
           const phonemizerMod = voiceRequire("phonemizer");
           this.phonemizeFn = phonemizerMod.phonemize ?? phonemizerMod.default?.phonemize ?? phonemizerMod;
         } catch {
+          if (isArmLinux) {
+            throw new Error(`Phonemizer (espeak-ng WASM) is not available on ARM Linux (${arch}). Voice feedback is disabled on this architecture.`);
+          }
           renderInfo("Installing phonemizer for voice synthesis...");
           try {
-            execSync14("npm install --no-audit --no-fund", {
+            execSync15("npm install --no-audit --no-fund", {
               cwd: voiceDir(),
               stdio: "pipe",
               timeout: 12e4
@@ -16301,7 +16565,8 @@ Error: ${err instanceof Error ? err.message : String(err)}`);
             const phonemizerMod = voiceRequire("phonemizer");
             this.phonemizeFn = phonemizerMod.phonemize ?? phonemizerMod.default?.phonemize ?? phonemizerMod;
           } catch (err) {
-            throw new Error(`Failed to install phonemizer. Try manually: cd ${voiceDir()} && npm install
+            const armHint = arch !== "x64" ? ` phonemizer WASM may not support ${process.platform}-${arch}.` : "";
+            throw new Error(`Failed to install phonemizer.${armHint} Try manually: cd ${voiceDir()} && npm install
 Error: ${err instanceof Error ? err.message : String(err)}`);
           }
         }
@@ -16983,7 +17248,7 @@ var init_edit_history = __esm({
 // packages/cli/dist/tui/dream-engine.js
 import { mkdirSync as mkdirSync10, writeFileSync as writeFileSync9, readFileSync as readFileSync15, existsSync as existsSync18, cpSync, rmSync, readdirSync as readdirSync8 } from "node:fs";
 import { join as join26, basename as basename7 } from "node:path";
-import { execSync as execSync15 } from "node:child_process";
+import { execSync as execSync16 } from "node:child_process";
 function adaptTool(tool) {
   return {
     name: tool.name,
@@ -17236,7 +17501,7 @@ var init_dream_engine = __esm({
           }
         }
         try {
-          const output = execSync15(cmd, {
+          const output = execSync16(cmd, {
             cwd: this.repoRoot,
             timeout: 3e4,
             encoding: "utf-8",
@@ -17456,17 +17721,17 @@ Dreams directory: ${this.dreamsDir}`);
         try {
           mkdirSync10(checkpointDir, { recursive: true });
           try {
-            const gitStatus = execSync15("git status --porcelain", {
+            const gitStatus = execSync16("git status --porcelain", {
               cwd: this.repoRoot,
               encoding: "utf-8",
               timeout: 1e4
             });
-            const gitDiff = execSync15("git diff", {
+            const gitDiff = execSync16("git diff", {
               cwd: this.repoRoot,
               encoding: "utf-8",
               timeout: 1e4
             });
-            const gitHash = execSync15("git rev-parse HEAD 2>/dev/null || echo 'no-git'", {
+            const gitHash = execSync16("git rev-parse HEAD 2>/dev/null || echo 'no-git'", {
               cwd: this.repoRoot,
               encoding: "utf-8",
               timeout: 5e3

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "open-agents-ai",
-  "version": "0.18.0",
+  "version": "0.19.0",
   "description": "AI coding agent powered by open-source models (Ollama/vLLM) — interactive TUI with agentic tool-calling loop",
   "type": "module",
   "main": "./dist/index.js",