open-agents-ai 0.184.78 → 0.184.80

package/dist/index.js

@@ -16303,20 +16303,51 @@ var init_desktop_click = __esm({
               const envModel = process.env["OLLAMA_VISION_MODEL"];
               const ollamaModel = envModel || (this._activeModelHasVision && this._activeModel ? this._activeModel : "moondream");
               const imageBase64 = readFileSync15(screenshotPath).toString("base64");
-              const res = await fetch(`${ollamaHost}/api/generate`, {
+              let res = await fetch(`${ollamaHost}/api/generate`, {
                 method: "POST",
                 headers: { "Content-Type": "application/json" },
-                body: JSON.stringify({ model: ollamaModel, prompt: `Point to ${target}`, images: [imageBase64], stream: false }),
+                body: JSON.stringify({ model: ollamaModel, prompt: `Locate the "${target}" in this image and give its x,y position as normalized coordinates.`, images: [imageBase64], stream: false }),
                 signal: AbortSignal.timeout(6e4)
               });
+              if (!res.ok && ollamaModel === "moondream") {
+                const errText = await res.text().catch(() => "");
+                if (res.status === 404 || /not found|does not exist/i.test(errText)) {
+                  try {
+                    const { execSync: es } = await import("node:child_process");
+                    es("ollama pull moondream", { timeout: 3e5, stdio: "pipe" });
+                    res = await fetch(`${ollamaHost}/api/generate`, {
+                      method: "POST",
+                      headers: { "Content-Type": "application/json" },
+                      body: JSON.stringify({ model: ollamaModel, prompt: `Locate the "${target}" in this image and give its x,y position as normalized coordinates.`, images: [imageBase64], stream: false }),
+                      signal: AbortSignal.timeout(6e4)
+                    });
+                  } catch {
+                  }
+                }
+              }
               if (res.ok) {
                 const data = await res.json();
                 const response = data.response || "";
-                const pointMatches = [...response.matchAll(/<point\s+x="([\d.]+)"\s+y="([\d.]+)"\s*\/?>/g)];
-                if (pointMatches.length > 0) {
-                  points = pointMatches.map((m) => ({ x: parseFloat(m[1]), y: parseFloat(m[2]) }));
+                const xmlMatches = [...response.matchAll(/<point\s+x="([\d.]+)"\s+y="([\d.]+)"\s*\/?>/g)];
+                if (xmlMatches.length > 0) {
+                  points = xmlMatches.map((m) => ({ x: parseFloat(m[1]), y: parseFloat(m[2]) }));
                   visionWorked = true;
                 }
+                if (!visionWorked) {
+                  const arrayMatch = response.match(/\[\s*([\d.]+)\s*,\s*([\d.]+)(?:\s*,\s*([\d.]+)\s*,\s*([\d.]+))?\s*\]/);
+                  if (arrayMatch) {
+                    const x1 = parseFloat(arrayMatch[1]);
+                    const y1 = parseFloat(arrayMatch[2]);
+                    if (arrayMatch[3] && arrayMatch[4]) {
+                      const x2 = parseFloat(arrayMatch[3]);
+                      const y2 = parseFloat(arrayMatch[4]);
+                      points = [{ x: (x1 + x2) / 2, y: (y1 + y2) / 2 }];
+                    } else {
+                      points = [{ x: x1, y: y1 }];
+                    }
+                    visionWorked = true;
+                  }
+                }
               }
             } catch {
             }
@@ -16480,12 +16511,28 @@ ${caption}`);
               const ollamaModel = envModel || (this._activeModelHasVision && this._activeModel ? this._activeModel : "moondream");
               const imageBase64 = imageBuffer.toString("base64");
               const ollamaPrompt = question || "Describe what you see on this desktop screenshot in detail. Include visible applications, windows, text, and UI elements.";
-              const res = await fetch(`${ollamaHost}/api/generate`, {
+              let res = await fetch(`${ollamaHost}/api/generate`, {
                 method: "POST",
                 headers: { "Content-Type": "application/json" },
                 body: JSON.stringify({ model: ollamaModel, prompt: ollamaPrompt, images: [imageBase64], stream: false }),
                 signal: AbortSignal.timeout(6e4)
               });
+              if (!res.ok && ollamaModel === "moondream") {
+                const errText = await res.text().catch(() => "");
+                if (res.status === 404 || /not found|does not exist/i.test(errText)) {
+                  try {
+                    const { execSync: es } = await import("node:child_process");
+                    es("ollama pull moondream", { timeout: 3e5, stdio: "pipe" });
+                    res = await fetch(`${ollamaHost}/api/generate`, {
+                      method: "POST",
+                      headers: { "Content-Type": "application/json" },
+                      body: JSON.stringify({ model: ollamaModel, prompt: ollamaPrompt, images: [imageBase64], stream: false }),
+                      signal: AbortSignal.timeout(6e4)
+                    });
+                  } catch {
+                  }
+                }
+              }
               if (res.ok) {
                 const data = await res.json();
                 if (data.response) {
@@ -51443,16 +51490,20 @@ async function handleUpdate(subcommand, ctx) {
   await new Promise((r) => setTimeout(r, 200));
   const { execPath, argv } = process;
   try {
-    const { execFileSync } = await import("node:child_process");
+    const { spawn: spawnChild } = await import("node:child_process");
     if (process.stdout.isTTY) {
       process.stdout.write("\x1B[?1002l\x1B[?1003l\x1B[?1006l\x1B[?25h\x1B[?1049l\x1B[0m");
     }
-    process.env.__OA_RESUMED = resumeFlag;
-    execFileSync(execPath, argv.slice(1), {
+    const child = spawnChild(execPath, argv.slice(1), {
       stdio: "inherit",
-      env: { ...process.env, __OA_RESUMED: resumeFlag }
+      env: { ...process.env, __OA_RESUMED: resumeFlag },
+      detached: false
     });
-    process.exit(0);
+    child.on("exit", (code) => {
+      process.exit(code ?? 0);
+    });
+    process.removeAllListeners("SIGINT");
+    process.removeAllListeners("SIGTERM");
   } catch {
     process.stderr.write("\x1B[0m\nRestart oa manually to use the new version.\n");
     process.exit(1);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "open-agents-ai",
-  "version": "0.184.78",
+  "version": "0.184.80",
   "description": "AI coding agent powered by open-source models (Ollama/vLLM) — interactive TUI with agentic tool-calling loop",
   "type": "module",
   "main": "./dist/index.js",