@duheso/zerocli 1.1.8 → 1.1.10

package/dist/cli.mjs

@@ -119226,7 +119226,7 @@ function buildProviderInfoLines() {
   const sLen = ` ● ${sL}    ${sReady}`.length;
   out.push(boxRow(sRow, W2, sLen));
   out.push(`${rgb(...BORDER)}╚${"═".repeat(W2 - 2)}╝${RESET}`);
-  out.push(`  ${DIM}${rgb(...DIMCOL)}zero ${RESET}${rgb(...ACCENT)}v${"1.1.8"}${RESET}`);
+  out.push(`  ${DIM}${rgb(...DIMCOL)}zero ${RESET}${rgb(...ACCENT)}v${"1.1.10"}${RESET}`);
   return out;
 }
 function printStartupScreen() {}
@@ -365596,7 +365596,7 @@ __export(exports_VerifyPlanExecutionTool, {
 var VerifyPlanExecutionTool_default = null, VerifyPlanExecutionTool = null;
 
 // src/tools/WindowsControlTool/constants.ts
-var TOOL_NAME = "windows_control", DEFAULT_WS_PORT = 50052, DEFAULT_REST_PORT = 50051, DEFAULT_QUALITY = 75, DEFAULT_SCALE = 0.5, RECONNECT_INTERVAL_MS = 5000, PING_INTERVAL_MS = 1e4, ENV_WINDOWS_HOST = "WINDOWS_CONTROL_HOST", ENV_WINDOWS_PORT = "WINDOWS_CONTROL_PORT", ENV_WINDOWS_TOKEN = "WINDOWS_CONTROL_TOKEN";
+var TOOL_NAME = "windows_control", DEFAULT_WS_PORT = 50052, DEFAULT_REST_PORT = 50051, DEFAULT_QUALITY = 75, DEFAULT_SCALE = 1, RECONNECT_INTERVAL_MS = 5000, PING_INTERVAL_MS = 1e4, ENV_WINDOWS_HOST = "WINDOWS_CONTROL_HOST", ENV_WINDOWS_PORT = "WINDOWS_CONTROL_PORT", ENV_WINDOWS_TOKEN = "WINDOWS_CONTROL_TOKEN";
 
 // src/tools/WindowsControlTool/prompt.ts
 var WINDOWS_CONTROL_PROMPT = `
@@ -365633,7 +365633,7 @@ in their .env file. That's it.
 
 **Screen:**
 - \`screenshot\` — See the screen (ALWAYS start with this). Returns the actual image.
-- \`extract_text\` — OCR: read text visible on screen or a region
+  **To read text on screen:** just take a screenshot — YOU (the LLM) can read it directly via vision. No OCR needed.
 - \`list_monitors\` — Multi-monitor layout info
 
 **Mouse:**
@@ -365673,7 +365673,13 @@ in their .env file. That's it.
 
 ### Coordinate System
 - (0,0) = top-left corner of primary monitor
-- Screenshot response includes cursor position and screen dimensions
+- **Screenshots are taken at scale=1.0 by default** (native resolution, 1:1 pixel mapping).
+  Coordinates you read from the screenshot can be passed directly to mouse_click/mouse_move.
+- If you use \`scale: 0.5\` for smaller images, the screenshot will be half-size.
+  In that case, **multiply all coordinates by 2** before passing to mouse actions.
+  The response includes \`nativeWidth\`, \`nativeHeight\`, and \`scale\` to help you calculate.
+  Formula: \`real_x = screenshot_x * (nativeWidth / width)\`
+- Cursor position in the response is always in **native screen coordinates**.
 - Use \`list_monitors\` for multi-monitor setups
 
 ### Common Patterns
@@ -365691,7 +365697,7 @@ in their .env file. That's it.
   key_combo(["ctrl","s"]) to save, key_combo(["alt","f4"]) to close, etc.
 
 **Find and click by text (when you can't see coordinates):**
-  extract_text → find position of target text → mouse_click at that position
+  screenshot → read text from the image → identify position → mouse_click at that position
 
 **Navigate menus:**
   mouse_click on menu → screenshot → mouse_click on menu item
@@ -366174,13 +366180,17 @@ async function handleCaptureAction(input, connection) {
         return { success: false, error: response.error || "Screenshot failed" };
       }
       const imageBase64 = response.imageData;
+      const scale = input.scale || DEFAULT_SCALE;
       return {
         success: true,
-        message: `Screenshot captured: ${response.width}x${response.height}`,
+        message: `Screenshot captured: ${response.width}x${response.height}` + (scale < 1 ? ` (scaled ${Math.round(scale * 100)}% — multiply coordinates by ${Math.round(1 / scale)} for clicks)` : ""),
         data: {
           imageBase64,
           width: response.width,
           height: response.height,
+          nativeWidth: Math.round(response.width / scale),
+          nativeHeight: Math.round(response.height / scale),
+          scale,
           cursor: { x: response.cursorX, y: response.cursorY },
           format: response.imageFormat || "jpeg"
         }
@@ -366234,22 +366244,6 @@ async function handleAutomationAction(input, connection) {
         value: input.value
       });
       return { success: true, message: `Set value on element ${input.runtime_id}` };
-    case "extract_text": {
-      const response = await connection.sendCommand("ExtractText", {
-        region: input.region || null,
-        monitorIndex: input.monitor_index ?? 0,
-        language: input.language || "en-US"
-      });
-      return {
-        success: true,
-        message: `OCR extracted ${response.lines?.length || 0} lines`,
-        data: {
-          fullText: response.fullText,
-          lines: response.lines,
-          languageDetected: response.languageDetected
-        }
-      };
-    }
     default:
       return { success: false, error: `Unknown automation action: ${input.action}` };
   }
@@ -366500,7 +366494,6 @@ var init_WindowsControlTool = __esm(() => {
     "get_ui_elements",
     "click_element",
     "set_element_value",
-    "extract_text",
     "get_clipboard",
     "set_clipboard",
     "upload_file",
@@ -366515,7 +366508,7 @@ var init_WindowsControlTool = __esm(() => {
   KEYBOARD_ACTIONS = new Set(["key_press", "key_combo", "key_hold", "type_text"]);
   WINDOW_ACTIONS = new Set(["list_windows", "focus_window", "move_window", "resize_window", "minimize", "maximize", "restore", "close_window", "snap_window"]);
   CAPTURE_ACTIONS = new Set(["screenshot", "list_monitors"]);
-  AUTOMATION_ACTIONS = new Set(["get_ui_elements", "click_element", "set_element_value", "extract_text"]);
+  AUTOMATION_ACTIONS = new Set(["get_ui_elements", "click_element", "set_element_value"]);
   CLIPBOARD_ACTIONS = new Set(["get_clipboard", "set_clipboard"]);
   DESKTOP_ACTIONS = new Set(["list_desktops", "switch_desktop", "create_desktop", "delete_desktop"]);
   FILE_PROCESS_ACTIONS = new Set(["upload_file", "download_file", "list_files", "list_processes", "kill_process", "launch_process"]);
@@ -366559,7 +366552,6 @@ var init_WindowsControlTool = __esm(() => {
     value: exports_external.string().optional().describe("Value to set on UI element"),
     max_depth: exports_external.number().optional().describe("UI tree traversal depth"),
     control_type_filter: exports_external.string().optional().describe("Filter UI elements by type"),
-    language: exports_external.string().optional().describe("OCR language (e.g. en-US, pt-BR)"),
     clipboard_text: exports_external.string().optional().describe("Text to set in clipboard"),
     include_image: exports_external.boolean().optional().describe("Include image from clipboard"),
     file_path: exports_external.string().optional().describe("File path on Windows"),
@@ -379565,7 +379557,7 @@ function getAnthropicEnvMetadata() {
 function getBuildAgeMinutes() {
   if (false)
     ;
-  const buildTime = new Date("2026-05-23T21:39:04.216Z").getTime();
+  const buildTime = new Date("2026-05-23T22:30:05.775Z").getTime();
   if (isNaN(buildTime))
     return;
   return Math.floor((Date.now() - buildTime) / 60000);
@@ -421589,7 +421581,7 @@ function buildPrimarySection() {
   });
   return [{
     label: "Version",
-    value: "1.1.8"
+    value: "1.1.10"
   }, {
     label: "Session name",
     value: nameValue
@@ -489935,7 +489927,7 @@ var init_bridge_kick = __esm(() => {
 var call60 = async () => {
   return {
     type: "text",
-    value: `${"1.1.8"} (built ${"2026-05-23T21:39:04.216Z"})`
+    value: `${"1.1.8"} (built ${"2026-05-23T22:30:05.775Z"})`
   };
 }, version2, version_default;
 var init_version = __esm(() => {
@@ -565959,7 +565951,7 @@ function WelcomeV2() {
             dimColor: true,
             children: [
               "v",
-              "1.1.8",
+              "1.1.10",
               " "
             ]
           })
@@ -566193,7 +566185,7 @@ function WelcomeV2() {
           dimColor: true,
           children: [
             "v",
-            "1.1.8",
+            "1.1.10",
             " "
           ]
         })
@@ -566440,7 +566432,7 @@ function AppleTerminalWelcomeV2(t0) {
         dimColor: true,
         children: [
           "v",
-          "1.1.8",
+          "1.1.10",
           " "
         ]
       });
@@ -566713,7 +566705,7 @@ function AppleTerminalWelcomeV2(t0) {
       dimColor: true,
       children: [
         "v",
-        "1.1.8",
+        "1.1.10",
         " "
       ]
     });
@@ -586958,7 +586950,7 @@ Usage: claude --remote "your task description"`, () => gracefulShutdown(1));
         pendingHookMessages
       }, renderAndRun);
     }
-  }).version("1.1.8", "-v, --version", "Output the version number");
+  }).version("1.1.10", "-v, --version", "Output the version number");
   program2.option("-w, --worktree [name]", "Create a new git worktree for this session (optionally specify a name)");
   program2.option("--tmux", "Create a tmux session for the worktree (requires --worktree). Uses iTerm2 native panes when available; use --tmux=classic for traditional tmux.");
   if (canUserConfigureAdvisor()) {
@@ -587624,7 +587616,7 @@ if (false) {}
 async function main2() {
   const args = process.argv.slice(2);
   if (args.length === 1 && (args[0] === "--version" || args[0] === "-v" || args[0] === "-V")) {
-    console.log(`${"1.1.8"} (ZeroCLI)`);
+    console.log(`${"1.1.10"} (ZeroCLI)`);
     return;
   }
   if (args.includes("--provider")) {
@@ -587766,4 +587758,4 @@ async function main2() {
 }
 main2();
 
-//# debugId=5E2ECDA4E7219CF964756E2164756E21
+//# debugId=BCBBAE4846FEE42064756E2164756E21

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@duheso/zerocli",
-  "version": "1.1.8",
-  "description": "Zero CLI to any LLM \u2014 OpenAI, Gemini, DeepSeek, Ollama, and 200+ models",
+  "version": "1.1.10",
+  "description": "Zero CLI to any LLM — OpenAI, Gemini, DeepSeek, Ollama, and 200+ models",
   "type": "module",
   "bin": {
     "zero": "bin/zero"
@@ -171,4 +171,4 @@
   "overrides": {
     "lodash-es": "4.18.1"
   }
-}
+}