opencode-qwen-cli-auth 2.3.3 → 2.3.4

package/dist/index.js

@@ -22,7 +22,7 @@ import { PROVIDER_ID, AUTH_LABELS, DEVICE_FLOW, PORTAL_HEADERS } from "./lib/con
 import { logError, logInfo, logWarn, LOGGING_ENABLED } from "./lib/logger.js";
 
 /** Request timeout for chat completions in milliseconds */
-const CHAT_REQUEST_TIMEOUT_MS = 120000;
+const CHAT_REQUEST_TIMEOUT_MS = 30000;
 /** Maximum number of retry attempts for failed requests */
 const CHAT_MAX_RETRIES = 3;
 /** Output token cap for coder-model (64K tokens) */
@@ -415,19 +415,19 @@ function sanitizeOutgoingPayload(payload) {
 function createQuotaDegradedPayload(payload) {
     const degraded = { ...payload };
     let changed = false;
-    // Remove tool-related fields (skip removing tools so Agents don't break)
-    // if ("tools" in degraded) {
-    //     delete degraded.tools;
-    //     changed = true;
-    // }
-    // if ("tool_choice" in degraded) {
-    //     delete degraded.tool_choice;
-    //     changed = true;
-    // }
-    // if ("parallel_tool_calls" in degraded) {
-    //     delete degraded.parallel_tool_calls;
-    //     changed = true;
-    // }
+    // Remove tool-related fields
+    if ("tools" in degraded) {
+        delete degraded.tools;
+        changed = true;
+    }
+    if ("tool_choice" in degraded) {
+        delete degraded.tool_choice;
+        changed = true;
+    }
+    if ("parallel_tool_calls" in degraded) {
+        delete degraded.parallel_tool_calls;
+        changed = true;
+    }
     // Disable streaming
     if (degraded.stream !== false) {
         degraded.stream = false;
@@ -595,7 +595,7 @@ function createSseResponseChunk(data) {
  * @param {boolean} streamMode - Whether to return streaming response
  * @returns {Response} Formatted completion response
  */
-function makeQwenCliCompletionResponse(model, content, context, streamMode, abortSignal) {
+function makeQwenCliCompletionResponse(model, content, context, streamMode) {
     if (LOGGING_ENABLED) {
         logInfo("Qwen CLI fallback returned completion", {
             request_id: context.requestId,
@@ -609,7 +609,7 @@ function makeQwenCliCompletionResponse(model, content, context, streamMode, abor
         const encoder = new TextEncoder();
         const stream = new ReadableStream({
             start(controller) {
-                // Send first chunk with empty content
+                // Send first chunk with content
                 controller.enqueue(encoder.encode(createSseResponseChunk({
                     id: completionId,
                     object: "chat.completion.chunk",
@@ -618,51 +618,28 @@ function makeQwenCliCompletionResponse(model, content, context, streamMode, abor
                     choices: [
                         {
                             index: 0,
-                            delta: { role: "assistant", content: "" },
+                            delta: { role: "assistant", content },
                             finish_reason: null,
                         },
                     ],
                 })));
-
-                const CHUNK_SIZE = 15;
-                const DELAY_MS = 20;
-                let position = 0;
-
-                function pushNextChunk() {
-                    if (abortSignal?.aborted) {
-                        try { controller.close(); } catch (e) { }
-                        return;
-                    }
-
-                    if (position >= content.length) {
-                        // Send stop chunk
-                        controller.enqueue(encoder.encode(createSseResponseChunk({
-                            id: completionId,
-                            object: "chat.completion.chunk",
-                            created,
-                            model,
-                            choices: [{ index: 0, delta: {}, finish_reason: "stop" }],
-                        })));
-                        controller.enqueue(encoder.encode("data: [DONE]\n\n"));
-                        try { controller.close(); } catch (e) { }
-                        return;
-                    }
-
-                    const nextSlice = content.slice(position, position + CHUNK_SIZE);
-                    position += CHUNK_SIZE;
-
-                    controller.enqueue(encoder.encode(createSseResponseChunk({
-                        id: completionId,
-                        object: "chat.completion.chunk",
-                        created,
-                        model,
-                        choices: [{ index: 0, delta: { content: nextSlice }, finish_reason: null }],
-                    })));
-
-                    setTimeout(pushNextChunk, DELAY_MS);
-                }
-
-                pushNextChunk();
+                // Send stop chunk
+                controller.enqueue(encoder.encode(createSseResponseChunk({
+                    id: completionId,
+                    object: "chat.completion.chunk",
+                    created,
+                    model,
+                    choices: [
+                        {
+                            index: 0,
+                            delta: {},
+                            finish_reason: "stop",
+                        },
+                    ],
+                })));
+                // Send DONE marker
+                controller.enqueue(encoder.encode("data: [DONE]\n\n"));
+                controller.close();
             },
         });
         return new Response(stream, {
@@ -724,8 +701,12 @@ async function runQwenCliFallback(payload, context, abortSignal) {
             command: QWEN_CLI_COMMAND,
         });
     }
-    // Use secure spawn logic across ALL OS, allowing .cmd locally on Windows by injecting shell correctly.
-    const isShellRequired = requiresShellExecution(QWEN_CLI_COMMAND);
+    if (requiresShellExecution(QWEN_CLI_COMMAND)) {
+        return {
+            ok: false,
+            reason: "cli_shell_execution_blocked_for_security",
+        };
+    }
     return await new Promise((resolve) => {
         let settled = false;
         let stdout = "";
@@ -755,7 +736,7 @@ async function runQwenCliFallback(payload, context, abortSignal) {
         }
         try {
             child = spawn(QWEN_CLI_COMMAND, args, {
-                shell: isShellRequired,
+                shell: false,
                 windowsHide: true,
                 stdio: ["ignore", "pipe", "pipe"],
             });
@@ -810,7 +791,7 @@ async function runQwenCliFallback(payload, context, abortSignal) {
             if (content) {
                 finalize({
                     ok: true,
-                    response: makeQwenCliCompletionResponse(model, content, context, streamMode, abortSignal),
+                    response: makeQwenCliCompletionResponse(model, content, context, streamMode),
                 });
                 return;
             }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "opencode-qwen-cli-auth",
-  "version": "2.3.3",
+  "version": "2.3.4",
   "description": "Qwen OAuth authentication plugin for opencode - use your Qwen account instead of API keys",
   "main": "./dist/index.js",
   "types": "./dist/index.d.ts",