fix(responses): normalize multimodal content in tool call output items

nginx.conf

@@ -33,12 +33,21 @@ http {
     server {
         listen 80;
 
+        # Allow large payloads for multimodal images / base64 payloads
+        client_max_body_size 32m;
+
         location / {
             limit_req zone=mylimit burst=20 nodelay;
 
             proxy_pass http://windsurf_backend;
             proxy_http_version 1.1;
 
+            # Timeouts for long-running streaming completions and thinking models
+            # 必须严格大于应用层 DEVIN_TIMEOUT_MS（默认 600s）。改任意一边都要重新核对。
+            proxy_connect_timeout 60s;
+            proxy_send_timeout 900s;
+            proxy_read_timeout 900s;
+
             # SSE and streaming requirements
             proxy_set_header Connection '';
             proxy_buffering off;

src/handlers/responses.js

@@ -38,9 +38,19 @@ function normalizeMessageContent(content) {
   if (typeof content === 'string') return content;
   if (!Array.isArray(content)) return stringifyMaybe(content);
 
+  const hasContentBlock = content.some(part => part && typeof part === 'object' && typeof part.type === 'string');
+  if (!hasContentBlock) {
+    return stringifyMaybe(content);
+  }
+
   const out = [];
   for (const part of content) {
-    if (!part || typeof part !== 'object') continue;
+    if (!part) continue;
+    if (typeof part === 'string') {
+      out.push({ type: 'text', text: part });
+      continue;
+    }
+    if (typeof part !== 'object') continue;
     if (part.type === 'input_text' || part.type === 'output_text' || part.type === 'text') {
       out.push({ type: 'text', text: part.text || '' });
     } else if (part.type === 'input_image') {
@@ -423,7 +433,7 @@ export function responsesToChat(body) {
         messages.push({
           role: 'tool',
           tool_call_id: item.call_id || item.id,
-          content: stringifyMaybe(item.output ?? ''),
+          content: normalizeMessageContent(item.output ?? ''),
         });
       } else if (item.type === 'custom_tool_call') {
         flushToolCalls.add({
@@ -437,7 +447,7 @@ export function responsesToChat(body) {
         messages.push({
           role: 'tool',
           tool_call_id: item.call_id || item.id,
-          content: stringifyMaybe(item.output ?? ''),
+          content: normalizeMessageContent(item.output ?? ''),
         });
       }
     }

test/responses.test.js

@@ -958,4 +958,59 @@ describe('H-1 (audit 2026-07-13): input_image string image_url normalizes to obj
     assert.equal(imgs[0].image_url.detail, 'high');
     assert.equal(imgs[1].image_url.url, 'https://z/w.png');
   });
+
+  it('normalizes multimodal function_call_output and custom_tool_call_output items', () => {
+    const { messages } = responsesToChat({
+      model: 'gpt-4o',
+      input: [
+        {
+          type: 'function_call_output',
+          call_id: 'call_img',
+          output: [
+            { type: 'input_text', text: 'Screenshot captured' },
+            { type: 'input_image', image_url: 'data:image/jpeg;base64,12345' },
+          ],
+        },
+        {
+          type: 'custom_tool_call_output',
+          call_id: 'call_custom_img',
+          output: [
+            { type: 'input_image', image_url: 'https://example.com/test.png' },
+          ],
+        },
+      ],
+    });
+    assert.equal(messages.length, 2);
+    assert.equal(messages[0].role, 'tool');
+    assert.equal(messages[0].tool_call_id, 'call_img');
+    assert.deepEqual(messages[0].content, [
+      { type: 'text', text: 'Screenshot captured' },
+      { type: 'image_url', image_url: { url: 'data:image/jpeg;base64,12345' } },
+    ]);
+    assert.equal(messages[1].role, 'tool');
+    assert.equal(messages[1].tool_call_id, 'call_custom_img');
+    assert.deepEqual(messages[1].content, [
+      { type: 'image_url', image_url: { url: 'https://example.com/test.png' } },
+    ]);
+  });
+
+  it('preserves non-content array tool outputs as JSON string', () => {
+    const { messages } = responsesToChat({
+      model: 'gpt-4o',
+      input: [
+        {
+          type: 'function_call_output',
+          call_id: 'call_list',
+          output: ['file1.txt', 'file2.txt'],
+        },
+        {
+          type: 'function_call_output',
+          call_id: 'call_objects',
+          output: [{ name: 'alice' }, { name: 'bob' }],
+        },
+      ],
+    });
+    assert.equal(messages[0].content, JSON.stringify(['file1.txt', 'file2.txt']));
+    assert.equal(messages[1].content, JSON.stringify([{ name: 'alice' }, { name: 'bob' }]));
+  });
 });