feat: 升级至 v0.3.15 — MiMo 多模态 + 工作空间数据库继承 + 引导窗口上下文配置 + TS 错误修复

- feat(mimo): 适配器支持图片输入，将 images 转为 OpenAI 兼容 content parts 数组
- feat(workspace): 切换工作空间时支持继承数据库 agent.db（SQLite backup API 原子导出，自动 checkpoint WAL）
- feat(onboarding): 引导弹窗 LLM 配置加上下文长度字段（DeepSeek/Agnes=1000000、MiMo=131072、Ollama=null）
- fix(types): 修复 16 个 TypeScript 编译错误（setConfigLoaded 接口声明、10 个 IPC 返回类型加 error 字段、MUI TextField readOnly slot 迁移）
- chore(workspace): 移除死代码 traces 目录（trace 数据实际存储于数据库 sessions.metadata 字段，traces 目录从未被任何代码读写）
- docs(mimo): API 文档补充多模态输入章节和 curl/Python 代码示例
- chore: 版本号 0.3.14 → 0.3.15

apis/mimo-api-docs-20260715.html

@@ -132,6 +132,7 @@
     <a href="#tools">🔧 工具调用 (Function Calling)</a>
     <a href="#websearch">🌐 联网搜索</a>
     <a href="#tts">🎙️ 语音合成 (TTS)</a>
+    <a href="#multimodal">🖼️ 多模态输入</a>
     <a href="#examples">💻 代码示例</a>
     <a href="#errors">⚠️ 错误码说明</a>
   </div>
@@ -331,6 +332,56 @@ Content-Type: application/json</pre>
   </div>
 </section>
 
+<!-- ===== MULTIMODAL ===== -->
+<section id="multimodal">
+  <h2>🖼️ 多模态输入</h2>
+  <p>当 <code>messages[].content</code> 为数组时，可混合传入文本、图像内容。采用 OpenAI 兼容的 content parts 数组格式。</p>
+
+  <h3>content parts 字段</h3>
+  <table>
+    <tr><th width="200">字段名</th><th width="90">类型</th><th>描述</th></tr>
+    <tr><td><code>content</code></td><td><span class="type-tag">string | array</span></td><td>纯文本字符串，或 content parts 数组</td></tr>
+    <tr><td><code>content[].type</code></td><td><span class="type-tag">string</span></td><td>part 类型：<code>"text"</code> / <code>"image_url"</code></td></tr>
+    <tr><td><code>content[].text</code></td><td><span class="type-tag">string</span></td><td>文本内容（type 为 <code>"text"</code> 时必填）</td></tr>
+    <tr><td><code>content[].image_url</code></td><td><span class="type-tag">object</span></td><td>图像对象（type 为 <code>"image_url"</code> 时必填）</td></tr>
+    <tr><td><code>content[].image_url.url</code></td><td><span class="type-tag">string</span></td><td>图像公网 URL 或 base64 data URI（格式：<code>data:image/png;base64,...</code>）</td></tr>
+    <tr><td><code>content[].image_url.detail</code></td><td><span class="type-tag">string</span></td><td>可选，清晰度：<code>"low"</code> / <code>"high"</code> / <code>"auto"</code>（默认 <code>"auto"</code>）</td></tr>
+  </table>
+
+  <div class="info-box">
+    <strong>📌 支持的图像格式：</strong>
+    <ul style="padding-left:20px;margin:8px 0 0;">
+      <li><strong>URL</strong>：公网可访问的 HTTPS 图片地址</li>
+      <li><strong>Base64 data URI</strong>：<code>data:image/{format};base64,{base64_data}</code>，支持 png / jpeg / gif / webp</li>
+    </ul>
+  </div>
+
+  <h3>多模态消息示例</h3>
+  <pre><span class="code-lang">json</span>{
+  "role": "user",
+  "content": [
+    { "type": "text", "text": "请描述这张图片的内容" },
+    {
+      "type": "image_url",
+      "image_url": {
+        "url": "https://example.com/image.jpg",
+        "detail": "auto"
+      }
+    }
+  ]
+}</pre>
+
+  <div class="warn-box">
+    <strong>⚠️ 多模态限制：</strong>
+    <ul style="padding-left:20px;margin:8px 0 0;">
+      <li>仅 <code>mimo-v2.5-pro</code> / <code>mimo-v2.5</code> 文本模型支持图像输入</li>
+      <li>TTS 系列模型（<code>mimo-v2.5-tts*</code>）不支持多模态图像输入</li>
+      <li>历史消息中的图片 content 数组应原样保留，不建议在多轮对话中删除</li>
+      <li>base64 data URI 会显著增加请求体大小，建议对大图先压缩或使用公网 URL</li>
+    </ul>
+  </div>
+</section>
+
 <!-- ===== RESPONSE ===== -->
 <section id="response">
   <h2>📤 响应对象（非流式输出）</h2>
@@ -746,6 +797,100 @@ audio_bytes = base64.b64decode(audio_data)
 with output("output.wav", "wb") as f:
     f.write(audio_bytes)
 print("音频已保存到 output.wav")</pre>
+
+  <h3>多模态图像输入</h3>
+  <div class="tabs">
+    <button class="tab-btn active" onclick="switchTab(event,'tab-mm-curl')">curl</button>
+    <button class="tab-btn" onclick="switchTab(event,'tab-mm-python')">Python (OpenAI SDK)</button>
+    <button class="tab-btn" onclick="switchTab(event,'tab-mm-base64')">Python (Base64 本地图片)</button>
+  </div>
+
+  <div id="tab-mm-curl" class="tab-content active">
+<pre><span class="code-lang">bash</span>curl --location --request POST 'https://api.xiaomimimo.com/v1/chat/completions' \
+--header "api-key: $MIMO_API_KEY" \
+--header "Content-Type: application/json" \
+--data-raw '{
+    "model": "mimo-v2.5-pro",
+    "messages": [
+        {
+            "role": "user",
+            "content": [
+                {"type": "text", "text": "请描述这张图片的内容"},
+                {
+                    "type": "image_url",
+                    "image_url": {
+                        "url": "https://example.com/image.jpg",
+                        "detail": "auto"
+                    }
+                }
+            ]
+        }
+    ],
+    "max_completion_tokens": 1024,
+    "thinking": {"type": "disabled"}
+}'</pre>
+  </div>
+
+  <div id="tab-mm-python" class="tab-content">
+<pre><span class="code-lang">python</span>from openai import OpenAI
+
+client = OpenAI(api_key="$MIMO_API_KEY", base_url="https://api.xiaomimimo.com/v1")
+
+response = client.chat.completions.create(
+    model="mimo-v2.5-pro",
+    messages=[
+        {
+            "role": "user",
+            "content": [
+                {"type": "text", "text": "请描述这张图片的内容"},
+                {
+                    "type": "image_url",
+                    "image_url": {
+                        "url": "https://example.com/image.jpg",
+                        "detail": "auto"
+                    }
+                }
+            ]
+        }
+    ],
+    max_completion_tokens=1024,
+    thinking={"type": "disabled"}
+)
+
+print(response.choices[0].message.content)
+print(f"图像 token: {response.usage.prompt_tokens_details.image_tokens}")</pre>
+  </div>
+
+  <div id="tab-mm-base64" class="tab-content">
+<pre><span class="code-lang">python</span>import base64
+from openai import OpenAI
+
+client = OpenAI(api_key="$MIMO_API_KEY", base_url="https://api.xiaomimimo.com/v1")
+
+# 读取本地图片并转为 base64 data URI
+with open("local_image.jpg", "rb") as f:
+    image_data = base64.b64encode(f.read()).decode("utf-8")
+image_data_uri = f"data:image/jpeg;base64,{image_data}"
+
+response = client.chat.completions.create(
+    model="mimo-v2.5-pro",
+    messages=[
+        {
+            "role": "user",
+            "content": [
+                {"type": "text", "text": "这张图里有什么？"},
+                {
+                    "type": "image_url",
+                    "image_url": {"url": image_data_uri}
+                }
+            ]
+        }
+    ],
+    max_completion_tokens=1024
+)
+
+print(response.choices[0].message.content)</pre>
+  </div>
 </section>
 
 <!-- ===== ERRORS ===== -->