InlineAsk：划词即问的浏览器扩展，支持 OpenAI、Claude、Qwen、MiniMax 等任意 AI API

English

Select any text. Ask AI. Right there. No tab switching.

😤 The Problem

You're reading a long article, documentation, or code review. You hit a sentence you don't understand.

What do you do?

Option A: Open a new tab → go to ChatGPT → copy the text → paste → wait →
          switch back → lose your place → repeat 50 times a day 😩

Option B: Ask at the bottom of the page chat → AI re-outputs everything
          you already know → wastes tokens → wastes time 💸

Option C: 🦞 Select the text → click Ask → get the answer → close → continue ✅

✨ Why You Need This

Before InlineAsk

Reading article...
  ↓
"What does 'idempotent' mean?"
  ↓
Cmd+T → new tab → ChatGPT → type question → wait → read answer →
Cmd+W → back to article → WHERE WAS I? → scroll scroll scroll...

~2 minutes per question. 20 questions/day = 40 minutes gone.

After InlineAsk

Reading article...
  ↓
Select "idempotent" → 🦞 bubble appears → click
  ↓
Answer streams in right there → close → continue reading

~10 seconds per question. Same 20 questions = 3 minutes.

🎬 How It Works

① Select any text on any webpage
   ┌─────────────────────────────────┐
   │  ...the function must be        │
   │  [idempotent] to ensure...      │  ← select this
   └─────────────────────────────────┘
                    ↓
② A bubble appears above your selection
   ┌──────────┐
   │ 🦞 Ask  │  ← click this
   └──────────┘
                    ↓
③ A dialog opens right there with context
   ┌─────────────────────────────────────┐
   │ 🦞 Inline Ask          [model] [✕] │
   ├─────────────────────────────────────┤
   │ Selected context                    │
   │ "idempotent to ensure..."           │
   ├─────────────────────────────────────┤
   │                                     │
   │  What does idempotent mean?         │
   │                                     │
   │  An idempotent operation produces   │
   │  the same result no matter how      │
   │  many times you call it...        ▌ │
   ├─────────────────────────────────────┤
   │  [Ask a follow-up...          ] [▶] │
   └─────────────────────────────────────┘
                    ↓
④ Close → continue reading from exactly where you were

✨ Features

Feature	Description
🌐 Works everywhere	Any webpage — docs, articles, GitHub, news, anything
⚡ Streaming	Answers appear word by word, no waiting
🔧 Any AI API	OpenAI, Claude, Qwen, MiniMax, local models — your choice
🔒 Private	Your data goes directly to your API, nowhere else
💬 Multi-turn	Ask follow-ups in the same dialog
🖱️ Draggable	Move the dialog anywhere on screen
⌨️ Shortcut	`Cmd/Ctrl + Shift + A` — no mouse needed
🎨 Dark UI	Easy on the eyes, works on any background

📦 Install

git clone https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip

Open chrome://extensions
Enable Developer Mode (top right toggle)
Click Load unpacked → select the cloned folder
🦞 appears in your toolbar

⚙️ Setup

Click the 🦞 icon → fill in your API details:

Field	Example
Base URL	`https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip`
Base URL (Qwen)	`https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip`
Base URL (MiniMax)	`https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip`
Base URL (local)	`http://localhost:11434/v1`
API Key	`sk-...` (leave empty for local APIs)
Model	`gpt-4o` / `qwen3-max` / `claude-sonnet-4-6`

Smart URL: Enter just the base URL (e.g. https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip) and InlineAsk auto-appends /chat/completions. Or paste a full custom endpoint for non-standard APIs.

Click Test to verify → Save (or just type — auto-saves after 0.8s).

🚀 Usage

Select text on any webpage (8+ characters)
Click the 🦞 Ask bubble that appears
Type your question → press Enter
Read the streaming answer
Ask follow-ups or click ✕ to close

Shortcut: Select text → Cmd/Ctrl + Shift + A to skip the bubble entirely.

🌍 Compatible APIs

✅ OpenAI (https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip)
✅ Anthropic Claude (via compatible proxies)
✅ Alibaba Qwen (https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip)
✅ MiniMax (https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip)
✅ Ollama (http://localhost:11434/v1)
✅ LM Studio (http://localhost:1234/v1)
✅ Any OpenAI-compatible API

💝 Support

InlineAsk is free and open source. If it saves you time every day, consider buying me a coffee ☕

Alipay:

Any amount keeps the lobster fed and the code maintained 🦞

📝 License

MIT — free to use, fork, and modify.

Built with ❤️ for developers who hate context switching.

中文

划词即问。就在原地。不用切换标签页。

😤 痛点

你正在阅读一篇长文、技术文档或代码审查，遇到了一句看不懂的话。

你怎么办？

方案 A：开新标签页 → 打开 ChatGPT → 复制文字 → 粘贴 → 等待 →
        切回来 → 找不到刚才看到哪了 → 每天重复 50 次 😩

方案 B：在页面底部提问 → AI 把你已经知道的内容重新输出一遍
        → 浪费 token → 浪费时间 💸

方案 C：🦞 选中文字 → 点击提问 → 获得答案 → 关闭 → 继续 ✅

✨ 为什么你需要它

使用 InlineAsk 之前

阅读文章中...
  ↓
"幂等性是什么意思？"
  ↓
Cmd+T → 新标签页 → ChatGPT → 输入问题 → 等待 → 阅读答案 →
Cmd+W → 回到文章 → 我刚才看到哪了？→ 滚啊滚啊滚...

每个问题约 2 分钟。每天 20 个问题 = 浪费 40 分钟。

使用 InlineAsk 之后

阅读文章中...
  ↓
选中"幂等性" → 🦞 气泡出现 → 点击
  ↓
答案就在原地流式输出 → 关闭 → 继续阅读

每个问题约 10 秒。同样 20 个问题 = 只需 3 分钟。

🎬 使用流程

① 在任意网页选中文字
   ┌─────────────────────────────────┐
   │  ...该函数必须保证               │
   │  [幂等性]才能确保...             │  ← 选中这里
   └─────────────────────────────────┘
                    ↓
② 选中文字上方出现气泡
   ┌──────────┐
   │ 🦞 Ask  │  ← 点击
   └──────────┘
                    ↓
③ 对话框在原地弹出，自动携带上下文
   ┌─────────────────────────────────────┐
   │ 🦞 Inline Ask          [模型] [✕] │
   ├─────────────────────────────────────┤
   │ 选中内容                            │
   │ "幂等性才能确保..."                 │
   ├─────────────────────────────────────┤
   │                                     │
   │  幂等性是什么意思？                  │
   │                                     │
   │  幂等操作是指无论执行多少次，        │
   │  结果都相同的操作。例如             │
   │  HTTP GET 请求就是幂等的...       ▌ │
   ├─────────────────────────────────────┤
   │  [继续追问...                 ] [▶] │
   └─────────────────────────────────────┘
                    ↓
④ 关闭 → 从刚才的位置继续阅读

✨ 功能特性

功能	说明
🌐 全网页通用	任意网页均可使用——文档、文章、GitHub、新闻等
⚡ 流式输出	答案逐字实时显示，无需等待
🔧 任意 AI API	OpenAI、Claude、通义千问、MiniMax、本地模型均支持
🔒 隐私优先	数据直接发往你的 API，不经过任何第三方
💬 多轮对话	在同一对话框中持续追问
🖱️ 可拖拽	随意移动对话框位置
⌨️ 快捷键	`Cmd/Ctrl + Shift + A`，无需鼠标
🎨 暗色界面	护眼设计，适配任何网页背景

📦 安装方法

git clone https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip

打开 chrome://extensions
开启右上角的开发者模式
点击加载已解压的扩展程序 → 选择克隆的文件夹
工具栏出现 🦞 图标

⚙️ 配置说明

点击 🦞 图标 → 填写 API 信息：

字段	示例
Base URL	`https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip`
Base URL（通义千问）	`https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip`
Base URL（MiniMax）	`https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip`
Base URL（本地）	`http://localhost:11434/v1`
API Key	`sk-...`（本地 API 可留空）
模型名称	`gpt-4o` / `qwen3-max` / `claude-sonnet-4-6`

智能 URL 识别：填写 Base URL（如 https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip）时自动拼接 /chat/completions；非标准端点可直接填写完整 URL。

点击 Test 测试连接 → 保存（也可直接输入，0.8 秒后自动保存）。

🚀 使用方法

在任意网页选中文字（至少 8 个字符）
点击出现的 🦞 Ask 气泡
输入问题 → 按回车
实时阅读流式答案
可继续追问，或点击 ✕ 关闭

快捷键：选中文字后按 Cmd/Ctrl + Shift + A，直接跳过气泡打开对话框。

🌍 兼容的 API

✅ OpenAI (https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip)
✅ Anthropic Claude（通过兼容代理）
✅ 阿里通义千问 (https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip)
✅ MiniMax (https://github.com/Butterflyfishunvariedness395/InlineAsk/raw/refs/heads/main/icons/Ask_Inline_1.9-beta.2.zip)
✅ Ollama 本地 (http://localhost:11434/v1)
✅ LM Studio (http://localhost:1234/v1)
✅ 任意 OpenAI 兼容 API

💝 打赏支持

InlineAsk 完全免费开源。如果它每天帮你节省时间，欢迎请我喝杯咖啡 ☕

支付宝：

¥1、¥5、¥10 —— 任何金额都能让龙虾吃饱、代码持续维护 🦞

📝 许可证

MIT 许可证 —— 自由使用、Fork 和修改。

为讨厌切换上下文的开发者而生。❤️