ima.copilot Sync

强烈建议在桌面端使用。 桌面端独有的「下载增强」功能可显著提升微信公众号内容和防盗链图片的下载成功率。

中文 | English

Install/安装

https://community.obsidian.md/plugins/ima-copilot-sync

中文

将腾讯 IMA 个人笔记和知识库同步到 Obsidian vault 的插件。

典型使用场景

微信公众号内容转发到ima后，自动同步公众号内容到Obsidian

功能特性

⚠ 单向同步：本插件仅支持 IMA → Obsidian 单向同步。在 Obsidian 中对笔记做的任何修改不会同步回 IMA，每次同步会用 IMA 服务端内容覆盖本地文件。

• 个人笔记同步：将 IMA 笔记本中的所有笔记自动下载到 Obsidian

• 图片和文件本地化：自动下载笔记中的图片并保存到本地附件目录

• 知识库同步：同步知识库中的所有类型条目

  • 笔记：完整同步内容并转为 Markdown
  • 网页：提取正文内容并转为 Markdown（但部分需要登录或者有反爬措施的无法绕过)
  • 微信文章：桌面端开启下载增强后通过无头浏览器渲染完整提取（含图片），支持所有微信文章格式（标准图文、图片分享页/小绿书等）；移动端回退到 meta 标签文本提取（仅文字）
  • 文件（PDF、Word、PPT、Excel 等）：个人/共享知识库可下载到本地；订阅/公共知识库仅同步 AI 摘要

• 增量同步：仅同步上次同步后有修改的笔记，减少不必要的请求

• 自动定时同步：按设定间隔自动在后台同步

• 安全凭证存储：凭证存储于 Obsidian 钥匙串（系统 Keychain），不以明文保存在配置文件中

• 知识库删除同步：支持删除/保留/标记三种模式处理 IMA 端已删除的条目

配置步骤

1. 获取 IMA OpenAPI 凭证

访问 https://ima.qq.com/agent-interface，登录后复制页面上的 Client ID 和 API Key。

2. 填入凭证

打开 Obsidian 设置 → ima.copilot Sync，在设置页面：

• 直接将复制的凭证文本粘贴到剪贴板，点击「粘贴并解析凭证」按钮自动填入
• 或手动在 Client ID 和 API Key 输入框中分别填写

凭证将安全存储于 Obsidian 钥匙串中，不会以明文保存在配置文件里。

点击「测试」按钮验证连接是否正常。

3. 选择同步内容

4. 附件设置

已知限制

• 订阅/公共知识库内容受限(个人知识库不受此限制)：IMA API 对订阅知识库有访问限制，各类型内容的同步能力如下：
  • 笔记：仅同步约 300 字预览，无法获取完整内容
  • 微信文章：桌面端可完整提取（需开启下载增强），支持所有微信文章格式；移动端回退到 meta 标签文本提取（仅文字，无图片）
  • 文件（PDF/Word 等）：仅同步 AI 摘要，无法下载原件
  • 网页：可抓取完整正文
• 知识库中部分条目如果 IMA API 未返回可访问的 URL，将仅同步标题（显示为占位符）

开发构建

# 安装依赖
npm install

# 开发模式（文件监听）
npm run dev

# 生产构建
npm run build

English

An Obsidian plugin to sync notes from Tencent IMA personal notebook and knowledge base into your Obsidian vault.

Typical Use Cases

Forward WeChat official account articles to IMA, then auto-sync the content to Obsidian.

Features

⚠ One-way sync only: This plugin syncs IMA → Obsidian only. Any edits made in Obsidian will not be synced back to IMA — each sync overwrites local files with the content from IMA.

• Personal notes sync: Automatically downloads all notes from your IMA notebook

• Image and file localization: Downloads inline images and file attachments to a local folder

• Knowledge base sync: Syncs all item types from your IMA knowledge base

  • Notes: Full content converted to Markdown
  • Webpages: Extracts main content and converts to Markdown (but sites requiring login or with anti-scraping measures cannot be bypassed)
  • WeChat articles: Full extraction via headless browser on desktop (with download enhancement enabled), supports all WeChat article formats; falls back to meta tag text extraction on mobile (text only, no images)
  • Files (PDF, Word, PPT, Excel, etc.): Personal/shared KBs can download locally; subscribed/public KBs get AI summary only

• Incremental sync: Only fetches notes modified since the last sync

• Auto periodic sync: Runs silently in the background on a configurable interval

• Secure credential storage: Credentials stored in Obsidian keychain (system Keychain), never saved in plaintext

• Knowledge base delete sync: Three modes (delete/keep/mark) for handling items deleted from IMA

Setup

1. Get IMA OpenAPI credentials

Visit https://ima.qq.com/agent-interface, log in, and copy your Client ID and API Key.

2. Enter credentials

Open Obsidian Settings → ima.copilot Sync:

• Paste the copied credential text to your clipboard and click 「粘贴并解析凭证」 to auto-fill
• Or enter the Client ID and API Key manually in their respective fields

Credentials are securely stored in the Obsidian keychain and never saved in plaintext.

Click 「测试」 to verify the connection.

3. Choose what to sync

4. Attachment settings

Known Limitations

• Subscribed/Public knowledge base content is limited (Personal knowledge base is not affected by this limitation): IMA API restricts access to subscribed knowledge bases. Sync capabilities by content type:
  • Notes: Only ~300 character preview, full content not available
  • WeChat articles: Full extraction on desktop (with download enhancement enabled), supports all WeChat article formats; falls back to meta tag text extraction on mobile (text only, no images)
  • Files (PDF/Word etc.): Only AI summary available, original files cannot be downloaded
  • Webpages: Full content can be fetched
• Some knowledge base items may only sync the title (shown as a placeholder) if the IMA API does not return an accessible URL

Development

# Install dependencies
npm install

# Development mode (watch)
npm run dev

# Production build
npm run build

License

MIT

This project includes defuddle (MIT) by Steph Ango (@kepano), used for HTML-to-Markdown conversion.