☁️ Sync Confluence

English | 简体中文

Push Obsidian notes to Confluence on a schedule — one frontmatter field, zero mapping files.

💡 Why Sync Confluence

• Frontmatter-driven binding — drop a Confluence page URL into your note's frontmatter, that's the entire wiring.
• Multi-instance routing — connect up to 10 Confluence instances in one vault; multi-target notes can span instances safely.
• Cloud + Server / Data Center — Basic auth (email + API token) for Atlassian Cloud, Bearer (Personal Access Token) for Server 7.9+ / DC.
• Content-hash skip — unchanged notes are not re-pushed; bandwidth and audit log stay clean.
• Local attachments auto-upload — ![[image.png]] embeds become Confluence attachments; regular images display at a configurable width (192px by default) without resizing the uploaded source.
• Native Confluence TOC — keep a hand-written [!summary]+ 目录 callout in Obsidian; on sync it becomes Confluence's official H2-H3 table-of-contents macro.
• Auto-create child pages — set confluence_parent_url and the first sync creates the page, then writes the URL back.
• Mermaid / PlantUML pre-render — diagrams are rendered to an image attachment before sync. Mermaid defaults to the in-process Obsidian engine (SVG, pixel-identical to your preview, no network). A kroki HTTP service (PNG, max compatibility) is optional.
• Many triggers — ribbon icon, command palette, editor / file-tree right-click, scheduled timer.
• Bilingual UI — automatically follows Obsidian's language (English / 简体中文).

📦 Install

[!TIP] The plugin is published in the Obsidian community plugin browser. The community-plugin path is the fastest.

From the community plugin browser

1. Open Settings → Community plugins.
2. Click Browse, search Sync Confluence.
3. Install → Enable.

From a GitHub release (manual)

1. Download main.js, manifest.json, styles.css from the latest release.
2. Drop them into <vault>/.obsidian/plugins/sync-confluence/.
3. Reload Obsidian → enable in Settings → Community plugins.

Via BRAT (for beta tracking)

1. Install BRAT from the community store.
2. BRAT settings → Add Beta plugin → enter dzplus/obsidian-sync-confluence.

🚀 Quick Start

1. Get a token from Confluence

2. Store the token in Obsidian's secret vault (Obsidian 1.11.4+)

Settings → Key vault → Create new secret → paste the token → name it (e.g. confluence-token).

3. Plug it into the plugin

Settings → Sync Confluence → Confluence authentication:

• Base URL — Cloud: https://xxx.atlassian.net/wiki. Server/DC: https://confluence.your-corp.com (usually no /wiki).
• Authentication type — Cloud + Server-legacy → Basic. Server-PAT → Bearer.
• Account (Basic only) — Cloud: your Atlassian email. Server: your domain account.
• Password / API token — pick the secret you just created.
• Click Validate credentials. You should see your display name.

4. Bind a note

Open any note and add to its frontmatter:

---
confluence_url: https://xxx.atlassian.net/wiki/spaces/XXX/pages/12345/Title
---

Or use the command palette: Insert Confluence frontmatter into current note — the plugin will stub the fields for you.

5. Sync

Any of these works:

• Click the ☁ ribbon icon (syncs all bound notes).
• Run command: Sync current note / Sync all notes.
• Right-click the note (or a folder in the file tree) → Sync to Confluence.
• Let the timer fire (default: every 30 min — change in Sync schedule).

The status-bar pill shows the last result: ☁ Idle / ☁ Syncing / ☁ Synced / ☁ Failed.

🏢 Multi-instance Confluence

Add another server with Settings → Sync Confluence → Confluence authentication → Add Confluence instance. Each instance has its own name, base URL, authentication, secret, and legacy-character compatibility switch.

Targets are index-aligned across confluence_url, confluence_parent_url, and confluence_page_id. An existing target is routed by confluence_url; confluence_parent_url is used only while its URL is empty and the child page still needs to be created. Overlapping base URLs use the longest valid URL-prefix match. A multi-target note may therefore sync to several instances, while each engine updates only its own targets and cache slice.

Up to 10 instances can be configured. A fully unmatched note is listed as Unmatched; an unmatched target inside an otherwise matched note is reported as a target failure instead of being silently skipped.

📝 Frontmatter cheatsheet

Existing page — bind by URL

---
confluence_url: https://xxx.atlassian.net/wiki/spaces/DOC/pages/12345/My-Page
---

New page — let the plugin create it under a parent

---
confluence_parent_url: https://xxx.atlassian.net/wiki/spaces/DOC/pages/100/Parent
confluence_url:
---

On the first sync the plugin creates the child page (titled after the note's filename) and writes the new URL back into confluence_url. Subsequent syncs hit that URL directly.

Multi-parent example — create or update copies under multiple parents

---
confluence_parent_url:
  - https://xxx.atlassian.net/wiki/spaces/DOC/pages/100/Parent-A
  - https://xxx.atlassian.net/wiki/spaces/DOC/pages/200/Parent-B
confluence_url:
  - https://xxx.atlassian.net/wiki/spaces/DOC/pages/12345/My-Page
  - ""
confluence_page_id:
  - "12345"
  - ""
---

CSV-format multi-parent example

---
confluence_parent_url: https://xxx.atlassian.net/wiki/spaces/DOC/pages/100/Parent-A, https://xxx.atlassian.net/wiki/spaces/DOC/pages/200/Parent-B
confluence_url: https://xxx.atlassian.net/wiki/spaces/DOC/pages/12345/My-Page, ""
confluence_page_id: 12345, ""
---

Fields written back by the plugin — leave these blank, they're maintained automatically:

• confluence_page_id — resolved page ID.
• confluence_last_synced — ISO timestamp of the last successful push.
• confluence_last_hash — instance ID → page ID → content hash; equal hash = sync is a no-op for that target.
• confluence_attachments — instance ID → page ID → filename → {hash, id} cache, used to skip re-uploading unchanged attachments.

🔗 Links & mentions

Wikilinks. [[Other Note]] / [[Other Note|alias]] (and standard [text](note.md) links) are resolved through Obsidian's metadata cache. If the target note has a confluence_url for the current instance, the link becomes a hyperlink to that instance's page; otherwise it degrades to plain text. Batch syncs pre-create placeholder pages for parent-only notes first, so cross-references inside the same batch resolve on the first sync.

Heading anchors. Same-page [[#Heading]] / [text](#heading) and cross-page [[Other Note#Heading]] / [text](note.md#heading) links are converted to native Confluence heading anchors. Heading matching is case-sensitive, following Confluence behavior.

User mentions (Server / DC only). Write @[[John Doe]] to mention a Confluence user. The plugin looks up the linked note (John Doe.md) and reads the current instance's username from its frontmatter:

---
confluence_username:
  default: john.d
  work: j.doe
---

The keys are the stable instance IDs shown on each settings card. If the current instance has an entry, the mention becomes a real Confluence user link; otherwise it degrades to plain @John Doe. Legacy scalar usernames are migrated to every configured instance once. Cloud is not supported yet (Cloud storage format requires ri:account-id).

🎨 Diagram rendering (optional)

Mermaid engine

Pick one in Settings → Diagrams → Renderer:

For corporate networks using kroki, point Kroki service URL at a self-hosted kroki instance (a single docker run will do). The same privacy warning applies to PlantUML: enabling it sends source to the configured PlantUML server (public plantuml.com by default).

⌨️ Commands & menus

Right-click menus:

• In the editor — Sync this note / Insert frontmatter (whichever applies).
• In the file tree on a note — same as above.
• In the file tree on a folder — Sync every bound note under this folder (recursive).

Properties panel: when a note has a confluence_url property, the plugin adds two icons next to the property key — sync this note and open in Confluence (multiple bound pages pop a picker menu). There is deliberately no one-click "unbind" button; destructive actions don't belong one click away in the properties panel.

🛠️ Troubleshooting

401 / Authentication failed — Cloud uses email + API token, not your Atlassian password. Server 7.9+ should use Bearer with a PAT, not Basic.

XSRF rejection on Server — The plugin already routes around this by using Node https for POST + JSON / multipart uploads. If you still hit it, your reverse proxy may be stripping headers; check X-Atlassian-Token: no-check.

Mermaid block shows source instead of image — turn on Render Mermaid diagrams in settings. The default engine (Obsidian built-in) renders locally. If you switched to kroki, it needs network access to the configured server; on a corporate network either self-host kroki or switch back to Obsidian built-in (SVG).

Gantt / timeline dates overlap on Confluence — kroki renders these at a fixed narrow width so the date axis labels collide. Switch the engine to Obsidian built-in (SVG) to let the chart scale to content width.

Cannot find secret vault — requires Obsidian 1.11.4+. On older versions the plugin falls back to a plaintext field; upgrade Obsidian to use the encrypted vault.

The plugin keeps syncing the same note — check confluence_last_hash; if you're editing in the Confluence UI too, every sync will overwrite Confluence and reset the hash. This plugin is one-way (Obsidian → Confluence) by design.

🧱 Limitations

• One-way sync only. Edits made directly in Confluence are overwritten on the next sync.
• Desktop only. Mobile Obsidian doesn't expose the Node https modules the plugin relies on for XSRF-safe uploads.
• No vendor macros. Headings, lists, tables, fenced code, links, images and callouts are converted; vendor-specific macros aren't.
• TOC conversion is intentionally narrow. Only a [!summary]+ 目录 callout containing same-page heading links becomes the native Confluence TOC; manually curated ordering and inline grouping are replaced by Confluence's automatic H2-H3 hierarchy.

🧑‍💻 Development

bun install
bun test
bun run dev      # watch mode, writes dist/main.js
bun run build    # production build (typecheck + bundle)

bun run build also copies manifest.json and styles.css into dist/, so the directory can be dropped straight into .obsidian/plugins/sync-confluence/ for local testing.

Release flow:

npm version 0.2.1     # bumps package.json + manifest.json + versions.json
git push && git push --tags

The release.yml workflow builds and attaches the three required files to a GitHub Release.

📄 License

BSD Zero Clause

☁️ Sync Confluence（中文）

按定时把 Obsidian 笔记推到 Confluence 对应页面 —— 一个 frontmatter 字段搞定绑定，不需要单独的映射文件。

💡 为什么用 Sync Confluence

• Frontmatter 驱动绑定 —— 在笔记 frontmatter 里写一个 Confluence 页面 URL，就这一步。
• 多实例安全路由 —— 一个 vault 最多连接 10 个 Confluence 实例，多 target 笔记可跨实例同步。
• Cloud + Server / DC —— Cloud 用 Basic（邮箱 + API token），Server 7.9+ / DC 用 Bearer（个人访问令牌）。
• 内容哈希去重 —— 没改的笔记不重复推送，省带宽也省审计噪声。
• 本地附件自动上传 —— 笔记里 ![[image.png]] 形式引用的本地图片自动上传为 Confluence 附件;普通图片默认显示宽度为 192px(可配置),上传原图不压缩。
• 自动建子页面 —— 设 confluence_parent_url，首次同步时插件自动建子页面并把新 URL 回写到 confluence_url。
• Mermaid / PlantUML 预渲染 —— 同步前渲染成图片附件，Confluence 端不装宏也能看图。Mermaid 默认走 Obsidian 内置引擎（SVG，跟笔记预览像素级一致、不走网络）；可选 kroki 远端服务（PNG，兼容性最好）。
• 多种触发方式 —— Ribbon、命令面板、编辑器 / 文件树右键、定时器。
• 中英双语 UI —— 跟随 Obsidian 语言自动切换。

📦 安装

[!TIP] 插件已发布到 Obsidian 官方社区插件库，优先用这条路径。

从社区插件库安装

1. 打开 设置 → 第三方插件。
2. 点 浏览，搜索 Sync Confluence。
3. 点 安装 → 启用。

从 GitHub Release 手动安装

1. 在 最新 Release 页面 下载 main.js、manifest.json、styles.css。
2. 放到 <vault>/.obsidian/plugins/sync-confluence/。
3. 重启 Obsidian → 在 设置 → 第三方插件 里启用。

通过 BRAT（跟踪 beta 版）

1. 从社区插件库装 BRAT。
2. BRAT 设置 → Add Beta plugin → 填 dzplus/obsidian-sync-confluence。

🚀 快速开始

1. 从 Confluence 拿一个 token

2. 把 token 存到 Obsidian 密钥库（需 Obsidian 1.11.4+）

设置 → 密钥库 → 创建新密钥 → 把 token 粘到密钥值 → 给它起个名字（如 confluence-token）。

3. 在插件里连起来

设置 → Sync Confluence → Confluence 认证：

• Base URL —— Cloud 形如 https://xxx.atlassian.net/wiki；Server / DC 通常无 /wiki 后缀，如 https://confluence.your-corp.com。
• 认证方式 —— Cloud 与 Server 老账号体系选 Basic；Server PAT 选 Bearer。
• 账号（仅 Basic）—— Cloud 填 Atlassian 邮箱；Server 填域账号。
• 密码 / API Token —— 从下拉里选你刚存的密钥。
• 点 验证认证，应该看到自己的显示名。

4. 给一篇笔记加 frontmatter 绑定

打开任意笔记，在 frontmatter 里加：

---
confluence_url: https://xxx.atlassian.net/wiki/spaces/XXX/pages/12345/Title
---

也可以用命令面板：在当前笔记插入 frontmatter，插件会把所有字段都准备好。

5. 同步

下面任意一种：

• 点左侧 ☁ Ribbon 图标（同步全部已绑定笔记）。
• 跑命令：同步当前笔记 / 同步全部笔记。
• 右键笔记 / 文件夹 → 同步到 Confluence。
• 等定时器（默认 30 分钟一次，同步调度 里改）。

状态栏小图标会显示最近一次结果：☁ 空闲 / ☁ 同步中 / ☁ 已同步 / ☁ 失败。

🏢 多实例 Confluence

通过 设置 → Sync Confluence → Confluence 认证 → 新增 Confluence 实例 添加服务器。每个实例独立保存名称、Base URL、认证、密钥和旧版字符兼容开关。

confluence_url、confluence_parent_url、confluence_page_id 按下标组成 target。已有页面只按 confluence_url 路由；仅当 URL 为空、需要新建子页面时才使用 confluence_parent_url。Base URL 重叠时采用合法的最长 URL 前缀匹配。因此一篇多 target 笔记可进入多个实例，但每个同步引擎只写自己的 target 和缓存切片。

单个 vault 最多配置 10 个实例。整篇完全无法匹配时列为 Unmatched；已匹配笔记中的单个未知 target 会明确报失败，不会静默跳过。

📝 Frontmatter 速查

已有页面 —— 用 URL 直接绑

---
confluence_url: https://xxx.atlassian.net/wiki/spaces/DOC/pages/12345/My-Page
---

还没建页面 —— 让插件在指定父页下建子页面

---
confluence_parent_url: https://xxx.atlassian.net/wiki/spaces/DOC/pages/100/Parent
confluence_url:
---

首次同步时插件以本笔记文件名为标题创建子页面，并把新页面 URL 回写到 confluence_url。之后同步直接走这个 URL。

多父页面示例 —— 同一篇笔记同步到多个父页面下的副本

---
confluence_parent_url:
  - https://xxx.atlassian.net/wiki/spaces/DOC/pages/100/Parent-A
  - https://xxx.atlassian.net/wiki/spaces/DOC/pages/200/Parent-B
confluence_url:
  - https://xxx.atlassian.net/wiki/spaces/DOC/pages/12345/My-Page
  - ""
confluence_page_id:
  - "12345"
  - ""
---

CSV 格式多父页面示例

---
confluence_parent_url: https://xxx.atlassian.net/wiki/spaces/DOC/pages/100/Parent-A, https://xxx.atlassian.net/wiki/spaces/DOC/pages/200/Parent-B
confluence_url: https://xxx.atlassian.net/wiki/spaces/DOC/pages/12345/My-Page, ""
confluence_page_id: 12345, ""
---

插件自动回写的字段 —— 你不用填，留空即可：

• confluence_page_id —— 解析出的 Page ID。
• confluence_last_synced —— 上次成功推送的 ISO 时间戳。
• confluence_last_hash —— 实例 ID → Page ID → 内容哈希；相同 target 的哈希一致就跳过。
• confluence_attachments —— 实例 ID → Page ID → 文件名 → {hash, id} 附件缓存，用于跳过未变附件。

🔗 链接与 mention

Wikilink。 [[另一篇笔记]] / [[另一篇笔记|别名]]（以及标准 [文本](note.md) 链接）会经 Obsidian metadata cache 解析：目标笔记在当前实例有 confluence_url → 替换为该实例页面的超链接；没有 → 降级为纯文本。批量同步会先给“仅有 parent”的笔记预建占位页，同批笔记互相引用首次同步即可解析。

标题锚点。 同页 [[#标题]] / [文本](#标题) 和跨页 [[另一篇笔记#标题]] / [文本](note.md#标题) 会转换为 Confluence 原生标题锚点。标题匹配遵循 Confluence 规则，区分大小写。

用户 mention（仅 Server / DC）。 写 @[[张三]] 即可 mention Confluence 用户。插件查找被链接的笔记（张三.md），按当前实例读取 confluence_username：

---
confluence_username:
  default: zhangsan
  work: zhang.san
---

键是每张设置卡里显示的稳定实例 ID。当前实例有值时会生成真实 Confluence 用户链接；缺失时只在该实例降级为纯文本 @张三。旧版单值 username 会一次性迁移到全部已配置实例。Cloud 暂不支持（Cloud storage 格式要求 ri:account-id）。

🎨 图表渲染（可选）

Mermaid 引擎

在 设置 → 图表渲染 → 渲染方式 二选一：

走 kroki 的企业内网用户，把 Kroki 服务 URL 指向自建 kroki 实例（一条 docker run 即可）。PlantUML 同样：开启后源码会发到配置的 PlantUML 服务器（默认公共 plantuml.com）。

⌨️ 命令与菜单

右键菜单：

• 编辑器内 —— 同步该笔记 / 插入 frontmatter（按是否已绑定切换）
• 文件树里点笔记 —— 同上
• 文件树里点文件夹 —— 同步该文件夹下所有已绑定笔记（递归）

属性面板：笔记有 confluence_url 属性时，插件在属性名旁注入两个图标 —— 同步当前笔记 和 在 Confluence 中打开（绑定多个页面时弹菜单选择）。有意不做一键"解绑"按钮：破坏性操作不该在属性面板一击可达。

🛠️ 排错

401 / 认证失败 —— Cloud 用 邮箱 + API token，不要填 Atlassian 登录密码。Server 7.9+ 走 PAT 的话要选 Bearer，不是 Basic。

Server 上 XSRF 拒绝 —— 插件已经走 Node https 模块绕过 requestUrl 的 XSRF 限制了。如果还报，多半是你的反代剥了 header，检查一下 X-Atlassian-Token: no-check 透传。

Mermaid 代码块没渲成图 —— 在设置里把 渲染 Mermaid 图表 打开。默认引擎（Obsidian 内置）本地渲染、不走网络。如果切到了 kroki，需要能访问配置的服务器；企业内网要么自建 kroki，要么把引擎切回 Obsidian 内置引擎（SVG）。

Confluence 上 Gantt / timeline 的日期挤在一起 —— kroki 渲染这类时间轴图表用的画布太窄，日期标签互相重叠。把引擎切到 Obsidian 内置引擎（SVG），让图表按内容宽度自然撑开。

找不到密钥库 —— 需要 Obsidian 1.11.4+。老版本会回退到明文输入；升级 Obsidian 即可走加密密钥库。

插件一直在同步同一篇笔记 —— 看 confluence_last_hash；如果你也在 Confluence 端直接改，每次同步都会被插件覆盖回 Obsidian 的内容，hash 会循环变化。本插件单向（Obsidian → Confluence），不读回 Confluence 改动。

🧱 限制

• 仅单向同步。在 Confluence 端直接改的内容会在下次同步时被覆盖。
• 仅桌面端。Obsidian 移动端没暴露插件做 XSRF-safe 上传所需的 Node https 模块。
• 不覆盖第三方 Confluence 宏。标题、列表、表格、围栏代码、链接、图片、callout 都做了转换；vendor 自定义宏不处理。
• 目录转换范围刻意收窄。只有正文含同页标题链接的 [!summary]+ 目录 callout 才会转成 Confluence 官方目录；手工排序和同一行分组会由 Confluence 自动生成的 H2-H3 层级取代。

🧑‍💻 开发

bun install
bun test
bun run dev      # watch 模式,写 dist/main.js
bun run build    # 生产构建(typecheck + 打包)

bun run build 会把 manifest.json 和 styles.css 一起拷到 dist/，整个目录可以直接拖到 .obsidian/plugins/sync-confluence/ 本地测试。

发版：

npm version 0.2.1     # 同步 package.json + manifest.json + versions.json
git push && git push --tags

release.yml 会自动构建并把三个必备文件挂到 GitHub Release 上。

📄 许可证

BSD Zero Clause