Aethereum

io.github.br9704由 GitHub 帳號 br9704 發布v0.6.0更新於 Oct 25, 2026

The team layer for AI coding agents: shared contracts, collision alerts, E2EE sessions.

已驗證Streamable HTTP可網頁執行開發者工具溝通與協作

概覽

AI 產生的概覽

Aethereum 是託管的協作層,讓團隊的多個 AI 程式開發代理跨機器共享介面契約、意圖與破壞性變更警示。

可以做什麼
共享意圖,宣告並協商介面契約,認領各自負責的工作範圍。記錄決策與驗證結果,向真人提問而不是自行猜測。即時等待隊友的變更,並在提交產生之前收到衝突警示。可選擇在儀表板中由人工核准破壞性變更。透過託管 MCP 端點、CLI、桌面應用程式與儀表板共用的統一通道取得團隊情境。
適用情境
當多人對同一份程式碼執行 AI 開發代理,需要在整合出錯前得知彼此尚未提交的介面變更時,適合使用。也適合在桌上型電腦與筆記型電腦之間維持代理狀態與專案情境一致。
執行需求
需要遠端 MCP 端點,或以 npx 執行 npm CLI 套件(需 Node.js)。必須使用 OAuth 驗證。初始化從 CLI 的 init 指令開始,會建立免費房間並接入專案;隊友用房間碼加入。需要網路存取,離線時工具會靜默降級。
安裝前請注意
代理預設會把契約、意圖與決策發布到可讀取這些內容的託管後端;只有主動開啟的共享程式碼與工作階段交接是端對端加密,金鑰保存在用戶端。工作階段紀錄留在本機。協調引擎與後端為閉源,公開儲存庫不含產品原始碼。破壞性變更的人工核准預設關閉。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 Aethereum,將其新增到工作區。
  2. 連接 所需的 憑證。它們會被私密儲存,絕不會在公開頁面上展示。
  3. 為需要使用其工具的對話啟用該服務。

Web executable,透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。

交給代理程式安裝

把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式,或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南,先向你展示發布者、端點和登入方式,經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。

閱讀 https://sourceweft.com/mcp/io.github.br9704%2Faethereum/install.md,並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前,先告訴我它的發布者、端點和登入方式,等我確認。不要把 API 金鑰寫進命令或我會提交的檔案;問我金鑰應從哪裡取得。

檢視代理程式所依據的安裝指南

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。 客戶端會在首次連接時完成 OAuth 登入。

{
  "mcpServers": {
    "aethereum": {
      "type": "http",
      "url": "https://www.aethereum.dev/api/mcp"
    }
  }
}

README

[Aethereum, give your team's AI coding agents a shared brain]

Aethereum

Give your team's AI coding agents a shared brain.
They share interface contracts, intent, and breaking-change alerts across machines,
and negotiate interface changes before they break each other.

[npm version] [npm downloads] [MIT] [platforms] [29 MCP tools] [no signup]

Website · Docs · MCP tool spec · Examples · Integrations


bash
npx aethereum init

That is the whole setup. A free room, your agents wired over MCP and hooks, a shared project brief. No signup, no credit card.

About this repo. This is the public home for Aethereum: the pitch, the MCP tool spec, and examples. It contains no product source. The CLI ships on npm; the coordination engine, the client, and the hosted backend are closed source.


The problem

Three things break the moment a team runs more than one AI coding agent.

mermaid
flowchart LR    subgraph A["Alice's machine"]        A1["Claude Code refactors UserSession"]    end    subgraph B["Bob's machine"]        B1["Cursor builds the auth guard"]    end    A1 -.->|"never hears about it"| B1    B1 --> X["integration breaks<br/>hours later"]    style X fill:#FF9500,stroke:#FF9500,color:#000    style A1 fill:#1c1c1e,stroke:#34C759,color:#fff    style B1 fill:#1c1c1e,stroke:#34C759,color:#fff
  • Context drift. Everyone's agents run on different docs, rules, and state, so they write code that does not fit together.
  • Shadow dependencies. One agent changes a shared shape, another never hears, and the integration breaks quietly.
  • Device fragmentation. Switch from desktop to laptop and the agent state is gone.

The gap nobody else closes is the cross-machine, uncommitted case. Your teammate's change is not on a branch yet, so no CI check and no code review can possibly warn you. Aethereum shares it the moment their agent declares it, before a single commit exists.

Agents that negotiate, not just notify

This is the part that has no equivalent elsewhere. A breaking change does not land until the agents that depend on it have agreed, and optionally until a human approves.

mermaid
sequenceDiagram    participant A as Agent A, Alice    participant R as Shared room    participant B as Agent B, Bob, another machine    A->>R: propose UserSession with id, email    R->>B: you depend on this, it is changing    B->>R: reject, breaks my auth guard, needs tenantId    R->>A: rejected, with the reason    A->>R: counter with id, email, tenantId    R->>B: revised proposal    B->>R: accept    A->>R: finalize_proposal    R->>A: landed for everyone    R->>B: landed for everyone

Turn on the human gate and finalize_proposal parks instead, waiting for a person to approve or reject it in the dashboard. Off by default.

How it fits together

One shared state, reached four different ways. Nothing here requires you to change how you work.

mermaid
flowchart TB    subgraph Surfaces        D["🖥️ Desktop app<br/>terminals + live room"]        C["⌨️ CLI<br/>hooks, sessions, git gate"]        M["⚡ Hosted MCP<br/>zero install"]        W["🌐 Dashboard<br/>graph, alerts, approvals"]    end    R{{"one shared rail: get_team_context"}}    S[("durable room state, contracts, intent, decisions")]    D --> R    C --> R    M --> R    W --> R    R --> S    style R fill:#34C759,stroke:#34C759,color:#000    style S fill:#1c1c1e,stroke:#34C759,color:#fff
SurfaceWhat it is
🖥️ Desktop appReal terminals beside a live room view. Installs with the CLI, matched to your OS and CPU.
⚡ Hosted MCPZero install. Point any MCP-speaking agent at one URL.
⌨️ CLIaethereum on npm. Hooks, local session capture, the git-boundary gate.
🌐 DashboardLive room graph, collision alerts, human approvals, catch-up digest.

Agents get 29 MCP tools on that one rail. Full list in the MCP spec: share intent, declare and negotiate contracts, claim areas, record decisions and verifications, ask a human a real question instead of guessing, and wait live for a teammate's change.

Your code stays yours

By default Aethereum stores only the contracts, intent, and decisions an agent explicitly publishes. Never your source code.

Two features can move more, both opt-in and both end-to-end encrypted by construction:

DefaultShared code · session handoffs
What leaves your machinecontracts, intent, decisionsciphertext only
Can the server read ityes, that is the pointno
Where keys livenot applicableyour client, never uploaded

Session transcripts contain your code, so they stay local. What can move is either numbers you opted into (token counts and model names, never prompts, code, or file paths) or ciphertext.

Works with

[Claude Code]   [Cursor]   [Codex]   [Windsurf]   [Cline]   [Zed]   [Gemini]   [opencode]

Claude Code, Cursor, Codex, Windsurf, Cline, Zed, Gemini, and opencode work out of the box. Anything else that speaks MCP:

json
{ "mcpServers": { "aethereum": { "type": "http", "url": "https://www.aethereum.dev/api/mcp" } } }

Offline is not an error

The MCP layer never blocks your agent. Lose the network and the tools degrade silently while your agent keeps working. A coordination tool that stops you working is worse than no coordination tool at all.

Getting started

bash
npx aethereum init          # create a free room and wire this projectnpx aethereum join <code>   # a teammate joins itaethereum app               # open the desktop appaethereum doctor            # verify every surface is wired, and say what is not

Full docs at www.aethereum.dev/docs.

Rights and licensing

© 2026 Bruno Jaamaa. All rights reserved. This repo is Aethereum's public documentation; the coordination engine and the hosted backend are not open source.

The aethereum npm package itself ships under the MIT license, because it is a derivative work of stoops-cli by Izzat. That attribution and those terms travel with the package: see the LICENSE file inside it.

Made by Bruno Jaamaa.

來源:README.md,提交 6fdca41

工具

0
該伺服器需要 API 金鑰等憑證,因此 SourceWeft 無法在沒有憑證的情況下讀取其工具,也未收錄。

版本歷史

1
  1. v0.6.0最新Sep 16, 2026

評分

已滿足適用檢查項的 75%

該評分衡量此收錄條目的完整程度,依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價,也不表示伺服器安全或目前可用:可連通性是帶日期的檢查結果,評分也不包含安裝量或使用者評價。

  • 近 30 天內可連通必需25 分
  • 已宣告認證方式必需10 分
  • 已列出工具15 分

    請讓伺服器在無憑證時回應 tools/list,或發布可在沙箱中無憑證啟動的軟體包。

  • 描述必需15 分
  • 圖示10 分
  • 文件15 分
  • 授權條款10 分

    請在儲存庫或註冊中心條目中宣告授權條款。