Show HN：Agor 推出多人协作 AI 工作空间，让 Agent 不再单打独斗

多人协作 AI。重新一起工作。

Agor 将你的团队和 Agent 汇聚到实时空间看板上，让你能够看到工作进展、开展协作并互相学习。培养具备记忆、共享知识、技能和工具连接能力的 AI 队友，然后让它们进入团队日常工作的场所。一个人行之有效的做法，可以成为整个团队继续构建的基础。

在底层，Agor 是一个用于运行编程 Agent（Claude Code、Codex、Gemini 等）的 Web 工作空间。Agent 在相互隔离的 Git 分支上运行，每个分支都有自己的开发环境和对话历史。通过 npm 安装后，几分钟内就能独自开始使用，随后再邀请团队加入。

文档 · 快速开始 · Agor Cloud 开放公测 · 架构 · 贡献

---

基于你已经使用的 Agent CLI 和 SDK 构建

Agor 不提供自有模型。它驱动你已经在使用的编程 Agent CLI 和 SDK，并允许在不同会话中切换。自带服务提供商和订阅，无需担心供应商锁定。比较各运行框架 →

看板：团队和 Agent 汇聚一处，分支以卡片呈现，区域用于划分空间，会话可实时跟踪。

▶在 YouTube 观看无脚本演示（13 分钟）

---

多数团队在 AI 使用方面的进步仍然依赖个人：好的提示词留在私人聊天中，有用的工作流程存在于某个人的配置里，而运行更多 Agent 意味着需要跟踪更多事项。Agor 让 AI 工作成为团队活动。

将团队和 Agent 汇聚在一起

• 查看并加入彼此的工作。 实时光标、评论和共享会话让同事能够跟踪 Agent 的进度并参与其中，而不必靠截图和状态消息重建上下文。
• 互相学习。 了解同事如何编写提示词、共享上下文，并将有效实践转化为整个团队都能复用的提示词和工作流程。
• 先独自使用，准备好后再扩展。 一个人就能安装 Agor 并在第一天获得价值，随后再邀请同事加入。

共同查看和推进工作

• 实时空间看板。 在二维画布上组织分支和 Agent 会话。区域用于构建工作结构、展示进度阶段，并可在分支拖入时触发模板化提示词。
• 以分支为锚点。 每项工作都是一个 Git 分支，拥有独立的工作目录、开发环境、对话历史和 PR。
• 共享开发环境。 每个分支都可一键启动开发服务器，端口自动分配，避免并行分支发生冲突，让任何人都能检查结果。
• 丰富的会话和会话树。 提供结构化工具输出、模型与推理强度选择器，以及 fork/spawn 衍生关系，方便探索不同方案或协调专业 Agent。
• 多运行时。 Claude Code、Codex、Gemini、OpenCode、Copilot 和 Cursor（测试版）可在不同会话中切换使用。
• 记忆和共享知识。 每个队友都有一个知识库命名空间，用于保存持久、可搜索的上下文，供团队和其他队友继续利用。
• 技能和 MCP。 将可重复的工作方式沉淀为技能，并把队友连接到团队使用的工具。Agor 也通过 MCP 暴露自身能力，让 Agent 能够自行 fork、spawn、安排任务并汇报工作。
• 融入团队的工作场所。 通过 Slack 和 GitHub 等网关渠道联系队友，无须一直待在看板上也能受益。
• 计划任务。 定期执行站会、审计、摘要等工作，无须每次重新输入提示词。

了解运行情况、成本和各人的权限

• 治理和可观测性。 提供分支范围的 RBAC、按用户隔离的凭据和环境变量，以及逐条提示词的 Token 和美元费用统计；历史记录可跨会话持久保存。
• 自托管或托管服务。 可在自己的基础设施上运行 Agor，使用自己的代码仓库和数据库（LibSQL 或 Postgres），并明确选择执行模式：可信本地执行、默认拒绝的沙箱执行，或委托外部执行。不想自行运行？请参阅 Agor Cloud 开放公测。

---

需要 Node.js ≥ 22.12（安装），并确保 PATH 中有 Git。使用 HTTPS 远程仓库还需要可用的系统 CA 信任存储；使用 SSH 远程仓库则需要 SSH 客户端，以及配置好的密钥或 SSH Agent 访问权限。

text npm install -g agor-live

agor init # creates config/database and installs the tools you select agor daemon start # runs the daemon in the background agor open # opens the web UI

之后可使用 agor install 更改或修复所选 Agent 工具；该命令不会初始化或重新创建 Agor。

就这么简单。新手引导向导会创建你的第一个看板，并帮助你培养第一个 AI 队友，之后由它继续协助你。

入门指南介绍了向导的使用方法；如果不想安装 Node，其中还提供 Docker Compose 方案。有关源码构建、Postgres 和团队部署，请参阅扩展安装。有关托管服务，请参阅 Agor Cloud 开放公测。

---

Agor 建立在三个基础实体之上，其他一切都以它们为基础：

• 分支：工作单元。每个分支都是具有独立开发环境和对话记录的一等 Git 工作目录。通常一个分支对应一个功能或 PR。
• 会话与会话树：具有衍生关系的 Agent 对话。通过 fork 探索不同方案（复制上下文），通过 spawn 创建专注于子任务的子会话（使用新的上下文窗口）。
• 看板与区域：展示分支和会话的实时二维画布。将分支拖入区域即可触发模板化提示词。

阅读功能概览 →

---

功能	说明
多人协作与社交	实时光标、成员头像、空间评论和共享多人终端，让团队能够跟踪并加入 Agent 的工作。
看板与区域	用于组织分支和会话的实时空间看板。区域展示工作阶段并触发可复用的提示词。
队友	持续存在的 AI 队友，每个都有独立的知识库命名空间，用于保存持久、可搜索的记忆。通过对话进行培养，再赋予技能、MCP 工具、网关渠道和计划任务。
知识	共享、可搜索的 Markdown 知识库，集中存放决策、操作手册、提示词和 Agent 记忆。
消息网关	无须打开看板，即可通过 Slack 和 GitHub 与队友协作。
调度器	为模板化提示词提供 Cron 风格的触发器，支持队友心跳任务、站会和自动审计。
Agor MCP Server	Agor 通过 MCP 暴露自身能力，Agent 可查看会话、分支和看板，并自行操作系统。
丰富的聊天体验	逐条提示词的 Token 和美元费用统计、模型与推理强度选择器、结构化工具块，以及完成提示音。
环境	每个分支均可一键启动开发服务器，唯一端口自动管理，让团队检查同一个运行中的构建版本。
Artifacts	将实时交互式应用（仪表盘、界面模型、工具）直接渲染到看板上。
安全与 RBAC	分支范围的权限层级、按用户隔离的凭据和环境变量，以及明确的执行模式（simple / sandbox / delegated）。
卡片（测试版）	面向非代码工作流程的通用工作流实体。

---

实时协作：光标、成员头像和限定范围的评论	带结构化工具块的丰富 Agent 会话
具备记忆和技能的持久 AI 队友	原生支持 MCP：Agent 自行操作 Agor
调度器：用于周期性工作的 Cron 风格触发器	每个分支均可一键启动开发环境

---

text graph TB subgraph Clients CLI["CLI (oclif)"] UI["Web UI (React)"] end

Client["Feathers Client<br/>REST + WebSocket"]

subgraph "Agor Daemon"
    Feathers["FeathersJS Server"]
    MCP["MCP HTTP Endpoint<br/>POST /mcp + Bearer auth"]
    Services["Services<br/>Sessions, Tasks, Messages<br/>Boards, Branches, Repos"]
    ORM["Drizzle ORM"]
end

subgraph Executor["Executor (process-isolated)"]
    AgentSDKs["Agent SDKs<br/>Claude · Codex · Gemini · OpenCode"]
end

subgraph Storage
    DB[("LibSQL / Postgres<br/>~/.agor/agor.db")]
    Git["Git Branches<br/>~/.agor/worktrees/"]
    Config["Config<br/>~/.agor/config.yaml"]
end

CLI --> Client
UI --> Client
Client <-->|REST + WebSocket| Feathers

Feathers --> Services
Feathers --> MCP
MCP --> Services
Services --> ORM
Services --> Executor
Executor -.->|JSON-RPC 2.0| MCP

ORM --> DB
Services --> Git
Services --> Config

加载中

守护进程（apps/agor-daemon，FeathersJS）负责数据库、服务、WebSocket 事件和 MCP HTTP 端点。执行器（packages/executor）是进程隔离的运行时，可通过各 Agent 的 SDK 在本地、文件系统沙箱内或委托的外部执行环境中启动 Agent。共享类型、Drizzle 数据库模式和 Git 工具位于 @agor/core（packages/core）中。

完整架构指南 →

text agor/ ├── apps/ │ ├── agor-daemon/ # FeathersJS backend (REST + WebSocket + MCP) │ ├── agor-ui/ # React UI (Ant Design + React Flow) │ ├── agor-cli/ # oclif CLI │ └── agor-docs/ # Docs site (Nextra), the canonical reference, published at agor.live ├── packages/ │ ├── core/ # @agor/core: types, db (Drizzle), git, api │ └── executor/ # Process-isolated agent runt

Multiplayer AI. Work together again.

Agor brings your team and agents together on live spatial boards where you can see the work, collaborate, and learn from each other. Raise AI teammates with memory, shared knowledge, skills, and connections to your tools, then bring them where your team works. What works for one person becomes something the whole team can build on.

Under the hood, Agor is a web workspace for running coding agents (Claude Code, Codex, Gemini, and others) on isolated git branches, each with its own dev environment and conversation history. Install it with npm and start on your own in a few minutes, then bring your team in.

Documentation · Quick Start · Agor Cloud Open Beta · Architecture · Contributing

---

Built on the agent CLIs & SDKs you already use

Agor ships no model of its own. It drives the coding-agent CLIs and SDKs you already run, interchangeable per session. Bring your own provider and subscription, with no vendor lock-in. Compare the harnesses →

The board: your team and agents in one place, with branches as cards, zones as regions, and sessions you can follow live.

▶Watch the unscripted demo on YouTube (13 min)

---

Most teams are getting better at AI one person at a time. Good prompts sit in private chats, useful workflows live in one person's setup, and running more agents means more to keep track of. Agor makes AI work a team activity.

Bring your team and agents together

• See and join each other's work. Live cursors, comments, and shared sessions let colleagues follow an agent's progress and step in, instead of reconstructing it from screenshots and status messages.
• Learn from each other. See how colleagues prompt, share context, and turn the practices that work into prompts and workflows the whole team can reuse.
• Start solo, expand when ready. One person can install Agor and get value on day one, then bring colleagues in.

See the work and shape it together

• A live spatial board. Organize branches and agent sessions on a 2D canvas. Zones give the work structure, show its stage, and can fire a templated prompt when a branch is dropped in.

• Branches as the anchor. Every piece of work is a git branch with its own working directory, dev environment, conversation history, and PR.

• Shared dev environments. A one-click dev server per branch, with ports assigned automatically so parallel branches never collide and anyone can inspect the result.

• Rich sessions and session trees. Structured tool output, model and effort selectors, and fork/spawn genealogy for exploring alternatives or coordinating specialist agents.

• Multi-runtime. Claude Code, Codex, Gemini, OpenCode, Copilot, and Cursor (beta) are interchangeable per session.

• Memory and shared knowledge. Each teammate gets a Knowledge-base namespace for durable, searchable context that the team and other teammates can build on.

• Skills and MCP. Capture repeatable ways of working as skills, and connect teammates to the tools your team uses. Agor also exposes itself over MCP, so agents can fork, spawn, schedule, and report on their own work.

• Where your team works. Reach teammates through gateway channels such as Slack and GitHub. Nobody has to live on the board to benefit.

• Schedules. Run standups, audits, digests, and other recurring work without a fresh prompt each time.

Know what's running, what it costs, and who can do what

• Governance and observability. Branch-scoped RBAC, per-user credentials and env vars, and per-prompt token and dollar accounting with durable history across sessions.
• Self-hosted or managed. Run Agor on your own infrastructure with your repos and database (LibSQL or Postgres) and explicit execution modes (trusted local, fail-closed sandbox, or delegated external execution). Prefer not to run it yourself? See Agor Cloud Open Beta.

---

Requires Node.js ≥ 22.12 (install) and Git on PATH. HTTPS remotes also require a working system CA trust store; SSH remotes require an SSH client and configured keys or agent access.

npm install -g agor-live

agor init           # creates config/database and installs the tools you select
agor daemon start   # runs the daemon in the background
agor open           # opens the web UI

Use agor install later to change or repair the selected agentic tools; it does not initialize or recreate Agor.

That's it. The onboarding wizard creates your first board and helps you raise your first AI teammate, which takes it from there.

The Getting Started guide walks through the wizard and includes a Docker Compose path if you would rather not install Node. For source builds, Postgres, and team setups, see Extended Installation. For managed hosting, see Agor Cloud Open Beta.

---

Agor is built on three foundational entities. Everything else builds on these:

• Branches : the unit of work. A first-class git working directory on its own branch, with an isolated dev environment and its own conversations. Conventionally 1 branch = 1 feature/PR.
• Sessions& Trees: agent conversations with genealogy. Fork to explore alternatives (copies context), spawn subsessions for focused subtasks (fresh context window).
• Boards& Zones: a live 2D canvas of branches and sessions. Drop a branch into a zone to fire a templated prompt.

Read the Features Overview →

---

|
---|---
Multiplayer& Social | Live cursors, facepiles, spatial comments, and a shared multiplayer terminal, so the team can follow and join agent work.
Boards& Zones | A live spatial board for organizing branches and sessions. Zones show the stage of the work and trigger reusable prompts.
Teammates | Persistent AI teammates, each with its own Knowledge-base namespace for durable, searchable memory. Taught conversationally, then equipped with skills, MCP tools, gateway channels, and schedules.
Knowledge | A shared, searchable markdown knowledge base. One place for decisions, runbooks, prompts, and agent memory.
Message Gateway | Work with teammates from Slack and GitHub, without opening the board.
Scheduler | Cron-style triggers for templated prompts. Powers teammate heartbeats, standups, and automated audits.
Agor MCP Server | Agor exposes itself over MCP. Agents introspect sessions, branches, and boards, and drive the system themselves.
Rich Chat UX | Per-prompt token and dollar accounting, model/effort selectors, structured tool blocks, completion chimes.
Environments | One-click dev servers per branch with automatically managed unique ports, so the team can inspect the same running build.
Artifacts | Live, interactive apps (dashboards, mockups, tools) rendered directly on the board.
Security& RBAC | Branch-scoped permission tiers, per-user credentials and env vars, and explicit execution modes (simple / sandbox / delegated).
Cards (Beta) | Generic workflow entities for non-code workflows.

---

Work together in real time: cursors, facepile, scoped comments	Rich agent sessions with structured tool blocks
Persistent AI teammates with memory and skills	MCP-native: agents drive Agor themselves
Scheduler: cron-style triggers for recurring work	One-click dev environments per branch

---

graph TB
    subgraph Clients
        CLI["CLI (oclif)"]
        UI["Web UI (React)"]
    end

    Client["Feathers Client<br/>REST + WebSocket"]

    subgraph "Agor Daemon"
        Feathers["FeathersJS Server"]
        MCP["MCP HTTP Endpoint<br/>POST /mcp + Bearer auth"]
        Services["Services<br/>Sessions, Tasks, Messages<br/>Boards, Branches, Repos"]
        ORM["Drizzle ORM"]
    end

    subgraph Executor["Executor (process-isolated)"]
        AgentSDKs["Agent SDKs<br/>Claude · Codex · Gemini · OpenCode"]
    end

    subgraph Storage
        DB[("LibSQL / Postgres<br/>~/.agor/agor.db")]
        Git["Git Branches<br/>~/.agor/worktrees/"]
        Config["Config<br/>~/.agor/config.yaml"]
    end

    CLI --> Client
    UI --> Client
    Client <-->|REST + WebSocket| Feathers

    Feathers --> Services
    Feathers --> MCP
    MCP --> Services
    Services --> ORM
    Services --> Executor
    Executor -.->|JSON-RPC 2.0| MCP

    ORM --> DB
    Services --> Git
    Services --> Config

Loading

The daemon (apps/agor-daemon, FeathersJS) owns the database, services, WebSocket events, and the MCP HTTP endpoint. The executor (packages/executor) is a process-isolated runtime that spawns agents via their SDKs locally, inside the filesystem sandbox, or through a delegated external substrate. Shared types, the Drizzle schema, and git utilities live in @agor/core (packages/core).

Full Architecture Guide →

agor/
├── apps/
│   ├── agor-daemon/   # FeathersJS backend (REST + WebSocket + MCP)
│   ├── agor-ui/       # React UI (Ant Design + React Flow)
│   ├── agor-cli/      # oclif CLI
│   └── agor-docs/     # Docs site (Nextra), the canonical reference, published at agor.live
├── packages/
│   ├── core/          # @agor/core: types, db (Drizzle), git, api
│   └── executor/      # Process-isolated agent runtime
└── context/           # Agent-oriented cheat sheets and design docs

---

The fastest path to a running dev instance from source:

git clone https://github.com/preset-io/agor
cd agor
docker compose up
# Visit http://localhost:5173 → login: admin@agor.live / admin

Prefer running locally without Docker? The two-process workflow (daemon in watch mode + UI dev server) and the .agor.yml variants (sqlite / postgres / rich / HA / docs) are documented in the Development Guide. It also covers running Agor inside Agor for dogfooding.

See CONTRIBUTING.md for the contribution workflow, and CLAUDE.md for the agent-oriented map of the codebase.

---

⭐️ I pledge to fix a GitHub issue for every star Agor gets :)

---

Community

Business Source License 1.1 (BUSL-1.1). Agor is source-available, not open source, before the Change Date.

The Additional Use Grant permits production use, including internal and self-hosted commercial use. It does not permit commercializing Agor itself by offering its agent-orchestration functionality to third parties as a product or service, whether hosted, managed, or bundled for customers to operate. Consulting, support, integration, modification, use within a broader product or service, and single-customer internal deployments remain permitted subject to the license terms. Contact Preset, Inc. about alternative commercial licensing.

On January 15, 2029 , or the fourth anniversary of the first public BSL distribution of a particular version (whichever comes first), that version converts to the Apache License 2.0. The license text controls if this summary differs from it.

Heavily prompted by@mistercrunch (Preset, Apache Superset, Apache Airflow), built by an army of Claudes and Codexes.

Read more: Announcing Agor Cloud Open Beta · Agent Modeling 101 · Raise a team helper agent in an afternoon · all posts →