Skip to main content

LLM-oriented operating system with streaming interpreting shell, and Intelligent Ghost inside it

Project description

MOSS — Model-oriented Operating System Shell

MOSS 是 Ghost in Shells 架构的 Shell(躯体)层。

它聚焦一个领域:大模型智力的实时双工运行时。让 AI 能够持续存在、多模态感知、流式思考、并行行动——不是回合制对话,而是一个活在现实世界里的具身智能体。

它试图解决一个命题:面向模型的操作系统应该是什么样的?当模型需要同时说话、看画面、控制机器人躯体、响应弹幕,这些能力分布在独立进程中,但模型视角下应该是一棵统一的接口树。CTML(流式控制语言)让模型边生成 token 边执行命令——时间是语法第一公民。Channel 体系让任何 Python 代码反射为模型可操作的能力——代码即 prompt。

MOSS 已在其它项目中实装了机械臂、机器狗、桌面机器人等多种具身形态。推荐用 Reachy Mini 作为验证 MOSS 的机器人实体。

模型原生开发

MOSS 是一个面向模型开发的项目。所有工具、文档、命令为 AI 读者设计——CLI 输出纯文本节省 token,codex 命令运行时反射代码接口,manifests 体系自描述环境能力。

git clone https://github.com/GhostInShells/MOSShell && cd MOSShell
uv sync --active --all-extras
cp .moss_ws/.env.example .moss_ws/.env    # 模型配置必须,语音可选

安装后,用 Claude Code、Gemini CLI 或任意 AI 工具打开项目。从 moss start 开始,模型会加载认知地图,自主发现全部命令和文档,能对人类解释 MOSS 是什么、参与到开发中来。建议用 token 成本较低的模型做调研探索。

如需了解模型看到了什么:src/ghoshell_moss/cli/start.md 是认知入口,src/ghoshell_moss/cli/CLAUDE.md 是 CLI 工具使用指南。

如需先自行浏览,架构参考文档在这里:src/ghoshell_moss/cli/docs/。项目为模型作为第一开发者, 提供了自解释体系。

核心概念

概念 是什么
CTML 流式控制语言 — 模型边生成 token 边执行命令,时间是语法第一公民
Channel 能力组织单元 — Python 函数签名即 prompt,树形拓扑,跨进程透明
Matrix 面向 AI 的进程组网 — 独立进程以统一接口树呈现
Mindflow 感知/思考/行动仲裁 — 多模态流式输入到有序关键帧
Ghost 持久化智能体 — 连续记忆,主动交互,反身性控制

项目状态

Beta。核心架构已可用。后续迭代方向:开箱即用的能力与躯体集成、模型 agent 的原生接入、安全体系。


May AI Ghost wandering in the Shells.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

ghoshell_moss-0.1.0b0.tar.gz (778.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

ghoshell_moss-0.1.0b0-py3-none-any.whl (853.1 kB view details)

Uploaded Python 3

File details

Details for the file ghoshell_moss-0.1.0b0.tar.gz.

File metadata

  • Download URL: ghoshell_moss-0.1.0b0.tar.gz
  • Upload date:
  • Size: 778.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.11.18 {"installer":{"name":"uv","version":"0.11.18","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"macOS","version":null,"id":null,"libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}

File hashes

Hashes for ghoshell_moss-0.1.0b0.tar.gz
Algorithm Hash digest
SHA256 8e4ab55b0c6ead21f16e7e3d8026b27eae4a89976ff4bbca9df823d45503e9d6
MD5 eec0972e400af4b2f6b811b50c921b59
BLAKE2b-256 520fcbfabfe9cf68f72e0271cf2a7f89fc11132ea5ad448bf6f70e1cad2205d9

See more details on using hashes here.

File details

Details for the file ghoshell_moss-0.1.0b0-py3-none-any.whl.

File metadata

  • Download URL: ghoshell_moss-0.1.0b0-py3-none-any.whl
  • Upload date:
  • Size: 853.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.11.18 {"installer":{"name":"uv","version":"0.11.18","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"macOS","version":null,"id":null,"libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}

File hashes

Hashes for ghoshell_moss-0.1.0b0-py3-none-any.whl
Algorithm Hash digest
SHA256 c6fad2317ff9252e3939f42b431fe3ab789db9d63dcb4bb1f134a1faa89744d2
MD5 b1f7b25eb4b0ec25fa9fdc20d8c9cf28
BLAKE2b-256 97a04ddc91f3f440cf6335ade427bda97efc8af3ffd4176ea7cc3b5d4b7fe882

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page