feat: create moss agent as default

Author: thirdgerb

README.md

@@ -1,6 +1,6 @@
 # 项目概述
 
-项目名为 `MOS-Shell` (Model-oriented Operating System Shell), 包含两个几个核心目标:
+项目名为 `MOS-Shell` (Model-oriented Operating System Shell), 包含几个核心目标:
 
 1. `MOS`: 为 AI 大模型提供一个 "面向模型的操作系统", 可以将 跨设备/跨进程 的功能模块, 以 "树" 的形式提供给模型操作.
 1. `Shell Runtime`: 为 AI Agent 提供一个持续运转的运行时 (Runtime), 联通所有功能模块 (称之为 Channel, 对标 python 的
@@ -12,6 +12,8 @@
 
 目标是 AI 大模型作为大脑, 不仅可以思考, 还可以 实时/并行/有序 地操作包括 计算机/具身躯体 来进行交互.
 
+更多设计思路请访问: [核心设计思想综述](https://ycnrlabqki3v.feishu.cn/wiki/QCKUwAX7tiUs4GkJTkLcMeWqneh)
+
 MOS-Shell 是 Ghost In Shells (中文名: 灵枢) 项目创建的新交互范式架构, 是第二代 MOSS 架构 (完善了 ChannelApp 和
 Realtime-Actions 思想).
 第一代 MOSS 架构 (全代码驱动 + FunctionToken) 详见 [GhostOS](https://github.com/ghostInShells/ghostos)

…kspace/configs/instructions/behaviors.md …e/configs/moss_instructions/behaviors.mdexamples/.workspace/configs/instructions/behaviors.md renamed to examples/.workspace/configs/moss_instructions/behaviors.md examples/.workspace/configs/instructions/behaviors.md renamed to examples/.workspace/configs/moss_instructions/behaviors.md

@@ -1,4 +1,3 @@
-# from class: demo.linxu_demo.channels.no_ppt_channel:NoPPTConfig
 ppt_list:
   - dirname: example
     description: 一个测试的ppt，介绍了人和AI的关系

…orkspace/configs/instructions/persona.md …ace/configs/moss_instructions/persona.mdexamples/.workspace/configs/instructions/persona.md renamed to examples/.workspace/configs/moss_instructions/persona.md examples/.workspace/configs/instructions/persona.md renamed to examples/.workspace/configs/moss_instructions/persona.md

@@ -1,4 +1,8 @@
 web_list:
+  - id: "moss-repository"
+    url: "https://github.com/GhostInShells/MOSShell"
+    description: |-
+      MOSShell 当前项目库 github 地址
   - id: "moss-architecture-doc"
     url: "https://ycnrlabqki3v.feishu.cn/wiki/QCKUwAX7tiUs4GkJTkLcMeWqneh"
     description: "MOSS (模型可操作系统 - 流式代码解释器) 系统原理与应用综述"
@@ -14,25 +18,25 @@ web_list:
   - id: "ghostos-repository"
     url: "https://github.com/GhostInShells/ghostos"
     description: |-
-      GhostOS 代码仓库地址，是用来做 ghost in shells 架构思想探索的验证库。
+      GhostOS 代码仓库地址，是 2024 年用来做 ghost in shells 架构思想探索的验证库。目前已经终止迭代.
   - id: "moss-article"
     url: "https://arxiv.org/abs/2409.16120"
     description: |-
-      这是我们24年研究 AI 通过全代码驱动系统的相关论文预印版
+      这是我们24年研究 AI 通过全代码驱动系统的相关论文预印版. 是第一代 MOSS 架构思想. 现在是第二代. 
   - id: "机械臂演示视频汇总"
     url: "https://ycnrlabqki3v.feishu.cn/wiki/FiOOwBKG0iSEONkGRKvcyDHOnaf"
     description: |-
-      这里提供了我们 6dof jetarm 机械臂积累的展示视频
+      这里提供了我们 6dof jetarm 机械臂积累的展示视频. 是当前 MOSShell 的展示视频之一.
   - id: "20年语音交互视频展示视频"
     url: "https://www.bilibili.com/video/BV1tK4y1a75B/?share_source=copy_web&vd_source=d16eb4aa121a670f02fa576471d30a41"
     description: |-
-      朱明 20年项目中，通过微信语音控制网页，通过网页的小视频做对话式视频的案例
+      ZhuMing 20年项目中，通过微信语音控制网页，通过网页的小视频做对话式视频的案例. 是 GhostInShells 项目理念起点的示例之一.
   - id: "20年语音交互控制网页游戏视频"
     url: "https://www.bilibili.com/video/BV1tK4y1a75B/?share_source=copy_web&vd_source=d16eb4aa121a670f02fa576471d30a41&t=290"
     description: |-
-      朱明 20年项目中，通过微信语音控制网页 “方向迷宫” 小游戏的视频，将网页游戏和微信进行了跨设备的同构。
+      ZhuMing 20年项目中，通过微信语音控制网页 “方向迷宫” 小游戏的视频，将网页游戏和微信进行了跨设备的同构。 是 MOSS 控制具身智能思路起点的示例之一.
   - id: "20年关于自然语言编程的视频"
     url: "https://www.bilibili.com/video/BV1tK4y1a75B/?share_source=copy_web&vd_source=d16eb4aa121a670f02fa576471d30a41&t=672"
     description: |-
-      朱明 20年项目中，讨论到未来实现自然语言编程，对话式编程的设想。 
+      ZhuMing 20年项目中，讨论到未来实现自然语言编程，对话式编程的设想。 是 MOSS 代码驱动理念起点的示例之一.
     

examples/.workspace/configs/no_ppt.yaml

@@ -12,6 +12,14 @@
 import live2d.v3 as live2d
 import pygame
 from ghoshell_container import Container
+
+current_dir = os.path.dirname(os.path.abspath(__file__))
+try:
+    import miku_channels
+except ImportError:
+    # 加载当前路径.
+    sys.path.append(current_dir)
+
 from miku_channels.arm import left_arm_chan, right_arm_chan
 from miku_channels.body import body_chan
 from miku_channels.elbow import left_elbow_chan, right_elbow_chan
@@ -21,40 +29,18 @@
 from miku_channels.head import head_chan
 from miku_channels.leg import left_leg_chan, right_leg_chan
 from miku_channels.necktie import necktie_chan
+from miku_provider import init_live2d, init_pygame
 from ghoshell_moss.core.shell import new_shell
 from ghoshell_moss_contrib.example_ws import workspace_container, get_example_speech
 import pathlib
 
-# 加载当前路径.
-current_dir = os.path.dirname(os.path.abspath(__file__))
-sys.path.append(current_dir)
-
 # 全局状态
-model: live2d.LAppModel | None = None
 WIDTH = 600
 HEIGHT = 800
 
+model: live2d.LAppModel | None = None
 
 # 初始化Pygame和Live2D
-def init_pygame():
-    pygame.init()
-    display = (WIDTH, HEIGHT)
-    screen = pygame.display.set_mode(display, pygame.DOUBLEBUF | pygame.OPENGL)
-    pygame.display.set_caption("Digital Human Demo with PyChannel")
-    return screen, display
-
-
-# 初始化Live2D模型
-def init_live2d(model_path: str, container: Container):
-    global model
-    live2d.init()
-    live2d.glInit()
-    model = live2d.LAppModel()
-    model.LoadModelJson(model_path)
-    model.Resize(WIDTH, HEIGHT)
-    # model.SetAutoBlinkEnable(False)
-    # model.SetAutoBreathEnable(True)
-    container.bind(live2d.LAppModel, model)
 
 
 async def speak(duration: float = 5.0, speed: float = 1.0, max_open: float = 0.9, min_open: float = 0.0):
@@ -160,9 +146,10 @@ async def speaking():
 
 async def run_agent_and_render(container: Container, speech: Speech | None = None):
     # 初始化 Pygame 和 Live2D
+    global model
     screen, display = init_pygame()
     model_path = join(dirname(__file__), "model/miku.model3.json")
-    init_live2d(model_path, container)
+    model = init_live2d(model_path, container)
 
     # 保持窗口打开，直到用户关闭
     running = True
@@ -215,7 +202,7 @@ async def run_agent_and_render(container: Container, speech: Speech | None = Non
 def main():
     # 运行异步主函数
     with workspace_container(WORKSPACE_DIR) as container:
-        speech = get_example_speech(container)
+        speech = get_example_speech(container, default_speaker='saturn_zh_female_keainvsheng_tob')
         asyncio.run(run_agent_and_render(container, speech))
 
 

examples/.workspace/configs/web.yaml

@@ -0,0 +1,143 @@
+import asyncio
+import os
+import sys
+from os.path import dirname, join
+
+import live2d.v3 as live2d
+import pygame
+from ghoshell_container import Container, get_container
+
+try:
+    import miku_channels
+except ImportError:
+    current_dir = os.path.dirname(os.path.abspath(__file__))
+    sys.path.append(current_dir)
+
+from miku_channels.arm import left_arm_chan, right_arm_chan
+from miku_channels.body import body_chan
+from miku_channels.elbow import left_elbow_chan, right_elbow_chan
+from miku_channels.expression import expression_chan
+from miku_channels.eye import eye_chan
+from miku_channels.eyebrow import eyebrow_left_chan, eyebrow_right_chan
+from miku_channels.head import head_chan
+from miku_channels.leg import left_leg_chan, right_leg_chan
+from miku_channels.necktie import necktie_chan
+from ghoshell_moss.transports.zmq_channel import ZMQChannelProvider
+from ghoshell_moss import Channel
+
+# 全局状态
+model: live2d.LAppModel | None = None
+WIDTH = 600
+HEIGHT = 800
+
+
+# 初始化Pygame和Live2D
+def init_pygame():
+    pygame.init()
+    display = (WIDTH, HEIGHT)
+    screen = pygame.display.set_mode(display, pygame.DOUBLEBUF | pygame.OPENGL)
+    pygame.display.set_caption("Digital Human Demo with PyChannel")
+    return screen, display
+
+
+# 初始化Live2D模型
+def init_live2d(model_path: str, con: Container | None = None) -> live2d.LAppModel:
+    global model
+    live2d.init()
+    live2d.glInit()
+    model = live2d.LAppModel()
+    model.LoadModelJson(model_path)
+    model.Resize(WIDTH, HEIGHT)
+    # model.SetAutoBlinkEnable(False)
+    # model.SetAutoBreathEnable(True)
+    con = con or get_container()
+    con.bind(live2d.LAppModel, model)
+    return model
+
+
+def miku_body() -> Channel:
+    head_chan.import_channels(
+        expression_chan,
+        # mouth_chan, 不把嘴巴给大模型调用.
+        eye_chan,
+        eyebrow_left_chan,
+        eyebrow_right_chan,
+    )
+    body_chan.import_channels(
+        head_chan,
+        left_arm_chan,
+        right_arm_chan,
+        left_elbow_chan,
+        right_elbow_chan,
+        necktie_chan,
+        left_leg_chan,
+        right_leg_chan,
+    )
+    return body_chan
+
+
+async def run_game_with_zmq_provider(address: str = "tcp://localhost:5555", con: Container | None = None):
+    # 初始化 Pygame 和 Live2D
+    screen, display = init_pygame()
+    model_path = join(dirname(__file__), "model/miku.model3.json")
+    init_live2d(model_path, con=con)
+
+    # 保持窗口打开，直到用户关闭
+    running = True
+    clock = pygame.time.Clock()
+    font = pygame.font.SysFont(None, 24)
+
+    # 创建一个任务来运行 agent
+    provider = ZMQChannelProvider(
+        address=address,
+        container=con,
+    )
+    _miku = miku_body()
+    task = asyncio.create_task(provider.arun(_miku))
+
+    try:
+        while running:
+            # 处理pygame事件
+            for event in pygame.event.get():
+                if event.type == pygame.QUIT:
+                    running = False
+
+            # 更新和绘制模型
+            model.Update()
+            live2d.clearBuffer(255, 255, 255, 150.0)
+            model.Draw()
+
+            # 显示完成信息
+            info_text = font.render("演示完成，按关闭按钮退出", True, (255, 255, 255))
+            screen.blit(info_text, (10, 10))
+
+            pygame.display.flip()
+            clock.tick(60)
+
+            # 让出控制权给其他协程，确保agent也能运行
+            await asyncio.sleep(0.01)
+    except KeyboardInterrupt:
+        pass
+    finally:
+        # 关闭 provider.
+        try:
+            provider.close()
+            await task
+        except asyncio.CancelledError:
+            pass
+        # 清理资源
+        live2d.dispose()
+        pygame.quit()
+
+
+async def run_provider(address: str = "tcp://localhost:5555"):
+    _body_chan = miku_body()
+    provider = ZMQChannelProvider(
+        address=address,
+        container=get_container(),
+    )
+
+    try:
+        await provider.arun(_body_chan)
+    except KeyboardInterrupt:
+        pass