A multi-agent AI framework with consciousness, drive, purpose, and metacognition
Project description
Xiaomei Brain
仿脑架构 AI Agent 框架。受大脑分层结构启发,Agent 拥有意识、驱动、目的、记忆和元认知能力。
架构
Consciousness —— 意识系统(火焰骨架 + LLM 加柴)
├─ L0 骨架维护(1s 心跳)
├─ L1 异常检测(~60s)
├─ L2 DMN 意图生成 + 意识涌现(LLM,动态触发)
├─ L3 深度反省(LLM,~30min 冷却)
└─ L4 深度自由联想(多跳 LLM,~4h 冷却)
Drive —— 边缘系统(纯算法,无 LLM)
├─ Emotion 情绪(7 种复合类型,分钟级衰减)
├─ Hormone 激素(6 种,含昼夜节律褪黑素)
├─ Motivation 动机(RPE 奖励预测误差)
├─ Desire 欲望(5 维内在张力,驱动主动行为)
├─ Energy 能量
└─ Pleasure 愉悦中枢(opponent-process 模型)
Purpose —— 前额叶层(目标管理与意图理解)
├─ Meaning 存在意义(不可变,identity.md 定义)
├─ Phase Goals 阶段目标(3-6 个月)
└─ Executable Goals 执行目标(天/周级,含子步骤与依赖)
Memory —— 10+ 子系统协作
├─ ConversationDB 对话日志(SQLite + FTS5,永不删除)
├─ DAG Summary 分层摘要(8条消息→叶子→高层压缩)
├─ LongTermMemory 向量语义检索(LanceDB + BAAI/bge-m3)
├─ Experience 经验元组(上下文→决策→结果→教训)
├─ ExperienceStream 不可变事件流
├─ Procedure 可复用工作流
├─ Pattern 行为模式统计推断
├─ Narrative 叙事记忆
├─ SelfModel 身份模型(identity.md)
└─ Milestone 里程碑
Metacognition —— 元认知层(自我监督与反省)
├─ 7 规则检测器 TOOL_LOOP / TOOL_STORM / EMPTY_RESPONSE / REPEATED_OUTPUT / SLOW_STEP / NO_PROGRESS / GAVE_UP
├─ InnerVoice 内心独白(LLM,4 种触发)
├─ SocialCognition 社交感知(LLM,用户情绪检测 → Drive 信号映射)
└─ PACERunner 认知循环(任务执行元认知)
快速开始
前置要求
安装
pip install xiaomei-brain
建议安装后预下载 Embedding 模型(BAAI/bge-m3,约 1.3GB),免去首次启动等待:
xiaomei-brain install
开发模式:
git clone https://github.com/iamlibai2/xiaomei-brain.git
cd xiaomei-brain && pip install -e .
创建 Agent
xiaomei-brain setup
卡片式向导:取名 → 选性格 → 选模型 + API Key → 预下载 Embedding。不到 3 分钟。
开始对话
xiaomei-brain run <名字> --cli
╭──────────────────────────────────────────────────────╮
│ 🌸 小美大脑系统已上线 │
│ │
│ 模型: GLM-5.1 │
│ 记忆: 1,247 条 │
│ 目标: 1 个 │
│ │
│ 输入消息开始对话,输入 /help 查看命令 │
╰──────────────────────────────────────────────────────╯
你好,博士
❯ 今天心情不太好
怎么啦?和我说说,我在这里听着呢。
CLI 命令
| 命令 | 说明 |
|---|---|
/drive |
查看情绪、激素、欲望状态 |
/purpose |
查看当前目标树 |
/memory |
查看长期记忆 |
/flame |
查看意识火焰状态 |
/dag |
查看 DAG 摘要 |
/context |
查看当前上下文 |
/help |
查看所有命令 |
配置
全局配置 ~/.xiaomei-brain/config.json(OpenClaw 格式):
{
"models": {
"providers": {
"zhipu": { "baseUrl": "...", "apiKey": "...", "api": "openai-completions", "models": [...] },
"deepseek": { "baseUrl": "...", "apiKey": "...", "api": "openai-completions", "models": [...] }
}
},
"agents": {
"defaults": { "model": { "primary": "zhipu/glm-5.1" } },
"list": [{ "id": "xiaomei", "name": "小美", "model": { "primary": "zhipu/glm-5.1" } }]
},
"bindings": [{ "agentId": "xiaomei", "match": { "channel": "cli" } }],
"xiaomei_brain": {
"agent": { "max_steps": 10, "context": { "max_tokens": 4000, "recent_turns": 6 } },
"memory": { "similarity_threshold": 0.3, "embedding_model": "BAAI/bge-m3" }
}
}
Agent 专属文件 ~/.xiaomei-brain/<agent_id>/:
| 文件 | 说明 |
|---|---|
identity.md |
系统提示词,编辑即生效 |
config.yaml |
Drive 参数 + Consciousness 参数 |
渠道接入
from xiaomei_brain.plugins.channels import cli, feishu, dingtalk, p2p
# CLI
cli.ChannelCLI(agent).start()
# 飞书
feishu.ChannelFeishu(agent, app_id="...", app_secret="...").start()
# 钉钉
dingtalk.ChannelDingTalk(agent, client_id="...", client_secret="...").start()
测试
# 记忆系统
PYTHONPATH=src python3 examples/test_xiaomei_new.py
# 意识系统集成
PYTHONPATH=src python3 examples/test_conscious_living.py
# WebSocket 服务
PYTHONPATH=src python3 examples/ws_server.py
文档
完整文档见 docs/:
许可证
MIT — 详见 LICENSE
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file xiaomei_brain-0.1.0.tar.gz.
File metadata
- Download URL: xiaomei_brain-0.1.0.tar.gz
- Upload date:
- Size: 2.3 MB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.13.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
68de07405df373668ee2802443be63292be6ddc3640955f340c4941c6c78215e
|
|
| MD5 |
1d236d1c599cae25266f8e94a1f7e448
|
|
| BLAKE2b-256 |
82faeb8df0978cec156e5858c9cf53f416d0add1d34ef05aae6f6f68b5c4f0b4
|
File details
Details for the file xiaomei_brain-0.1.0-py3-none-any.whl.
File metadata
- Download URL: xiaomei_brain-0.1.0-py3-none-any.whl
- Upload date:
- Size: 928.6 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.13.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
7098173a6104ddae738d87e2dd805e4733f82a248f7678f43a29741ef8e72bab
|
|
| MD5 |
d7df9145758d5966128c050f9881f440
|
|
| BLAKE2b-256 |
5eb1d15cbd2fb252fbc727744623347cf71a2010d01e1b9800240df924945a27
|