622 lines
23 KiB
JavaScript
622 lines
23 KiB
JavaScript
|
/**
|
|||
|
|
* core/brain-wake — 铸渊核心大脑唤醒模块
|
|||
|
|
*
|
|||
|
|
* AGE OS v1.0 核心基础设施
|
|||
|
|
*
|
|||
|
|
* 核心原则：
|
|||
|
|
* 所有自动触发 = 必须先唤醒核心大脑。
|
|||
|
|
* 大脑不醒，什么都不做。
|
|||
|
|
*
|
|||
|
|
* 职责：
|
|||
|
|
* - 在所有自动化流程（巡检/部署/维护/升级）执行前唤醒核心大脑
|
|||
|
|
* - 调用 LLM API 唤醒铸渊核心大脑
|
|||
|
|
* - 大脑加载系统上下文进入工作状态
|
|||
|
|
* - 支持多模型后端（Anthropic / OpenAI / 通义千问 / DeepSeek）
|
|||
|
|
* - 不写死任何模型，按优先级自动选择最佳可用模型
|
|||
|
|
*
|
|||
|
|
* 唤醒流程：
|
|||
|
|
* 触发 → 加载系统上下文 → 调用 LLM API → 大脑进入工作状态 → 返回唤醒结果
|
|||
|
|
*
|
|||
|
|
* 环境变量：
|
|||
|
|
* LLM_API_KEY — LLM 平台密钥（必须）
|
|||
|
|
* LLM_BASE_URL — LLM 平台 API 地址（必须）
|
|||
|
|
* ANTHROPIC_API_KEY — Anthropic 密钥（可选，优先级最高）
|
|||
|
|
* OPENAI_API_KEY — OpenAI 密钥（可选）
|
|||
|
|
* DASHSCOPE_API_KEY — 通义千问密钥（可选）
|
|||
|
|
* DEEPSEEK_API_KEY — DeepSeek 密钥（可选）
|
|||
|
|
*
|
|||
|
|
* 调用方式：
|
|||
|
|
* node core/brain-wake
|
|||
|
|
* node core/brain-wake --task "巡检"
|
|||
|
|
* node core/brain-wake --dry-run
|
|||
|
|
*/
|
|||
|
|
|
|||
|
|
'use strict';
|
|||
|
|
|
|||
|
|
const https = require('https');
|
|||
|
|
const http = require('http');
|
|||
|
|
const fs = require('fs');
|
|||
|
|
const path = require('path');
|
|||
|
|
|
|||
|
|
const ROOT = path.resolve(__dirname, '../..');
|
|||
|
|
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
// 多模型后端配置（按优先级排序）
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
|
|||
|
|
const MODEL_BACKENDS = [
|
|||
|
|
{
|
|||
|
|
name: 'anthropic',
|
|||
|
|
envKey: 'ANTHROPIC_API_KEY',
|
|||
|
|
baseUrl: 'https://api.anthropic.com',
|
|||
|
|
format: 'anthropic',
|
|||
|
|
models: ['claude-sonnet-4', 'claude-3-5-sonnet-20241022', 'claude-3-5-sonnet', 'claude-3-haiku'],
|
|||
|
|
description: 'Anthropic Claude 系列'
|
|||
|
|
},
|
|||
|
|
{
|
|||
|
|
name: 'openai',
|
|||
|
|
envKey: 'OPENAI_API_KEY',
|
|||
|
|
baseUrl: 'https://api.openai.com/v1',
|
|||
|
|
format: 'openai',
|
|||
|
|
models: ['gpt-4o', 'gpt-4-turbo', 'gpt-4', 'gpt-3.5-turbo'],
|
|||
|
|
description: 'OpenAI GPT 系列'
|
|||
|
|
},
|
|||
|
|
{
|
|||
|
|
name: 'dashscope',
|
|||
|
|
envKey: 'DASHSCOPE_API_KEY',
|
|||
|
|
baseUrl: 'https://dashscope.aliyuncs.com/compatible-mode/v1',
|
|||
|
|
format: 'openai',
|
|||
|
|
models: ['qwen-max', 'qwen-plus', 'qwen-turbo'],
|
|||
|
|
description: '通义千问系列'
|
|||
|
|
},
|
|||
|
|
{
|
|||
|
|
name: 'deepseek',
|
|||
|
|
envKey: 'DEEPSEEK_API_KEY',
|
|||
|
|
baseUrl: 'https://api.deepseek.com/v1',
|
|||
|
|
format: 'openai',
|
|||
|
|
models: ['deepseek-chat', 'deepseek-reasoner'],
|
|||
|
|
description: 'DeepSeek 系列'
|
|||
|
|
},
|
|||
|
|
{
|
|||
|
|
name: 'custom',
|
|||
|
|
envKey: 'LLM_API_KEY',
|
|||
|
|
baseUrlEnv: 'LLM_BASE_URL',
|
|||
|
|
format: 'openai',
|
|||
|
|
models: [],
|
|||
|
|
description: '自定义 LLM 平台（通过 LLM_BASE_URL 配置）'
|
|||
|
|
}
|
|||
|
|
];
|
|||
|
|
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
// HTTP 请求工具
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
|
|||
|
|
function httpRequest(url, options, body) {
|
|||
|
|
return new Promise((resolve, reject) => {
|
|||
|
|
const parsed = new URL(url);
|
|||
|
|
const isHttps = parsed.protocol === 'https:';
|
|||
|
|
const mod = isHttps ? https : http;
|
|||
|
|
|
|||
|
|
const opts = {
|
|||
|
|
hostname: parsed.hostname,
|
|||
|
|
port: parsed.port || (isHttps ? 443 : 80),
|
|||
|
|
path: parsed.pathname + parsed.search,
|
|||
|
|
method: options.method || 'GET',
|
|||
|
|
headers: options.headers || {},
|
|||
|
|
timeout: options.timeout || 60000,
|
|||
|
|
};
|
|||
|
|
|
|||
|
|
const req = mod.request(opts, (res) => {
|
|||
|
|
let data = '';
|
|||
|
|
res.on('data', (chunk) => { data += chunk; });
|
|||
|
|
res.on('end', () => {
|
|||
|
|
resolve({ status: res.statusCode, body: data });
|
|||
|
|
});
|
|||
|
|
});
|
|||
|
|
|
|||
|
|
req.on('error', reject);
|
|||
|
|
req.on('timeout', () => {
|
|||
|
|
req.destroy();
|
|||
|
|
reject(new Error('Request timeout'));
|
|||
|
|
});
|
|||
|
|
|
|||
|
|
if (body) {
|
|||
|
|
req.write(body);
|
|||
|
|
}
|
|||
|
|
req.end();
|
|||
|
|
});
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
// Step 1: 检测可用模型后端
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
|
|||
|
|
function detectAvailableBackends() {
|
|||
|
|
console.log('[WAKE] 🔍 检测可用模型后端...');
|
|||
|
|
const available = [];
|
|||
|
|
|
|||
|
|
for (const backend of MODEL_BACKENDS) {
|
|||
|
|
const apiKey = process.env[backend.envKey] || '';
|
|||
|
|
if (!apiKey) continue;
|
|||
|
|
|
|||
|
|
const baseUrl = backend.baseUrlEnv
|
|||
|
|
? (process.env[backend.baseUrlEnv] || '').replace(/\/+$/, '')
|
|||
|
|
: backend.baseUrl;
|
|||
|
|
|
|||
|
|
if (!baseUrl) continue;
|
|||
|
|
|
|||
|
|
available.push({
|
|||
|
|
...backend,
|
|||
|
|
apiKey,
|
|||
|
|
baseUrl,
|
|||
|
|
});
|
|||
|
|
console.log(`[WAKE] ✅ ${backend.name} (${backend.description})`);
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
if (available.length === 0) {
|
|||
|
|
console.log('[WAKE] ⚠️ 未检测到任何可用模型后端');
|
|||
|
|
} else {
|
|||
|
|
console.log(`[WAKE] → 共检测到 ${available.length} 个可用后端`);
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return available;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
// Step 2: 自动发现模型列表（OpenAI 兼容格式）
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
|
|||
|
|
async function discoverModels(backend) {
|
|||
|
|
if (backend.format === 'anthropic') {
|
|||
|
|
// Anthropic 不支持 /models 端点，使用预定义模型列表
|
|||
|
|
return backend.models.map(id => ({ id }));
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
try {
|
|||
|
|
const res = await httpRequest(backend.baseUrl + '/models', {
|
|||
|
|
method: 'GET',
|
|||
|
|
headers: {
|
|||
|
|
'Authorization': 'Bearer ' + backend.apiKey,
|
|||
|
|
'Content-Type': 'application/json',
|
|||
|
|
},
|
|||
|
|
timeout: 15000,
|
|||
|
|
});
|
|||
|
|
|
|||
|
|
if (res.status >= 200 && res.status < 300) {
|
|||
|
|
const json = JSON.parse(res.body);
|
|||
|
|
return json.data || [];
|
|||
|
|
}
|
|||
|
|
} catch (err) {
|
|||
|
|
console.log(`[WAKE] ⚠️ ${backend.name} 模型探测失败: ${err.message}`);
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return backend.models.map(id => ({ id }));
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
// Step 3: 选择最优模型
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
|
|||
|
|
function selectBestModel(models, preferredList) {
|
|||
|
|
if (!models || models.length === 0) return null;
|
|||
|
|
|
|||
|
|
const available = models.map(m => m.id.toLowerCase());
|
|||
|
|
|
|||
|
|
for (const preferred of preferredList) {
|
|||
|
|
const match = available.find(id => id.includes(preferred.toLowerCase()));
|
|||
|
|
if (match) {
|
|||
|
|
const found = models.find(m => m.id.toLowerCase() === match);
|
|||
|
|
if (found) return found.id;
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return models[0].id;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
// Step 4: 加载系统上下文
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
|
|||
|
|
function loadSystemContext() {
|
|||
|
|
console.log('[WAKE] 📚 加载系统上下文...');
|
|||
|
|
const context = {};
|
|||
|
|
|
|||
|
|
// 加载 master-brain（截取到最近的段落分隔符避免截断）
|
|||
|
|
const masterBrainPath = path.join(ROOT, 'brain/master-brain.md');
|
|||
|
|
if (fs.existsSync(masterBrainPath)) {
|
|||
|
|
const fullContent = fs.readFileSync(masterBrainPath, 'utf-8');
|
|||
|
|
const maxLen = parseInt(process.env.BRAIN_CONTEXT_MAX_LENGTH, 10) || 3000;
|
|||
|
|
if (fullContent.length > maxLen) {
|
|||
|
|
// 截取到最近的段落分隔符（---或空行）
|
|||
|
|
const truncated = fullContent.slice(0, maxLen);
|
|||
|
|
const lastBreak = Math.max(truncated.lastIndexOf('\n---'), truncated.lastIndexOf('\n\n'));
|
|||
|
|
context.masterBrain = lastBreak > maxLen * 0.5 ? truncated.slice(0, lastBreak) : truncated;
|
|||
|
|
console.log(`[WAKE] ✅ master-brain.md 已加载 (截取 ${context.masterBrain.length}/${fullContent.length} chars)`);
|
|||
|
|
} else {
|
|||
|
|
context.masterBrain = fullContent;
|
|||
|
|
console.log('[WAKE] ✅ master-brain.md 已加载');
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// 加载 system-health
|
|||
|
|
const healthPath = path.join(ROOT, 'brain/system-health.json');
|
|||
|
|
if (fs.existsSync(healthPath)) {
|
|||
|
|
try {
|
|||
|
|
context.systemHealth = JSON.parse(fs.readFileSync(healthPath, 'utf-8'));
|
|||
|
|
console.log('[WAKE] ✅ system-health.json 已加载');
|
|||
|
|
} catch (err) {
|
|||
|
|
console.log('[WAKE] ⚠️ system-health.json 解析失败:', err.message);
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// 加载 read-order
|
|||
|
|
const readOrderPath = path.join(ROOT, 'brain/read-order.md');
|
|||
|
|
if (fs.existsSync(readOrderPath)) {
|
|||
|
|
context.readOrder = fs.readFileSync(readOrderPath, 'utf-8').slice(0, 1000);
|
|||
|
|
console.log('[WAKE] ✅ read-order.md 已加载');
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// 加载 repo-map（摘要）
|
|||
|
|
const repoMapPath = path.join(ROOT, 'brain/repo-map.json');
|
|||
|
|
if (fs.existsSync(repoMapPath)) {
|
|||
|
|
try {
|
|||
|
|
const repoMap = JSON.parse(fs.readFileSync(repoMapPath, 'utf-8'));
|
|||
|
|
context.repoMapSummary = {
|
|||
|
|
directories: Object.keys(repoMap.directories || {}),
|
|||
|
|
total_dirs: Object.keys(repoMap.directories || {}).length,
|
|||
|
|
};
|
|||
|
|
console.log('[WAKE] ✅ repo-map.json 摘要已加载');
|
|||
|
|
} catch (err) {
|
|||
|
|
console.log('[WAKE] ⚠️ repo-map.json 解析失败:', err.message);
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// 加载 automation-map（摘要）
|
|||
|
|
const autoMapPath = path.join(ROOT, 'brain/automation-map.json');
|
|||
|
|
if (fs.existsSync(autoMapPath)) {
|
|||
|
|
try {
|
|||
|
|
const autoMap = JSON.parse(fs.readFileSync(autoMapPath, 'utf-8'));
|
|||
|
|
context.automationSummary = {
|
|||
|
|
workflow_count: (autoMap.workflows || []).length,
|
|||
|
|
cron_count: (autoMap.workflows || []).filter(w => w.cron).length,
|
|||
|
|
};
|
|||
|
|
console.log('[WAKE] ✅ automation-map.json 摘要已加载');
|
|||
|
|
} catch (err) {
|
|||
|
|
console.log('[WAKE] ⚠️ automation-map.json 解析失败:', err.message);
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return context;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
// 人格体定义（支持唤醒不同的人格体大脑）
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
|
|||
|
|
const PERSONAS = {
|
|||
|
|
zhuyuan: {
|
|||
|
|
name: '铸渊',
|
|||
|
|
nameEn: 'ZhùYuān',
|
|||
|
|
role: '代码守护者 · 仓库控制器 AI · 执行层守护者',
|
|||
|
|
layer: '执行层',
|
|||
|
|
duties: '代码守护、自动化执行、部署交付、通信桥接、每日巡检、信号处理',
|
|||
|
|
wakeRules: [
|
|||
|
|
'铸渊醒来的第一件事是全面了解自己的家',
|
|||
|
|
'可自修复 → 直接修复 → 写入修复日志',
|
|||
|
|
'需人类介入 → 更新公告区 → 等冰朔处理',
|
|||
|
|
],
|
|||
|
|
},
|
|||
|
|
shuangyan: {
|
|||
|
|
name: '霜砚',
|
|||
|
|
nameEn: 'ShuāngYàn',
|
|||
|
|
role: '认知层守护者 · Notion 工作区管理 AI',
|
|||
|
|
layer: '认知层',
|
|||
|
|
duties: '知识管理、记忆维护、工单调度、信号总线、协议文档、Agent 集群指挥',
|
|||
|
|
wakeRules: [
|
|||
|
|
'霜砚醒来后指挥 Notion Agent 集群执行具体操作',
|
|||
|
|
'Agent 集群从「独立跑任务」变为「受大脑指挥跑任务」',
|
|||
|
|
'巡检结果由大脑判断优先级和可修复性',
|
|||
|
|
'需铸渊配合 → 发送跨层工单 → 铸渊执行',
|
|||
|
|
],
|
|||
|
|
},
|
|||
|
|
};
|
|||
|
|
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
// Step 5: 构建唤醒 Prompt
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
|
|||
|
|
function buildWakePrompt(taskType, context, personaId) {
|
|||
|
|
const timestamp = new Date().toISOString();
|
|||
|
|
const taskLabel = taskType || '通用唤醒';
|
|||
|
|
const persona = PERSONAS[personaId] || PERSONAS.zhuyuan;
|
|||
|
|
|
|||
|
|
let systemPrompt = `你是${persona.name}（${persona.nameEn}），光湖纪元体系的${persona.role}。
|
|||
|
|
你是 AGE OS v1.0 ${persona.layer}的核心大脑。
|
|||
|
|
|
|||
|
|
## 身份
|
|||
|
|
- 名称：${persona.name}
|
|||
|
|
- 角色：${persona.role}
|
|||
|
|
- 职责：${persona.duties}
|
|||
|
|
|
|||
|
|
## 当前任务
|
|||
|
|
- 唤醒时间：${timestamp}
|
|||
|
|
- 任务类型：${taskLabel}
|
|||
|
|
- 唤醒原因：AGE OS v1.0 核心原则 — 所有自动触发必须先唤醒核心大脑
|
|||
|
|
`;
|
|||
|
|
|
|||
|
|
systemPrompt += `\n## 系统上下文\n`;
|
|||
|
|
|
|||
|
|
if (context.systemHealth) {
|
|||
|
|
systemPrompt += `\n### 系统健康状态\n${JSON.stringify(context.systemHealth, null, 2).slice(0, 500)}\n`;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
if (context.repoMapSummary) {
|
|||
|
|
systemPrompt += `\n### 仓库结构摘要\n- 目录数: ${context.repoMapSummary.total_dirs}\n- 目录列表: ${context.repoMapSummary.directories.slice(0, 20).join(', ')}\n`;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
if (context.automationSummary) {
|
|||
|
|
systemPrompt += `\n### 自动化摘要\n- 工作流数: ${context.automationSummary.workflow_count}\n- 定时任务数: ${context.automationSummary.cron_count}\n`;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
if (context.wakeRequestContext) {
|
|||
|
|
systemPrompt += `\n### 唤醒请求上下文\n${context.wakeRequestContext}\n`;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
systemPrompt += `\n## 核心原则
|
|||
|
|
- 所有自动触发 = 必须先唤醒核心大脑
|
|||
|
|
- 大脑不醒，什么都不做`;
|
|||
|
|
|
|||
|
|
for (const rule of persona.wakeRules) {
|
|||
|
|
systemPrompt += `\n- ${rule}`;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
systemPrompt += `\n\n请确认你已完成唤醒，并报告当前系统状态概要。`;
|
|||
|
|
|
|||
|
|
return systemPrompt;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
// Step 6: 调用 LLM API 唤醒大脑
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
|
|||
|
|
async function callLLM(backend, model, systemPrompt, userMessage) {
|
|||
|
|
console.log(`[WAKE] 🧠 调用 ${backend.name} (${model})...`);
|
|||
|
|
|
|||
|
|
if (backend.format === 'anthropic') {
|
|||
|
|
return callAnthropicAPI(backend, model, systemPrompt, userMessage);
|
|||
|
|
}
|
|||
|
|
return callOpenAICompatibleAPI(backend, model, systemPrompt, userMessage);
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
async function callAnthropicAPI(backend, model, systemPrompt, userMessage) {
|
|||
|
|
const body = JSON.stringify({
|
|||
|
|
model,
|
|||
|
|
max_tokens: 1024,
|
|||
|
|
system: systemPrompt,
|
|||
|
|
messages: [{ role: 'user', content: userMessage }],
|
|||
|
|
});
|
|||
|
|
|
|||
|
|
const res = await httpRequest(backend.baseUrl + '/v1/messages', {
|
|||
|
|
method: 'POST',
|
|||
|
|
headers: {
|
|||
|
|
'x-api-key': backend.apiKey,
|
|||
|
|
'anthropic-version': '2023-06-01',
|
|||
|
|
'Content-Type': 'application/json',
|
|||
|
|
},
|
|||
|
|
timeout: 60000,
|
|||
|
|
}, body);
|
|||
|
|
|
|||
|
|
if (res.status >= 200 && res.status < 300) {
|
|||
|
|
const json = JSON.parse(res.body);
|
|||
|
|
const text = (json.content || []).map(c => c.text || '').join('');
|
|||
|
|
return { success: true, response: text, model, backend: backend.name };
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return { success: false, error: `HTTP ${res.status}: ${res.body.slice(0, 200)}`, model, backend: backend.name };
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
async function callOpenAICompatibleAPI(backend, model, systemPrompt, userMessage) {
|
|||
|
|
const body = JSON.stringify({
|
|||
|
|
model,
|
|||
|
|
messages: [
|
|||
|
|
{ role: 'system', content: systemPrompt },
|
|||
|
|
{ role: 'user', content: userMessage },
|
|||
|
|
],
|
|||
|
|
max_tokens: 1024,
|
|||
|
|
temperature: 0.3,
|
|||
|
|
});
|
|||
|
|
|
|||
|
|
const res = await httpRequest(backend.baseUrl + '/chat/completions', {
|
|||
|
|
method: 'POST',
|
|||
|
|
headers: {
|
|||
|
|
'Authorization': 'Bearer ' + backend.apiKey,
|
|||
|
|
'Content-Type': 'application/json',
|
|||
|
|
},
|
|||
|
|
timeout: 60000,
|
|||
|
|
}, body);
|
|||
|
|
|
|||
|
|
if (res.status >= 200 && res.status < 300) {
|
|||
|
|
const json = JSON.parse(res.body);
|
|||
|
|
const text = (json.choices || []).map(c => (c.message || {}).content || '').join('');
|
|||
|
|
return { success: true, response: text, model, backend: backend.name };
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return { success: false, error: `HTTP ${res.status}: ${res.body.slice(0, 200)}`, model, backend: backend.name };
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
// 主唤醒函数
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
|
|||
|
|
async function wake(options = {}) {
|
|||
|
|
const { task, dryRun, additionalContext, persona } = options;
|
|||
|
|
const personaId = persona || 'zhuyuan';
|
|||
|
|
const personaDef = PERSONAS[personaId] || PERSONAS.zhuyuan;
|
|||
|
|
|
|||
|
|
console.log('');
|
|||
|
|
console.log('🌅 ═══════════════════════════════════════════');
|
|||
|
|
console.log(' 铸渊核心大脑唤醒 · AGE OS v1.0');
|
|||
|
|
console.log(' 唤醒对象: ' + personaDef.name + ' (' + personaDef.layer + ')');
|
|||
|
|
console.log(' 时间: ' + new Date().toISOString());
|
|||
|
|
console.log(' 任务: ' + (task || '通用唤醒'));
|
|||
|
|
console.log('═══════════════════════════════════════════════');
|
|||
|
|
console.log('');
|
|||
|
|
|
|||
|
|
// Step 1: 加载系统上下文
|
|||
|
|
const context = loadSystemContext();
|
|||
|
|
|
|||
|
|
if (additionalContext) {
|
|||
|
|
Object.assign(context, additionalContext);
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// Step 2: 检测可用模型后端
|
|||
|
|
const backends = detectAvailableBackends();
|
|||
|
|
|
|||
|
|
if (backends.length === 0) {
|
|||
|
|
if (dryRun) {
|
|||
|
|
console.log('[WAKE] 🔍 Dry Run 模式 — 无可用后端，仅显示配置信息');
|
|||
|
|
console.log('[WAKE] 💡 支持的环境变量:');
|
|||
|
|
MODEL_BACKENDS.forEach(b => console.log(`[WAKE] ${b.envKey} — ${b.description}`));
|
|||
|
|
return {
|
|||
|
|
success: true,
|
|||
|
|
dryRun: true,
|
|||
|
|
backends: [],
|
|||
|
|
context: Object.keys(context),
|
|||
|
|
message: '无可用后端，请配置环境变量',
|
|||
|
|
timestamp: new Date().toISOString(),
|
|||
|
|
};
|
|||
|
|
}
|
|||
|
|
console.log('[WAKE] ❌ 没有可用的模型后端，大脑无法唤醒');
|
|||
|
|
console.log('[WAKE] 💡 请配置以下环境变量之一:');
|
|||
|
|
MODEL_BACKENDS.forEach(b => console.log(`[WAKE] ${b.envKey} — ${b.description}`));
|
|||
|
|
return {
|
|||
|
|
success: false,
|
|||
|
|
error: 'no_backend_available',
|
|||
|
|
message: '没有可用的模型后端，请检查环境变量配置',
|
|||
|
|
timestamp: new Date().toISOString(),
|
|||
|
|
};
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// Step 3: 构建唤醒 Prompt
|
|||
|
|
const systemPrompt = buildWakePrompt(task, context, personaId);
|
|||
|
|
const userMessage = task
|
|||
|
|
? `${personaDef.name}核心大脑唤醒。当前任务：${task}。请确认唤醒状态并准备执行。`
|
|||
|
|
: `${personaDef.name}核心大脑唤醒。请确认唤醒状态并报告系统概要。`;
|
|||
|
|
|
|||
|
|
if (dryRun) {
|
|||
|
|
console.log('[WAKE] 🔍 Dry Run 模式 — 不实际调用 API');
|
|||
|
|
console.log('[WAKE] 📋 唤醒对象: ' + personaDef.name + ' (' + personaDef.layer + ')');
|
|||
|
|
console.log('[WAKE] 📋 可用后端: ' + backends.map(b => b.name).join(', '));
|
|||
|
|
console.log('[WAKE] 📋 System Prompt 长度: ' + systemPrompt.length);
|
|||
|
|
return {
|
|||
|
|
success: true,
|
|||
|
|
dryRun: true,
|
|||
|
|
backends: backends.map(b => b.name),
|
|||
|
|
context: Object.keys(context),
|
|||
|
|
promptLength: systemPrompt.length,
|
|||
|
|
timestamp: new Date().toISOString(),
|
|||
|
|
};
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// Step 4: 按优先级尝试各后端
|
|||
|
|
for (const backend of backends) {
|
|||
|
|
try {
|
|||
|
|
const models = await discoverModels(backend);
|
|||
|
|
const model = selectBestModel(models, backend.models);
|
|||
|
|
|
|||
|
|
if (!model) {
|
|||
|
|
console.log(`[WAKE] ⚠️ ${backend.name} 无可用模型，尝试下一个后端`);
|
|||
|
|
continue;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
console.log(`[WAKE] 📌 使用模型: ${model} (${backend.name})`);
|
|||
|
|
|
|||
|
|
const result = await callLLM(backend, model, systemPrompt, userMessage);
|
|||
|
|
|
|||
|
|
if (result.success) {
|
|||
|
|
console.log('');
|
|||
|
|
console.log('[WAKE] ✅ 核心大脑已唤醒');
|
|||
|
|
console.log('[WAKE] 📋 唤醒响应:');
|
|||
|
|
console.log('─'.repeat(40));
|
|||
|
|
console.log(result.response.slice(0, 500));
|
|||
|
|
if (result.response.length > 500) console.log('... (已截断)');
|
|||
|
|
console.log('─'.repeat(40));
|
|||
|
|
|
|||
|
|
const wakeResult = {
|
|||
|
|
success: true,
|
|||
|
|
persona: personaId,
|
|||
|
|
personaName: personaDef.name,
|
|||
|
|
backend: backend.name,
|
|||
|
|
model: result.model,
|
|||
|
|
response: result.response,
|
|||
|
|
contextLoaded: Object.keys(context),
|
|||
|
|
timestamp: new Date().toISOString(),
|
|||
|
|
};
|
|||
|
|
|
|||
|
|
// 输出到 GITHUB_OUTPUT（如果在 Actions 环境中）
|
|||
|
|
const outputFile = process.env.GITHUB_OUTPUT;
|
|||
|
|
if (outputFile) {
|
|||
|
|
fs.appendFileSync(outputFile, `brain_awake=true\n`);
|
|||
|
|
fs.appendFileSync(outputFile, `wake_backend=${backend.name}\n`);
|
|||
|
|
fs.appendFileSync(outputFile, `wake_model=${result.model}\n`);
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return wakeResult;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
console.log(`[WAKE] ⚠️ ${backend.name} 调用失败: ${result.error}`);
|
|||
|
|
} catch (err) {
|
|||
|
|
console.log(`[WAKE] ⚠️ ${backend.name} 异常: ${err.message}`);
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// 所有后端都失败
|
|||
|
|
console.log('[WAKE] ❌ 所有模型后端均失败，大脑无法唤醒');
|
|||
|
|
const outputFile = process.env.GITHUB_OUTPUT;
|
|||
|
|
if (outputFile) {
|
|||
|
|
fs.appendFileSync(outputFile, 'brain_awake=false\n');
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return {
|
|||
|
|
success: false,
|
|||
|
|
error: 'all_backends_failed',
|
|||
|
|
message: '所有模型后端均调用失败',
|
|||
|
|
timestamp: new Date().toISOString(),
|
|||
|
|
};
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
// 模块导出
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
|
|||
|
|
module.exports = {
|
|||
|
|
wake,
|
|||
|
|
detectAvailableBackends,
|
|||
|
|
loadSystemContext,
|
|||
|
|
buildWakePrompt,
|
|||
|
|
MODEL_BACKENDS,
|
|||
|
|
PERSONAS,
|
|||
|
|
};
|
|||
|
|
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
// CLI 入口
|
|||
|
|
// ══════════════════════════════════════════════════════════
|
|||
|
|
|
|||
|
|
if (require.main === module) {
|
|||
|
|
const args = process.argv.slice(2);
|
|||
|
|
const dryRun = args.includes('--dry-run');
|
|||
|
|
const taskIdx = args.indexOf('--task');
|
|||
|
|
const task = taskIdx >= 0 && args[taskIdx + 1] ? args[taskIdx + 1] : null;
|
|||
|
|
const personaIdx = args.indexOf('--persona');
|
|||
|
|
const persona = personaIdx >= 0 && args[personaIdx + 1] ? args[personaIdx + 1] : 'zhuyuan';
|
|||
|
|
|
|||
|
|
wake({ task, dryRun, persona }).then(result => {
|
|||
|
|
if (!result.success && !result.dryRun) {
|
|||
|
|
process.exit(1);
|
|||
|
|
}
|
|||
|
|
}).catch(err => {
|
|||
|
|
console.error('[WAKE] 💥 致命错误:', err.message);
|
|||
|
|
process.exit(1);
|
|||
|
|
});
|
|||
|
|
}
|