OPC智脑开源！创业诊断AI Skills，一人创业者的避坑导航仪

用户输入 → 五维度评分 → 加权计算 → 阶段判定 → 硬性条件修正 → 矛盾检测 → 输出结果

import { classifyStage, getStageInfo } from 'opc-skills';

const result = classifyStage({
  demandValidation: 25,   // 有初步想法，但未验证
  solutionMaturity: 10,   // 没有产品
  complianceReadiness: 5, // 未注册
  userAcquisition: 0,     // 无用户
  scalabilityLevel: 0,    // 无规模化
});

console.log(result.stage);        // 'IDEA'
console.log(result.stageLabel);   // '构思期'
console.log(result.confidence);   // 0.72
console.log(result.recommendedSkill); // 'skill1-idea-feasibility'
console.log(result.hasContradiction); // false

// 获取阶段详细信息
const info = getStageInfo(result.stage);
console.log(info.goal);               // '验证需求真伪，确认个人匹配度'
console.log(info.graduationCriteria);  // '需求验证通过 + 个人匹配度≥60 + 存在可行赛道'

{
  "demandValidation": {
    "demandType": "rigid | improvement | fake",
    "demandTypeLabel": "刚需 | 改善型 | 伪需求",
    "reasoning": "判定理由",
    "evidenceForRigid": ["支持刚需的证据"],
    "evidenceAgainstRigid": ["反对刚需的证据"]
  },
  "personalMatch": {
    "skillMatch": 75,
    "timeMatch": 60,
    "budgetMatch": 80,
    "overallScore": 72
  },
  "feasibilityLevel": "feasible | challenging | infeasible",
  "summary": "可行性研判总结",
  "actionPlan": ["具体行动1", "具体行动2"]
}

skill1-idea-feasibility/
├── SKILL.md      # Skill实现（IDE使用，包含完整Prompt逻辑）
├── skill.json    # 元数据配置（ID、触发词、输入输出定义）
├── prompt.md     # Prompt文本（AI平台使用）
└── index.ts      # TypeScript代码（程序调用）

import { executeSkill1, checkSkill1Prerequisites } from 'opc-skills';

// 先检查前置条件
const check = checkSkill1Prerequisites({
  projectInfo: {
    projectName: '微团队CRM',
    description: '专为3-5人微团队的轻量CRM',
    industry: 'SaaS工具',
    targetUsers: '3-5人微团队创始人',
    valueProposition: '用大CRM 1/10的价格解决核心需求',
    founderBackground: '5年SaaS产品经理',
    budget: 10000,
    availableTime: 30,
  },
  demandDescription: '微团队用不起大CRM，用Excel又太乱',
  personalSkills: ['产品经理', '需求分析', '基础编程'],
  budget: 10000,
  availableTime: 30,
});

if (check.canExecute) {
  const result = executeSkill1(input);
  // result.prompt → 发送给任何大模型
  // result.data → 结构化数据
  // result.success → 是否执行成功
}

git clone https://gitee.com/zx_allen_li/opc_skills.git
cd opc_skills

bash install-codearts.sh /path/to/your-project

your-project/
├── AGENTS.md                              # 完整版配置
└── .codeartsdoer/
    ├── agents/
    │   └── opc-zhinao.json                # 自动生成的Agent配置
    └── skills/
        ├── skill1-idea-feasibility/SKILL.md
        ├── skill2-mvp-design/SKILL.md
        ├── skill3-opc-compliance/SKILL.md
        ├── skill4-seed-coldstart/SKILL.md
        ├── skill5-scale-growth/SKILL.md
        ├── feasibility-scoring/SKILL.md
        ├── report-export/SKILL.md
        └── user-feedback/SKILL.md

bash install-codebuddy.sh /path/to/your-project

your-project/
├── AGENTS.md
└── .codebuddy/
    ├── skills-registry.json          # 自动生成的Skill注册表
    └── skills/
        ├── skill1-idea-feasibility/
        │   └── SKILL.md
        ├── skill2-mvp-design/
        │   └── SKILL.md
        ├── skill3-opc-compliance/
        │   └── SKILL.md
        ├── skill4-seed-coldstart/
        │   └── SKILL.md
        ├── skill5-scale-growth/
        │   └── SKILL.md
        ├── feasibility-scoring/
        │   └── SKILL.md
        ├── report-export/
        │   └── SKILL.md
        └── user-feedback/
            └── SKILL.md

# 安装到指定项目
bash install-codebuddy.sh /path/to/your-project

# 安装到当前目录
bash install-codebuddy.sh

# 跳过备份直接覆盖
bash install-codebuddy.sh /path/to/your-project --skip-backup

bash install-prompt.sh /path/to/your-project

your-project/
├── AGENTS.md
├── .cursorrules              # 自动从AGENTS.md生成
└── skills/
    ├── skill1-idea-feasibility/SKILL.md
    └── ...

your-project/
├── AGENTS.md
├── .github/
│   └── copilot-instructions.md   # 自动从AGENTS.md生成
└── skills/
    ├── skill1-idea-feasibility/SKILL.md
    └── ...

# 码道IDE
cat .codeartsdoer/agents/opc-zhinao.json
# 应该看到 "instructions": "AGENTS.md"

# CodeBuddy
cat .codebuddy/skills-registry.json
# 应该看到8个skill名称

# 通用检查skills数量（码道IDE）
ls -d .codeartsdoer/skills/*/ | wc -l
# 通用检查skills数量（CodeBuddy）
ls -d .codebuddy/skills/*/ | wc -l
# 应该输出：8

npm install opc-skills

import {
  // 阶段判定
  classifyStage,
  detectContradiction,
  getStageInfo,
  getAllStageInfo,

  // Skill执行
  executeSkill1,
  executeSkill2,
  executeSkill3,
  executeSkill4,
  executeSkill5,

  // 前置条件检查
  checkSkill1Prerequisites,
  checkSkill2Prerequisites,
  checkSkill3Prerequisites,
  checkSkill4Prerequisites,
  checkSkill5Prerequisites,

  // 输出Schema
  skill1OutputSchema,
  skill2OutputSchema,
  skill3OutputSchema,
  skill4OutputSchema,
  skill5OutputSchema,

  // 校验引擎
  validateAIOutput,
  checkUnderstanding,

  // 类型
  StageEnum,
  type BusinessDimensions,
  type Skill1Input,
  type SkillResult,
} from 'opc-skills';

// 1. 判定阶段
const stageResult = classifyStage({
  demandValidation: 30,
  solutionMaturity: 20,
  complianceReadiness: 10,
  userAcquisition: 5,
  scalabilityLevel: 0,
});
// → 构思期，推荐Skill1

// 2. 检查前置条件
const check = checkSkill1Prerequisites({
  projectInfo: { projectName: '...', description: '...', ... },
  demandDescription: '...',
  personalSkills: ['...'],
  budget: 10000,
  availableTime: 30,
});

// 3. 执行Skill，生成Prompt
if (check.canExecute) {
  const skillResult = executeSkill1(input);

  // 方式A：把Prompt发送给任何大模型
  console.log(skillResult.prompt);

  // 方式B：获取结构化数据
  console.log(skillResult.data);
}

// 4. 校验AI输出
const validation = validateAIOutput('skill1-idea-feasibility', aiOutput);
if (!validation.isValid) {
  console.log(validation.issues); // 查看校验问题
}

import OpenAI from 'openai';
import { executeSkill1, skill1OutputSchema } from 'opc-skills';

const openai = new OpenAI({ apiKey: 'your-api-key' });

// 方式1：直接使用Prompt
const skillResult = executeSkill1(input);
const response = await openai.chat.completions.create({
  model: 'gpt-4o',
  messages: [
    { role: 'system', content: '你是一位OPC创业诊断专家' },
    { role: 'user', content: skillResult.prompt },
  ],
});

// 方式2：使用Function Calling实现结构化输出
const response2 = await openai.chat.completions.create({
  model: 'gpt-4o',
  messages: [
    { role: 'system', content: '你是一位OPC创业诊断专家' },
    { role: 'user', content: skillResult.prompt },
  ],
  functions: [{
    name: 'feasibility_report',
    description: '输出OPC项目可行性分析报告',
    parameters: skill1OutputSchema,  // 直接使用opc-skills提供的Schema
  }],
  function_call: { name: 'feasibility_report' },
});

const report = JSON.parse(response2.choices[0].message.function_call.arguments);

def main(demand_validation, solution_maturity, compliance_readiness, user_acquisition, scalability_level):
    score = (demand_validation * 0.30 + solution_maturity * 0.25 
           + compliance_readiness * 0.15 + user_acquisition * 0.15 
           + scalability_level * 0.15)

    if score <= 20:   return {'stage': 'IDEA', 'skill': 'skill1-idea-feasibility'}
    elif score <= 40: return {'stage': 'MVP', 'skill': 'skill2-mvp-design'}
    elif score <= 60: return {'stage': 'ENTITY', 'skill': 'skill3-opc-compliance'}
    elif score <= 80: return {'stage': 'VALIDATION', 'skill': 'skill4-seed-coldstart'}
    else:             return {'stage': 'SCALE', 'skill': 'skill5-scale-growth'}

npx install-opc /path/to/your-project

你：我想做一个面向程序员的AI代码审查工具
OPC智脑：自动判定为构思期，触发Skill1
→ 输出三维评分 + 可行性等级 + 可执行建议

你：需求验证通过了，帮我设计MVP
OPC智脑：自动判定为原型期，触发Skill2
→ 输出功能裁剪建议 + 三层产品体系 + 交付成本测算

你：我做了一个小程序，给几个朋友用过，但还没注册公司也没有付费用户，我卡在哪了？
OPC智脑：自动采集信息→五维度评分→判定原型期→检测矛盾→输出卡点分析
→ 当前诊断 + 整改方案 + 前置布局