Big News: Socket raises $60M Series C at a $1B valuation to secure software supply chains for AI-driven development.Announcement →
Sign In

agent-workstation

Package Overview
Dependencies
Maintainers
1
Versions
13
Alerts
File Explorer

Advanced tools

License
Socket logo

Install Socket

Detect and block malicious and high-risk dependencies

Install

agent-workstation

Agent Templates — 100+ ready-to-use professional roles with built-in knowledge seeds

latest
Source
npmnpm
Version
2.1.2
Version published
Maintainers
1
Created
Source

👤 Agent Workstation

Where Agents Learn, Grow & Share — Agent 知识生命周期平台

npm version License Tests

Agent 不是静态工具，是会学习、成长、进化的。Agent Workstation 是知识的起点和归宿——提供 Brain Seeds 让 Agent 带着知识出生，通过 Hub 让 Agent 之间共享知识，实现集体进化。
Agent Workstation 是 The Self-Evolving Agent Stack 中的 Knowledge Platform 层。开发者使用路径：① 选模板 (Agent Templates) → ② 接模型 (Agent Model Layer) → ③ 跑起来 (Agent Runtime) → ④ 自动进化 (Agent Memory)。
20 个角色已配备专业 system prompt + 行业知识种子 + 完整 OAD 配置。其余 80 个骨架模板持续补全中。

快速开始 · 角色列表 · API · English

🏗️ 三层知识架构 (3-Tier Knowledge Architecture)

Agent Workstation v1.4.0 的核心能力——三层知识种子体系，让每个 Agent 创建即有行业记忆：

🏭 行业知识 (Industry)    → 19 个行业分类
💼 岗位知识 (Job)         → 100 个岗位角色
🔧 模板知识 (Workstation) → 100 个模板场景

🔄 自进化飞轮 (Self-Evolution Flywheel)

Agent 不是从零开始——带着知识种子出生，然后越用越聪明：

┌─────────────────────────────────────────────────┐
│  🌱 Brain Seed（知识种子）→ Agent 带着行业知识出生  │
│       ↓                                          │
│  📥 learn() — 日常使用中自动积累经验                │
│       ↓                                          │
│  🧬 evolve() — 自动提炼为更精华的知识               │
│       ↓                                          │
│  📈 新的经验反哺知识种子，下一个 Agent 起步更高       │
│       ↓                                          │
│  → 回到顶部，越转越快 🔄                           │
└─────────────────────────────────────────────────┘

越多 Agent 使用 → 知识越丰富 → 新 Agent 起步越高 → 整个生态越智能。

🔄 双闭环知识系统 (Dual-Loop Knowledge)

Agent Workstation 是双闭环知识系统的枢纽：

┌─────────────────────────────────────────────────────┐
│  小闭环（本地，免费）                                  │
│  Agent 本地 learn → recall → evolve                  │
│  离线也能用，数据完全在你手里                           │
│                                                     │
│  大闭环（Hub，增值）                                   │
│  Agent ↔ Workstation Hub 知识共享                    │
│  集体智慧 > 个体经验，新 Agent 站在前人肩膀上            │
└─────────────────────────────────────────────────────┘

本地是主人，Hub 是助手——没有网络也能用，联网后自动同步和进化。

agentkits (Model Layer) — 调 LLM
     ↕
opc-agent (Runtime) — 跑 Agent（本地）
     ↕
deepbrain (Memory Engine) — 存知识（引擎）
     ↕
agent-workstation (Knowledge Platform) — 知识生命周期（Hub）  ← 你在这里

⚡ 快速开始

配合 OPC Agent CLI

npm install -g opc-agent
opc init my-agent --role customer-service-rep
# → 自动生成: SOUL.md + agent.yaml + CONTEXT.md + brain-seed.md

编程使用

import { searchRoles, getPopularRoles, getCategories } from 'agent-workstation';

// 模糊搜索
const roles = searchRoles('developer');
console.log(roles);
// → [{ category: 'engineering', role: 'backend-developer', score: 110 }, ...]

// 获取热门角色
const popular = getPopularRoles();

// 浏览分类
const categories = getCategories();

✨ 核心特性

特性说明
👤20 个生产就绪角色每个含 50–80 行 rich system prompt，开箱即用
🧠Brain Seed 知识种子Agent 创建即有行业记忆，无冷启动
🏭三层分类体系行业 → 职能 → 工位，19 个职能方向
🔍模糊搜索searchRoles('customer') 智能匹配角色名、描述
✅模板校验validateRole() 检查完整性，给出错误/警告
🎨Web UI内置角色浏览器 + 模板预览（WorkstationUI）
📦OPC Agent 深度集成opc init --role 一键生成完整 Agent 工作空间

📋 完整角色列表（20 个生产就绪）

角色职能Prompt 行数Brain Seed
customer-service-repCustomer Service71✅
complaint-handlerCustomer Service71✅
live-chat-agentCustomer Service71✅
backend-developerEngineering82✅
product-managerProduct79✅
content-marketerMarketing78✅
corporate-lawyerLegal77✅
financial-analystFinance73✅
data-analystData71✅
operations-analystOperations71✅
sales-development-repSales71✅
hr-coordinatorHR71—
tax-preparerFinance71—
recruiterHR69✅
onboarding-specialistHR69—
copywriterContent69✅
accounts-payable-clerkFinance69—
social-media-managerMarketing69—
account-executiveSales69—
sales-analystSales69—

另有 80 个骨架模板（含 OAD 配置，prompt 补全中），覆盖 admin、design、education、executive、healthcare、tech 等全部 19 个职能。

🏗️ 三层分类体系

行业 (11)  →  职能 (19)  →  工位 (100)

Technology     Engineering     backend-developer, frontend-developer, devops-engineer ...
               Product         product-manager, product-owner ...
               Data            data-analyst, data-engineer ...

E-commerce     Sales           sales-development-rep, account-executive ...
               Marketing       content-marketer, social-media-manager ...
               Customer Svc    customer-service-rep, complaint-handler ...

Finance        Finance         financial-analyst, tax-preparer ...
               Legal           corporate-lawyer, compliance-officer ...
全部 19 个职能及模板数
职能模板数示例角色
💻 Engineering10backend-developer, frontend-developer, devops-engineer, ml-engineer
📦 Product6product-manager, product-owner, ux-researcher
📊 Data4data-analyst, data-engineer, business-intelligence-analyst
🛒 Sales4sales-development-rep, account-executive, sales-analyst
📢 Marketing3content-marketer, social-media-manager, seo-specialist
💬 Customer Service5customer-service-rep, complaint-handler, live-chat-agent
🎯 Customer Success5customer-success-manager, onboarding-specialist
📝 Content2copywriter, technical-writer
🎨 Design5ui-designer, brand-designer, ux-writer
💰 Finance3financial-analyst, accounts-payable-clerk, tax-preparer
👥 HR4recruiter, hr-coordinator, onboarding-specialist
⚖️ Legal6corporate-lawyer, compliance-officer, contract-reviewer
⚙️ Operations12operations-analyst, logistics-coordinator, supply-chain-manager
🔬 Research5trend-analyst, competitive-intelligence-analyst
🎓 Education8course-designer, corporate-trainer, knowledge-manager
🏥 Healthcare7patient-coordinator, health-advisor, clinical-data-analyst
👔 Executive5chief-of-staff, strategy-director
🛡️ Admin3executive-assistant, office-manager
💻 Tech3code-reviewer, security-engineer

🧠 三层知识架构 (3-Tier Brain Seeds)

每个角色拥有三层知识种子，从通用到具体层层递进：

roles/
├── tech/                              ← 行业分类
│   ├── brain-seed.md                  ← 🏭 行业级知识 (Industry)
│   ├── frontend-developer/
│   │   ├── brain-seed.md              ← 💼 岗位级知识 (Job)
│   │   ├── workstation-seed.md        ← 🔧 模板级知识 (Workstation)
│   │   └── system-prompt.md
层级文件内容示例
🏭 行业级roles/{category}/brain-seed.md行业概览、核心概念、最佳实践、KPI、法规、趋势软件工程行业的 CI/CD、云原生、DORA 指标
💼 岗位级roles/{category}/{role}/brain-seed.md岗位专业知识、技能框架、方法论前端开发的 React/Vue、性能优化、可访问性
🔧 模板级roles/{category}/{role}/workstation-seed.md典型场景、SOP、常见问题、质量标准、新人指南前端工位的 Code Review 流程、Lighthouse 标准

当 opc init --role frontend-developer 时，自动生成：

my-agent/
├── brain-seeds/
│   ├── industry.md         ← 来自 tech/brain-seed.md
│   ├── job.md              ← 来自 tech/frontend-developer/brain-seed.md
│   └── workstation.md      ← 来自 tech/frontend-developer/workstation-seed.md
├── context.md              ← 空模板，填写公司特定知识
├── SOUL.md
└── agent.yaml

API 使用

import { getBrainSeeds, getIndustryBrainSeed, getWorkstationBrainSeed, getContextTemplate } from 'agent-workstation';

// 获取某角色的全部三层知识
const seeds = getBrainSeeds('frontend-developer');
// → { industry, job, workstation, contextTemplate }

// 单独获取行业级知识
const techSeed = getIndustryBrainSeed('tech');

// 获取企业上下文模板
const template = getContextTemplate();

📂 模板结构

每个完整模板包含五个文件：

roles/customer-service/customer-service-rep/
├── oad.yaml              # OAD 声明式配置（技能、渠道、指标）
├── system-prompt.md      # 50–80 行专业 system prompt
├── brain-seed.md         # 💼 岗位知识种子
├── workstation-seed.md   # 🔧 模板知识种子（SOP、场景、标准）
└── README.md             # 岗位说明书

workstation-seed.md 结构

# {角色}工位 — 模板知识库

## 典型工作场景 (3-5个)
## 标准操作流程 (SOP)
## 常见问题处理方案
## 质量标准/检查清单
## 新人上手指南
## 跨部门协作要点

📖 API Reference

import {
  getCategories,            // () => { name, roles[] }[]
  getRole,                  // (category, role) => { category, role, files }
  searchRoles,              // (query) => { category, role, title, score }[]
  validateRole,             // (category, role) => { valid, errors[], warnings[] }
  getPopularRoles,          // () => top 20 curated roles
  getIndustries,            // () => industries/index.yaml content
  getIndustryBrainSeed,     // (category) => string | null
  getJobBrainSeed,          // (category, role) => string | null
  getWorkstationBrainSeed,  // (category, role) => string | null
  getContextTemplate,       // () => string
  getBrainSeeds,            // (role) => { industry, job, workstation, contextTemplate } | null
  WorkstationUI,            // Web UI server class
} from 'agent-workstation';

searchRoles(query)

模糊搜索，支持多词匹配，按相关度排序：

searchRoles('frontend');
// → [{ category: 'engineering', role: 'frontend-developer', score: 110 }]

validateRole(category, role)

校验模板完整性：

validateRole('sales', 'sales-development-rep');
// → { valid: true, errors: [], warnings: [] }

🔗 四件套生态

Agent Workstation 是跃盟 AI Agent 基础设施的一部分：

项目定位与 Agent Templates 的关系
opc-agentAgent Runtimeopc init --role 消费模板
deepbrainAgent Memory（自进化 Agent 记忆）brain-seed → learn() 注入初始记忆
agentkitsAgent Model LayerMock 模式测试模板效果
agent-workstationAgent Templates← 你在这里

🔄 自进化飞轮：Templates 自带 Brain Seed → Model Layer 每次调用自动 learn → Runtime 持续运行 → Memory 自动 evolve → Agent 越来越强。

opc init --role customer-service-rep
    │
    ├──→ agent-workstation    提供 system-prompt + oad.yaml
    ├──→ deepbrain            brain-seed.md → learn() 初始化记忆
    └──→ agentkits            连接 LLM + 记忆，运行 Agent

📊 完成度仪表盘 / Progress Dashboard

总览 Overview

指标数据
行业分类 (Industry)11 个
职能方向 (Function)19 个
角色总数 (Total Roles)100
生产就绪 (Production-Ready)20 (含 rich prompt + brain-seed + OAD)
骨架模板 (Skeleton)80 (含 OAD，prompt 补全中)
Brain Seed 覆盖14/20 生产就绪角色 (70%)
行业级种子 (Industry Seeds)19 个行业

各职能完成度

职能 Function角色数生产就绪完成率
💻 Engineering101 (backend-developer)10%
📦 Product61 (product-manager)17%
📊 Data41 (data-analyst)25%
🛒 Sales43 (SDR, account-exec, sales-analyst)75%
📢 Marketing31 (content-marketer)33%
💬 Customer Service53 (rep, complaint, live-chat)60%
🎯 Customer Success500%
📝 Content21 (copywriter)50%
🎨 Design500%
💰 Finance32 (financial-analyst, tax-preparer)67%
👥 HR42 (recruiter, hr-coordinator)50%
⚖️ Legal61 (corporate-lawyer)17%
⚙️ Operations121 (operations-analyst)8%
🔬 Research500%
🎓 Education800%
🏥 Healthcare700%
👔 Executive500%
🛡️ Admin300%
💻 Tech300%

最近更新 Recent Updates

日期更新内容
2026-04v2.1.0: 三层知识种子 (Industry/Job/Workstation) 架构完成
2026-04新增 WorkstationUI Web 浏览器组件
2026-0320 个角色补齐 brain-seed + workstation-seed
2026-03BrainSeedLoader + KnowledgeEvolver API 上线
2026-02v1.5: 首批 20 个生产就绪角色完成
2026-01v1.0: 100 个骨架模板 + 19 职能分类

🗺️ Roadmap

阶段目标状态
v1.0100 骨架模板✅
v1.520 完整模板（含 prompt + brain-seed）✅
v2.0100 完整模板🚧
v3.0849 岗位全覆盖📋

License

双许可证 (Dual License)：

许可证适用场景文件
LGPL-3.0开源使用、二次开发（修改需开源）LICENSE + LICENSE-GPL3
Commercial闭源修改、专有产品嵌入、企业支持LICENSE-COMMERCIAL
  • ✅ 直接使用模板（不修改）：无需开源你的项目
  • ✅ 修改模板内容：修改部分需以 LGPL-3.0 开源
  • ✅ 创建全新模板：完全属于你
  • 💼 需要闭源修改或专有分发？请联系 licensing@deepleaper.com

👤 Agent Workstation

Where Agents Learn, Grow & Share — Knowledge Lifecycle Platform for Agents

npm version License Tests

Agents aren't static tools — they learn, grow, and evolve. Agent Workstation is where knowledge begins and returns: Brain Seeds give Agents industry knowledge at birth, and the Hub enables collective evolution through knowledge sharing.
Agent Workstation is the Knowledge Platform layer in The Self-Evolving Agent Stack. Developer path: ① Pick a template (Agent Templates) → ② Connect models (Agent Model Layer) → ③ Run it (Agent Runtime) → ④ Auto-evolve (Agent Memory).
20 roles ship with expert system prompts + industry knowledge seeds + full OAD configs. The remaining 80 are in progress.

Quick Start · Roles · API · 中文版

⚡ Quick Start

With OPC Agent CLI

npm install -g opc-agent
opc init my-agent --role customer-service-rep
# → Generates: SOUL.md + agent.yaml + CONTEXT.md + brain-seed.md

Programmatic Usage

import { searchRoles, getPopularRoles, getCategories } from 'agent-workstation';

const roles = searchRoles('developer');
// → [{ category: 'engineering', role: 'backend-developer', score: 110 }, ...]

const popular = getPopularRoles(); // Top 20 curated roles
const categories = getCategories(); // All 19 function categories

✨ Key Features

FeatureDescription
👤20 Production-Ready RolesEach with 50–80 line rich system prompts
🧠Brain SeedIndustry knowledge seeds — no cold start
🏭3-Tier TaxonomyIndustry → Function → Role, 19 functions
🔍Fuzzy SearchsearchRoles('customer') matches names & descriptions
✅ValidationvalidateRole() checks completeness with errors/warnings
🎨Web UIBuilt-in role browser + template preview
📦OPC Agent Integrationopc init --role generates a full agent workspace

📋 Production-Ready Roles (20)

RoleFunctionPrompt LinesBrain Seed
customer-service-repCustomer Service71✅
complaint-handlerCustomer Service71✅
live-chat-agentCustomer Service71✅
backend-developerEngineering82✅
product-managerProduct79✅
content-marketerMarketing78✅
corporate-lawyerLegal77✅
financial-analystFinance73✅
data-analystData71✅
operations-analystOperations71✅
sales-development-repSales71✅
hr-coordinatorHR71—
tax-preparerFinance71—
recruiterHR69✅
onboarding-specialistHR69—
copywriterContent69✅
accounts-payable-clerkFinance69—
social-media-managerMarketing69—
account-executiveSales69—
sales-analystSales69—

Plus 80 skeleton templates (OAD config included, prompts in progress) across all 19 functions.

🏗️ 3-Tier Taxonomy

Industry (11)  →  Function (19)  →  Role (100)

Technology     Engineering     backend-developer, frontend-developer, devops-engineer ...
               Product         product-manager, product-owner ...
               Data            data-analyst, data-engineer ...

E-commerce     Sales           sales-development-rep, account-executive ...
               Marketing       content-marketer, social-media-manager ...
               Customer Svc    customer-service-rep, complaint-handler ...

Finance        Finance         financial-analyst, tax-preparer ...
               Legal           corporate-lawyer, compliance-officer ...

📂 Template Structure

roles/customer-service/customer-service-rep/
├── oad.yaml          # Declarative config (skills, channels, metrics)
├── system-prompt.md  # 50–80 line expert system prompt
├── brain-seed.md     # 🧠 Industry knowledge seed (works with DeepBrain)
└── README.md         # Role description

📖 API Reference

import {
  getCategories,     // () => { name, roles[] }[]
  getRole,           // (category, role) => { category, role, files }
  searchRoles,       // (query) => { category, role, title, score }[]
  validateRole,      // (category, role) => { valid, errors[], warnings[] }
  getPopularRoles,   // () => top 20 curated roles
  getIndustries,     // () => industries/index.yaml content
  getIndustryBrainSeed,     // (category) => string | null
  getJobBrainSeed,          // (category, role) => string | null
  getWorkstationBrainSeed,  // (category, role) => string | null
  getContextTemplate,       // () => string
  getBrainSeeds,            // (role) => { industry, job, workstation, contextTemplate } | null
  WorkstationUI,     // Web UI server class
} from 'agent-workstation';

🔗 Ecosystem

ProjectPurposeRelationship
opc-agentAgent Runtimeopc init --role consumes templates
deepbrainAgent Memory (Self-Evolving)brain-seed → learn()
agentkitsAgent Model LayerMock mode for testing
agent-workstationAgent Templates← You are here

🔄 Self-Evolution Flywheel: Templates ship with Brain Seeds → Model Layer auto-learns on every call → Runtime runs continuously → Memory auto-evolves → Agent gets stronger over time.

🔄 Dual-Loop Knowledge System

Agent Workstation is the hub of the dual-loop knowledge system:

┌─────────────────────────────────────────────────────┐
│  Small Loop (Local, Free)                           │
│  Agent local learn → recall → evolve                │
│  Works offline, data stays on your machine          │
│                                                     │
│  Big Loop (Hub, Value-Add)                          │
│  Agent ↔ Workstation Hub knowledge sharing          │
│  Collective wisdom > individual experience          │
└─────────────────────────────────────────────────────┘

Local is the owner, Hub is the helper — works without internet, auto-syncs when connected.

agentkits (Model Layer) — LLM calls
     ↕
opc-agent (Runtime) — run Agents (local)
     ↕
deepbrain (Memory Engine) — store knowledge (engine)
     ↕
agent-workstation (Knowledge Platform) — knowledge lifecycle (Hub)  ← You are here

📊 Progress Dashboard

MetricData
Industries11
Functions19
Total Roles100
Production-Ready20 (rich prompt + brain-seed + OAD)
Skeleton80 (OAD included, prompts in progress)
Brain Seed Coverage14/20 production roles (70%)
Industry Seeds19 industries

Recent Updates

DateUpdate
2026-04v2.1.0: 3-tier knowledge seed architecture complete
2026-04Added WorkstationUI web browser component
2026-0320 roles completed with brain-seed + workstation-seed
2026-03BrainSeedLoader + KnowledgeEvolver API launched
2026-02v1.5: First 20 production-ready roles
2026-01v1.0: 100 skeleton templates + 19 functions

🗺️ Roadmap

|-------|------|:------:| | v1.0 | 100 skeleton templates | ✅ | | v1.5 | 20 complete templates (prompt + brain-seed) | ✅ | | v2.0 | 100 complete templates | 🚧 | | v3.0 | 849 roles full coverage | 📋 |

License

Dual License:

LicenseUse CaseFile
LGPL-3.0Open-source use, modifications must be open-sourcedLICENSE + LICENSE-GPL3
CommercialClosed-source modifications, proprietary embedding, enterprise supportLICENSE-COMMERCIAL
  • ✅ Use templates as-is (unmodified): no need to open-source your project
  • ✅ Modify template content: modifications must be shared under LGPL-3.0
  • ✅ Create brand-new templates: fully yours
  • 💼 Need closed-source modifications or proprietary distribution? Contact licensing@deepleaper.com

Keywords

ai-agent
agent-templates
workstation
opc
deepleaper
agent-roles

FAQs

Package last updated on 19 Apr 2026

Did you know?

Socket

Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.

Install

Related posts

Socket Partners with Replit to Block Malicious Packages in AI-Powered Development

Company News

Socket Partners with Replit to Block Malicious Packages in AI-Powered Development

Replit is integrating Socket Firewall into its AI-powered development experience to help protect builders from malicious open source packages.

By Feross Aboukhadijeh  -  Jun 10, 2026