JoeMCP

JoeMCP

Connects AI assistants to JoeAPI construction management system, enabling management of clients, contacts, proposals, estimates, action items, project schedules, and financial data through natural language.

Category
访问服务器

README

JoeAPI MCP Server

Model Context Protocol (MCP) server for JoeAPI construction management system. Exposes construction management tools to Claude and other AI assistants.

Overview

This MCP server provides:

  • 18 pre-built workflows for common construction management tasks
  • 60+ individual tools for direct API access (CRUD operations)
  • Multi-transport support: Smithery (cloud) and STDIO (local)

Architecture

mcp/
├── index.ts                           # Main MCP server (Smithery + STDIO)
├── local-server.ts                    # STDIO transport runner
├── claude-desktop-config.example.json # Claude Desktop config
├── README.md                          # This file
└── server.ts                          # Legacy (deprecated)

mcp-build/                             # Compiled JavaScript (production)
├── index.js                           # Compiled server
├── local-server.js                    # Compiled STDIO runner
└── *.d.ts                             # TypeScript definitions

Deployment Options

1. Smithery (Cloud) - EXISTING

Your MCP server is already deployed on Smithery for cloud access.

Access: Via Smithery marketplace URL: https://smithery.ai/server/@your-username/joeapi Transport: HTTP Use case: Remote access, team collaboration

2. Local STDIO - THIS SETUP

Run MCP server locally on your machine for development/testing.

Transport: STDIO (Standard Input/Output) Use case: Local development, faster iteration, offline work


Local STDIO Setup

Prerequisites

  1. JoeAPI running locally:

    npm run dev  # Start JoeAPI on http://localhost:8080
    
  2. Build MCP server:

    npm run mcp:build
    
  3. Configure Claude Desktop:

    • Copy mcp/claude-desktop-config.example.json
    • Update the command path to match your installation
    • Add to Claude Desktop config

Option A: Development Mode (TypeScript)

For rapid development with auto-recompile:

npm run mcp:local

This runs tsx mcp/local-server.ts - TypeScript executed directly (no build needed).

Option B: Production Mode (Compiled JavaScript)

For production use:

# Build once
npm run mcp:build

# Run compiled version
npm run mcp:start

This runs node mcp-build/local-server.js - faster startup, production-ready.


Claude Desktop Configuration

Step 1: Find Claude Config Directory

macOS:

~/Library/Application Support/Claude/

Windows:

%APPDATA%\Claude\

Linux:

~/.config/Claude/

Step 2: Edit claude_desktop_config.json

{
  "mcpServers": {
    "joeapi-local": {
      "command": "node",
      "args": [
        "/ABSOLUTE/PATH/TO/joeapi/mcp-build/local-server.js"
      ],
      "env": {
        "JOEAPI_BASE_URL": "http://localhost:8080"
      },
      "disabled": false
    }
  }
}

Important:

  • Use absolute paths (not relative like ~/ or ./)
  • Change /ABSOLUTE/PATH/TO/joeapi/ to your actual path
  • Example: /Users/joe/dev/joeapi/mcp-build/local-server.js

Step 3: Restart Claude Desktop

Close and reopen Claude Desktop. The MCP server will connect automatically.

Step 4: Verify Connection

In Claude Desktop, type:

Can you use the find_workflow tool?

If connected, Claude will have access to all JoeAPI tools.


Available Tools

Workflow Discovery (ALWAYS START HERE!)

find_workflow - Discover pre-built workflows

  • Set autoExecute: false to see ALL 18 workflows
  • Set autoExecute: true with workflow name to get step-by-step instructions

18 Pre-Built Workflows

  1. win_loss_rate - Calculate proposal win/loss statistics
  2. sales_pipeline - Analyze active proposals in pipeline
  3. work_in_process_report - WIP report for active projects
  4. job_costing_detail - Detailed job costing analysis
  5. project_profitability - Profit analysis by project
  6. cost_variance_analysis - Actual vs. estimated cost variance
  7. cash_flow_forecast - Project cash flow projections
  8. schedule_variance_analysis - Schedule delays and impacts
  9. client_portal_update - Generate client progress updates
  10. subcontractor_performance - Analyze subcontractor metrics
  11. material_tracking - Track material costs and usage
  12. labor_productivity - Analyze labor efficiency
  13. cost_per_square_foot - Calculate $/sqft by trade
  14. change_order_tracking - Track change orders/revisions
  15. upgrade_pricing - Price client upgrade requests
  16. update_schedule - Extend/adjust project schedules
  17. plan_takeoff - Estimate from building plans
  18. estimate_from_previous - Create estimate from past job

60+ Individual Tools (Categories)

  • Clients - CRUD operations for clients
  • Contacts - Manage client contacts
  • SubContractors - Subcontractor management
  • Proposals - Create and track proposals
  • ProposalLines - Line items within proposals
  • Estimates - Project estimates
  • ProjectManagements - Active project data
  • ProjectSchedules - Project timelines
  • ProjectScheduleTasks - Individual tasks
  • ActionItems - Track action items with:
    • Comments
    • Supervisors
    • Cost changes
    • Schedule changes
  • Transactions - QuickBooks transaction data
  • JobBalances - Current job balances
  • CostVariance - Cost variance analysis
  • Invoices - Invoice management
  • ScheduleRevisions - Schedule change tracking
  • ProjectDetails - Comprehensive project data
  • ProposalPipeline - Pipeline analytics
  • EstimateRevisions - Estimate change history
  • CostRevisions - Cost revision tracking
  • Deposits - Deposit/retainer management
  • ProposalTemplatePricing - Standard pricing templates

Environment Variables

Required

JOEAPI_BASE_URL=http://localhost:8080  # Local JoeAPI server

Optional (if JoeAPI requires auth)

JOEAPI_API_KEY=your_api_key           # API key for JoeAPI
JOEAPI_USER_ID=1                      # Dev user ID

Set these in:

  1. Claude Desktop config (recommended):

    "env": {
      "JOEAPI_BASE_URL": "http://localhost:8080"
    }
    
  2. Shell environment (alternative):

    export JOEAPI_BASE_URL=http://localhost:8080
    

Troubleshooting

MCP Server Not Connecting

1. Check JoeAPI is running:

curl http://localhost:8080/health
# Should return: {"status":"healthy",...}

2. Check MCP server builds:

npm run mcp:build
# Should complete without errors

3. Check Claude Desktop logs:

macOS:

tail -f ~/Library/Logs/Claude/mcp*.log

Windows:

Get-Content $env:APPDATA\Claude\logs\mcp*.log -Wait

4. Test MCP server manually:

npm run mcp:local
# Should output: "JoeAPI MCP Server running locally"
# Press Ctrl+C to stop

Tools Not Appearing in Claude

1. Verify config path is absolute:

"args": ["/Users/joe/dev/joeapi/mcp-build/local-server.js"]

Not: ["~/dev/joeapi/mcp-build/local-server.js"]

2. Check disabled flag:

"disabled": false  // Should be false

3. Restart Claude Desktop:

  • Close completely (Cmd+Q on macOS)
  • Reopen

4. Check for errors in MCP logs

JoeAPI Connection Errors

Error: JoeAPI error (401)

  • JoeAPI requires authentication
  • Add API key or dev user ID to config

Error: JoeAPI error (404)

  • Endpoint not found
  • Check JoeAPI version matches MCP server

Error: ECONNREFUSED

  • JoeAPI not running
  • Start with: npm run dev

Development

Adding New Tools

  1. Edit mcp/index.ts:

    {
      name: 'my_new_tool',
      description: 'Description of tool',
      inputSchema: {
        type: 'object',
        properties: {
          param1: { type: 'string', description: '...' }
        },
        required: ['param1']
      }
    }
    
  2. Add handler in CallToolRequest:

    case 'my_new_tool': {
      const { param1 } = args as { param1: string };
      result = await callJoeAPI(baseUrl, `/api/v1/endpoint`);
      break;
    }
    
  3. Rebuild:

    npm run mcp:build
    
  4. Restart Claude Desktop

Adding New Workflows

  1. Add to prompts array in mcp/index.ts:

    {
      name: 'my_workflow',
      description: 'Brief description',
      arguments: [...],
      prompt: 'WORKFLOW: My Workflow\n\nPURPOSE: ...\n\nSTEPS:\n1. ...'
    }
    
  2. Rebuild and restart


Scripts Reference

# Development
npm run mcp:local     # Run TypeScript directly (tsx)
npm run dev          # Run JoeAPI server

# Production
npm run mcp:build    # Compile TypeScript → JavaScript
npm run mcp:start    # Run compiled JavaScript

# Both
npm run verify-db    # Test database connection

Comparison: Smithery vs Local STDIO

Feature Smithery (Cloud) Local STDIO
Setup Already deployed Requires local setup
Access Remote, anywhere Local machine only
Speed Network latency Instant (local)
Availability Always on Requires JoeAPI running
Updates Auto-deployed Manual build required
Collaboration Team access Single user
Best for Production, team Development, testing

Support

  • MCP SDK Docs: https://modelcontextprotocol.io
  • Claude Desktop MCP Guide: https://docs.claude.com/docs/mcp
  • JoeAPI Issues: Check GitHub repo
  • MCP Server Code: mcp/index.ts

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选