PrintSmith MCP Server
Connects Claude to PrintSmith Vision for read-only AI-assisted print shop management. It enables users to query customer data, production job statuses, and financial summaries through natural language.
README
PrintSmith MCP Server
An MCP (Model Context Protocol) server that connects Claude to PrintSmith Vision, enabling AI-assisted print shop management.
All operations are READ-ONLY - this server only queries data, never modifies it.
Features
| Tool | Description |
|---|---|
lookup_customer |
Search customers by name, ID, or account number |
get_job_status |
Get production status of a specific job |
list_jobs |
List jobs filtered by status (in production, ready for pickup, etc.) |
get_customer_jobs |
Get all jobs for a specific customer |
get_ar_summary |
Accounts receivable summary |
get_estimate |
Get details of a specific estimate |
list_pending_estimates |
List estimates awaiting conversion |
health_check |
Verify PrintSmith connection |
Quick Start (Mock Data)
Test the server without a PrintSmith connection:
cd printsmith-mcp
pip install -r requirements.txt
USE_MOCK_DATA=true python src/server.py
Deployment on Proxmox LXC
Step 1: Create the LXC Container
On your Proxmox host:
# Download the setup script
scp scripts/setup-lxc.sh root@proxmox:/tmp/
# SSH to Proxmox and run
ssh root@proxmox
cd /tmp
# Edit variables at top of script, then run:
bash setup-lxc.sh
This creates an Ubuntu 24.04 LXC container with Python installed.
Step 2: Deploy Application Files
From your local machine (where you downloaded this repo):
# Set your container ID
CTID=200
# Copy files to container
pct push $CTID src/server.py /opt/printsmith-mcp/src/server.py
pct push $CTID src/printsmith_client.py /opt/printsmith-mcp/src/printsmith_client.py
pct push $CTID requirements.txt /opt/printsmith-mcp/requirements.txt
pct push $CTID scripts/install.sh /opt/printsmith-mcp/scripts/install.sh
Or use the deploy script:
bash scripts/deploy-to-lxc.sh 200
Step 3: Install Inside Container
pct enter 200
cd /opt/printsmith-mcp
bash scripts/install.sh
Step 4: Configure PrintSmith Connection
nano /opt/printsmith-mcp/.env
Edit these values:
PRINTSMITH_BASE_URL=https://your-printsmith-server.com
PRINTSMITH_API_TOKEN=your-api-token-here
USE_MOCK_DATA=false
Step 5: Start the Service
systemctl start printsmith-mcp
systemctl status printsmith-mcp
# View logs
journalctl -u printsmith-mcp -f
Step 6: Test
# From inside container
curl http://localhost:8080/health
# From Proxmox host (replace IP)
curl http://10.0.0.50:8080/health
Connecting to Claude
Option A: Claude Desktop (if LXC has STDIO access)
Not recommended for LXC. Use Option B.
Option B: Remote MCP Connection (Recommended)
The server runs in HTTP/SSE mode, accessible at:
http://<container-ip>:8080/sse
To use with Claude, you'll need an MCP client that supports remote servers. Check Anthropic's documentation for current options.
Option C: For Development/Testing
You can also run the server locally in STDIO mode for Claude Desktop:
MCP_TRANSPORT=stdio python src/server.py
PrintSmith API Reference
Getting Your API Token
- Log into PrintSmith Vision as administrator
- Go to Setup → Preferences → API Settings
- Generate or copy your API token
API Endpoints Used (Read-Only)
| Endpoint | Method | Description |
|---|---|---|
/AccountAPI/{token} |
GET | Customer/account data |
/InvoiceAPI/{token} |
GET | Invoice data |
/EstimateAPI/{token} |
GET | Estimate data |
/JobAPI/{token} |
GET | Job/production data |
/ContactAPI/{token} |
GET | Contact data |
Authentication
PrintSmith uses token-in-URL authentication:
GET https://your-server.com/AccountAPI/YOUR_API_TOKEN?account_id=1234
Common Query Parameters
| Parameter | Description |
|---|---|
account_id |
Primary key ID |
account_account_id |
Display account number |
contact_id |
Contact primary key |
start_date |
Range filter start (YYYY-MM-DD) |
end_date |
Range filter end (YYYY-MM-DD) |
status |
Status filter |
Response Format
All responses are JSON:
{
"id": "1234",
"name": "Acme Corporation",
"balance": 2450.00,
"credit_status": "good"
}
Error Responses
{
"status": "Failure",
"message": "Account with number(account_account_id): 9999 was not found."
}
Environment Variables
| Variable | Default | Description |
|---|---|---|
PRINTSMITH_BASE_URL |
(required) | PrintSmith server URL |
PRINTSMITH_API_TOKEN |
(required) | API authentication token |
PRINTSMITH_VERIFY_SSL |
true |
Verify SSL certificates |
PRINTSMITH_TIMEOUT |
30 |
Request timeout in seconds |
MCP_TRANSPORT |
stdio |
Transport: stdio or http |
MCP_HTTP_PORT |
8080 |
HTTP server port |
MCP_HTTP_HOST |
0.0.0.0 |
HTTP server bind address |
USE_MOCK_DATA |
false |
Use mock data (no PrintSmith needed) |
Project Structure
printsmith-mcp/
├── src/
│ ├── server.py # MCP server (main entry point)
│ └── printsmith_client.py # PrintSmith API client (read-only)
├── scripts/
│ ├── setup-lxc.sh # Create LXC container on Proxmox
│ ├── deploy-to-lxc.sh # Copy files to container
│ └── install.sh # Install inside container
├── requirements.txt
└── README.md
Security Considerations
- API Token: Stored in
.envfile with restricted permissions (600) - Read-Only: All operations are GET requests only
- Network: Consider firewall rules to restrict access to MCP port
- SSL: Enable
PRINTSMITH_VERIFY_SSLin production
For a multi-tenant SaaS, you'll need additional:
- Per-customer credential storage (encrypted)
- Authentication for the MCP endpoint
- Rate limiting
- Audit logging
Troubleshooting
Container won't start
pct status 200
pct config 200
journalctl -u pve-container@200
Service fails to start
pct enter 200
journalctl -u printsmith-mcp -n 50
Can't connect to PrintSmith
# Test from inside container
curl -v "https://your-printsmith-server.com/AccountAPI/YOUR_TOKEN?limit=1"
SSL certificate errors
# Temporarily disable SSL verification
echo "PRINTSMITH_VERIFY_SSL=false" >> /opt/printsmith-mcp/.env
systemctl restart printsmith-mcp
Next Steps
- Test with mock data - Verify everything works
- Connect to real PrintSmith - Add your credentials
- Identify high-value features - What do users need most?
- Add authentication - Secure the MCP endpoint
- Build multi-tenant - Credential management for multiple customers
Resources
- MCP Documentation
- MCP Python SDK
- PrintSmith Vision - Contact EFI/ePS for API documentation
推荐服务器
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。
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。