PlantUML CSA MCP Server
Generate Control System Architecture (CSA) diagrams using PlantUML via the Model Context Protocol (MCP). Supports ISA-95 Purdue model, industrial symbols, and multiple protocols.
README
PlantUML CSA MCP Server
Generate Control System Architecture (CSA) diagrams using PlantUML via the Model Context Protocol (MCP).
Features
- ISA-95 Purdue Model - Zone packages with color-coded levels (0-4)
- Industrial Symbols - 36+ component types (10 controllers, 26 devices)
- Protocol Visualization - 12 protocols with distinct line colors
- Architecture Templates - 5 pre-defined templates for common configurations
- Bootstrap from Skills - Generate topology from equipment-list + instrument-io skills
- Multiple Layout Engines - Graphviz, Smetana, ELK
- Shareable URLs - Generate PlantUML server links
- Same Schema as FreeCAD - Interoperable YAML topology format
Project Structure
plantuml-csa-mcp-server/
├── src/
│ ├── __init__.py # Package entry point
│ ├── __main__.py # Module runner (python -m src)
│ ├── server.py # FastMCP server with 8 MCP tools
│ ├── models/
│ │ └── csa_topology.py # Pydantic models (same as FreeCAD CSA)
│ ├── converter/
│ │ ├── topology_to_puml.py # YAML topology → PlantUML source
│ │ ├── sprites.py # ISA symbol definitions (36+ types)
│ │ └── layout_hints.py # Purdue-aware layout directives
│ ├── renderer/
│ │ └── plantuml_runner.py # PlantUML CLI wrapper (JAR/native/Docker)
│ ├── encoder/
│ │ └── plantuml_encoder.py # Text encoding for shareable URLs
│ ├── templates/
│ │ └── architecture_templates.py # 5 architecture templates
│ └── bootstrap/
│ └── csa_bootstrap.py # Bootstrap from equipment/IO lists
├── tests/ # 56 tests
├── docs/
│ └── completed-plans/ # Implementation plan
├── pyproject.toml
├── CLAUDE.md # Development guide
└── README.md
Installation
# Clone and install
cd plantuml-csa-mcp-server
uv sync
# Verify PlantUML is available
uv run python -c "from src.renderer import PlantUMLRunner; print(PlantUMLRunner().check_available())"
Quick Start
1. Add to MCP Configuration
Add to your .mcp.json:
{
"mcpServers": {
"plantuml-csa-mcp": {
"type": "stdio",
"command": "uv",
"args": ["--directory", "/path/to/plantuml-csa-mcp-server", "run", "python", "-m", "src"],
"description": "PlantUML CSA diagram generation (ISA-95 Purdue model)"
}
}
}
2. Generate a Diagram
# topology.yaml
schema_version: "1.0"
metadata:
project_name: "Sample WWTP CSA"
zones:
- id: "level_0"
purdue_level: 0
- id: "level_1"
purdue_level: 1
controllers:
- id: "PLC-101"
type: PLC
zone: "level_1"
devices:
- id: "RIO-101"
type: RemoteIO
parent_controller: "PLC-101"
zone: "level_0"
links:
- source: "PLC-101"
target: "RIO-101"
protocol: "Ethernet_IP"
Then use the MCP tools:
# Get PlantUML source for version control
csa_get_plantuml_source(topology_yaml=yaml_content)
# Generate SVG diagram
csa_generate_diagram(topology_yaml=yaml_content, format="svg")
# Get shareable URL
csa_encode_plantuml(plantuml_source=puml_source)
MCP Tools Reference
| Tool | Purpose |
|---|---|
csa_generate_diagram |
Render topology YAML to SVG/PNG |
csa_get_plantuml_source |
Get .puml source for version control |
csa_validate_topology |
Validate YAML against schema |
csa_list_symbols |
List available ISA-style symbols |
csa_encode_plantuml |
Generate shareable PlantUML URLs |
csa_list_templates |
List architecture templates |
csa_render_preview |
Quick preview for iteration |
csa_check_plantuml |
Check PlantUML availability |
csa_bootstrap_from_io |
Bootstrap topology from equipment/IO lists |
Architecture Templates
Pre-defined templates for common control system configurations:
| Template | Description | Use Case |
|---|---|---|
centralized |
Central MCC + Central PLC | Small plants (<5 MLD) |
central_mcc_distributed_io |
Central MCC + Distributed IO | Medium plants (5-20 MLD) |
fully_distributed |
Remote panels per area | Large plants (>20 MLD) |
hybrid_safety |
Central Safety + Distributed Process | SIL/SIS requirements |
vendor_package_integration |
OEM packages via OPC-UA | Multiple vendor packages |
Bootstrap from Skill Outputs
Generate CSA topology from equipment-list-skill and instrument-io-skill outputs:
csa_bootstrap_from_io(
equipment_list_qmd=equipment_qmd_content,
instrument_database_yaml=io_database_yaml,
project_name="WWTP Control System",
architecture_template="fully_distributed"
)
# Returns: {topology_yaml, suggestions, io_summary, equipment_mapping}
Supported Components
Controllers (10 types)
PLC, DCS, PAC, Safety_PLC, Soft_PLC, Edge_Controller, Motion_Controller, Redundant_PLC, RTU, SIS
Devices (26 types)
RemoteIO, HMI, SCADA, Historian, OPC_UA_Server, Gateway, VFD, Soft_Starter, MCC, Industrial_PC, Switch, Managed_Switch, Router, Firewall, Wireless_AP, Media_Converter, Network_TAP, Motor_Starter, Engineering_WS, Panel_PC, Data_Logger, Junction_Box, Marshalling_Cabinet, Local_Panel, Remote_Panel, Instrument_Rack
Protocols (12 types)
Ethernet_IP, Profinet, Modbus_TCP, Modbus_RTU, Profibus, DeviceNet, ControlNet, HART, Foundation_Fieldbus, OPC_UA, MQTT, BACnet
Testing
# Run all tests
uv run pytest tests/ -v
# Run with coverage
uv run pytest tests/ --cov=src
vs FreeCAD CSA
| Aspect | PlantUML CSA | FreeCAD CSA |
|---|---|---|
| Output | SVG/PNG | TechDraw PDF |
| Version Control | Plain text .puml | Binary .FCStd |
| Dependencies | Java/PlantUML | FreeCAD runtime |
| Best For | Documentation | CAD deliverables |
Use Both: Same YAML topology works with both renderers. Use PlantUML for rapid iteration and documentation, FreeCAD for final engineering drawings.
Workflow Integration
This server is part of the puran-water control system architecture workflow:
┌─────────────────────────┐ ┌──────────────────────────┐ ┌─────────────────────┐
│ equipment-list-skill │ ──► │ instrument-io-skill │ ──► │ csa-diagram-skill │
│ (equipment + feeder) │ │ (IO lists, patterns) │ │ (CSA generation) │
└─────────────────────────┘ └──────────────────────────┘ └─────────────────────┘
│
▼
┌─────────────────────┐
│ plantuml-csa-mcp │
│ (this server) │
└─────────────────────┘
│
▼
┌─────────────────────┐
│ CSA Topology YAML │
│ PlantUML PNG/SVG │
│ Shareable URLs │
└─────────────────────┘
Related Projects
Upstream (Data Sources)
- equipment-list-skill - Equipment lists with feeder types
- instrument-io-skill - IO lists with DI/DO/AI/AO patterns
Companion Skill
- csa-diagram-skill - Claude Code skill that orchestrates this MCP server
Similar Pattern (Electrical)
- plantuml-sld-mcp-server - Single-Line Diagram generation (same YAML → PlantUML pattern)
- electrical-distribution-skill - SLD skill using plantuml-sld-mcp-server
Alternative Renderer
- freecad-csa-workbench - Same YAML topology rendered in FreeCAD for CAD deliverables
License
MIT
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。