arcpy-mcp-server
Enables AI agents to perform spatial analysis in ArcGIS Pro using natural language by exposing all ArcGIS Pro toolboxes as MCP tools.
README
arcpy-mcp-server -- AI + GIS Smart Spatial Analysis
Let LLMs (GPT-4o, Claude, DeepSeek) directly operate ArcGIS Pro -- drive 1300+ spatial analysis tools with natural language.
What is this?
arcpy-mcp-server exposes all ArcGIS Pro toolboxes as MCP (Model Context Protocol) tools, enabling:
You: "Buffer analysis 500m around the university"
AI: Auto-calls arcpy -> ArcGIS Pro -> generates result
No tool names to memorize. No Python to write. No menus to click.
Features
- 1300+ tools across 15 modules: management(392), sa(355), ddd(144), stats(44), ga(39), analysis(38) + 9 more
- MCP standard: works with AutoClaw, Claude Code, Codex, any MCP client
- Zero dependencies: uses ArcGIS Pro's built-in Python 3.9
- Security whitelist: configurable path and tool restrictions
- One-click install: 30 seconds from clone to running
Supported Modules
| Module | Tools | Highlights |
|---|---|---|
| management | 392 | Data management, projection, fields, topology |
| sa (Spatial Analyst) | 355 | Slope, aspect, reclassify, viewshed, hydrology |
| ddd (3D Analyst) | 144 | DEM, contours, viewshed2, skyline, point cloud |
| nax (Network Analyst) | 61 | Route, service area, OD matrix |
| conversion | 57 | JSON/KML/CAD/Excel/raster conversion |
| na | 55 | Legacy network analysis |
| stats (Spatial Statistics) | 44 | Moran's I, Gi*, KDE, Ripley's K, GWR |
| cartography | 44 | Contour labels, aggregation, simplification |
| ga (Geostatistical) | 39 | Kriging, IDW, EBK, cross-validation |
| analysis | 38 | Buffer, Clip, Intersect, Erase, Union, Thiessen |
| edit | 18 | Align, extend, trim, split, merge |
| server | 18 | Publish services, cache, tiles |
| stpm (SpaceTime) | 16 | Space-time cube, emerging hotspots |
| geocoding | 13 | Address matching, reverse geocoding |
| sharing | 6 | Web layers, web maps, scene sharing |
Quick Start
Prerequisites
- Windows 10/11
- ArcGIS Pro 3.x (with valid license)
- Python 3.9+
Install
git clone https://github.com/zhaojj662/arcpy-mcp-server.git
cd arcpy-mcp-server
install.bat
Start Server
Double-click start_server.bat or run:
start_server.bat
Server runs on http://127.0.0.1:8765.
Verify:
Invoke-RestMethod http://127.0.0.1:8765/health
# {"status":"ok","server":"arcpy-mcp-server","version":"2.0","tools":1300}
Configure AI Client
AutoClaw MCP Settings:
| Field | Value |
|---|---|
| Name | arcpy |
| Mode | Local process (stdio) |
| Command | python |
| Args | path/to/mcp_bridge.py |
Claude Code (claude_desktop_config.json):
{
"mcpServers": {
"arcpy": {
"command": "python",
"args": ["C:/Users/%USERNAME%/arcpy-mcp-server/mcp_bridge.py"]
}
}
}
Try It
"Buffer roads.shp by 500 meters"
"Calculate slope and aspect from this DEM"
"Hotspot analysis on crime points"
"Kriging interpolation on PM2.5 monitoring data"
"Batch viewshed for 100 observer points"
Architecture
AI Client <-- stdio --> mcp_bridge.py <-- HTTP --> server.py (arcpy) --> ArcGIS Pro 3.x
Security
Edit server.py to restrict access:
ALLOWED_PATHS = ["C:/GIS-AI-Course/"] # Only allow these directories
INCLUDE_MODULES = ["analysis", "sa"] # Only expose these modules
API
| Endpoint | Method | Description |
|---|---|---|
| /health | GET | Server status, tool count |
| /tools | GET | List available tools (first 200) |
| /modules | GET | List all modules with tool counts |
| /call | POST | Execute a tool |
FAQ
Q: Is ArcGIS Pro required? A: Yes. arcpy runs only inside the ArcGIS Pro Python environment.
Q: Supported ArcGIS Pro versions? A: 3.x series (tested on 3.0).
Q: Can I limit which tools AI can access?
A: Yes. Edit INCLUDE_MODULES in server.py.
License
MIT -- see LICENSE
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。