ArcGIS Pro CLI

Give AI agents eyes into ArcGIS Pro.

pip install arcgispro-cli

# Optional: install TUI dependencies
pip install arcgispro-cli[tui]

arcgis install

First 5 Minutes (Quickstart)

1. Open an ArcGIS Pro project (.aprx)
2. Click Snapshot in the CLI ribbon tab
3. In a terminal, run:

   arcgis status
   arcgis layers
   arcgis layer "Parcels"
   

More: docs/quickstart.md

What's New in v0.4.0

• Enhanced TUI with map preview support and improved banner rendering
• Mermaid project structure export (project-structure.mmd + markdown wrapper)
• Best-effort stable IDs for maps/layers/tables to improve snapshot tracking
• Geoprocessing history export scaffold for richer context artifacts
• Reliability fixes including Python 3.9 compatibility and improved terminal/add-in robustness

How It Works

ProExporter (Pro add-in) creates detailed flat files that explain the state of your ArcGIS Pro project. The arcgis CLI tool facilitates frictionless reasoning over the context. Fewer assumptions and annoying follow-up questions. Helps the AI help you.

(Backwards compatible aliases: arcgispro, agp.)

1. Open a project in ArcGIS Pro
2. Click Snapshot in the CLI ribbon tab
3. Ask questions:

   arcgis layers              # What layers do I have?
   arcgis layer "Parcels"     # Tell me about this layer
   arcgis fields "Parcels"    # What fields are in it?
   

CLI Commands

Setup

Query

Add --json to any query command for machine-readable output.

Troubleshooting

arcgispro launches ArcGIS Pro instead of the CLI?

This happens if C:\Program Files\ArcGIS\Pro\bin is on your PATH. Options:

• Prefer arcgis (new default): arcgis layers, arcgis launch
• Use agp (alias): agp layers, agp launch
• Or fix PATH order: ensure Python Scripts comes before ArcGIS Pro bin

Requirements

• Windows 10/11
• ArcGIS Pro 3.x
• Python 3.9+

Development

To build the add-in from source, you'll need:

• Visual Studio 2022 with ArcGIS Pro SDK extension
• .NET 8 SDK

# Clone and install CLI in dev mode
git clone https://github.com/danmaps/arcgispro_cli.git
cd arcgispro_cli/cli
pip install -e .

# Build add-in in Visual Studio
# Open ProExporter/ProExporter.sln
# Build → Build Solution (Release)

License

MIT

Using with AI Agents

This tool is designed to make ArcGIS Pro sessions observable for AI coding assistants.

What Gets Exported

When you click Snapshot in ArcGIS Pro, the project structure is:

project_root/
├── AGENTS.md              # AI agent skill file (start here!)
├── YourProject.aprx       # ArcGIS Pro project file
└── .arcgispro/
    ├── config.yml         # Export settings (auto-export, toggles)
    ├── meta.json          # Export timestamp, tool version
    ├── context/
    │   ├── project.json       # Project name, path, geodatabases
    │   ├── maps.json          # Map names, spatial references, scales
    │   ├── layers.json        # Full layer details with field schemas
    │   ├── tables.json        # Standalone tables
    │   ├── connections.json   # Database connections
    │   ├── layouts.json       # Print layouts
    │   └── notebooks.json     # Jupyter notebooks
    ├── images/
    │   ├── map_*.png          # Screenshots of each map view
    │   └── layout_*.png       # Screenshots of each layout
    └── snapshot/
        ├── context.md         # Human-readable summary
        ├── project-structure.mmd # Mermaid diagram source
        └── project-structure.md  # Mermaid diagram markdown

The AGENTS.md file teaches AI agents how to use the CLI and interpret the exported data; no user explanation needed.

Configuration

Edit .arcgispro/config.yml to control export behavior:

# Auto-export on project open (default: false)
autoExportEnabled: false
autoExportLocalOnly: true   # Skip network drives
autoExportMaxLayers: 50     # Safety limit

# Content toggles
exportImages: true          # Map/layout screenshots
exportNotebooks: true       # Jupyter notebook metadata
exportFields: true          # Layer field schemas

Claude Code / Copilot CLI / Gemini CLI

These tools can read files and run commands in your working directory. Navigate to your ArcGIS Pro project folder and start your AI session:

cd /path/to/your/project
claude   # or: copilot, gemini

Example prompts:

What layers are in this project?
> AI runs: arcgis layers

What fields are in the Parcels layer?
> AI runs: arcgis fields "Parcels"

Which layers have broken data sources?
> AI runs: arcgis layers --broken

Give me the full project context
> AI runs: arcgis context

Look at the map screenshot and describe what you see
> AI reads: .arcgispro/images/map_*.png

Tips for Best Results

1. Click Snapshot in Pro before starting your AI session - ensures context is fresh

2. Ask naturally - the CLI commands map to common questions:

   • "What layers do I have?" → arcgis layers
   • "Tell me about the Parcels layer" → arcgis layer Parcels
   • "What's the schema?" → arcgis fields Parcels

3. Use --json for programmatic access - AI can parse structured output:

   arcgis layers --json
   arcgis layer "Parcels" --json
   

4. Check images for visual context - map screenshots help AI understand spatial data

5. Be bold. Try pasting in a question you'd normally answer by working in ArcGIS Pro manually.

   • "Jeff wants an updated map of the project area with an imagery basemap instead of streets"
     • AI generates a (working) python script that exports the PDF directly, using your existing map and layout. You get to go to lunch early, and get a raise.

Custom Agent Integration

The JSON files are designed for programmatic access:

import json
from pathlib import Path

context_dir = Path(".arcgispro/context")
layers = json.loads((context_dir / "layers.json").read_text(encoding="utf-8-sig"))

for layer in layers:
    print(f"{layer['name']}: {layer.get('featureCount', 'N/A')} features")
    for field in layer.get('fields', []):
        print(f"  - {field['name']} ({field['fieldType']})")