Hard-coded paths
Move machine-specific values into configuration so an AI can change the setup without rewriting the operation.
Give your coding agent enough structure to help without turning a working GIS script into a fragile mess.
ArcPy scripts rarely fail because the buffer tool is hard. They fail because paths, assumptions, geoprocessing, and business rules got tangled together. This is the practical starting point for separating them before the next AI-assisted change.
Your script may work today. The problem is that the next change now requires you—or an AI assistant—to understand everything at once.
Move machine-specific values into configuration so an AI can change the setup without rewriting the operation.
Make fields, geometry, spatial reference, and dataset choices visible before they become quiet bad output.
Keep ArcGIS Pro-dependent work isolated so the rest of the logic can be reasoned about and tested normally.
Give Claude, Copilot, or another coding agent project instructions that define what it may change and what it must preserve.
Use three ArcPy refactors as concrete patterns instead of asking an AI to invent an architecture from scratch.
Start with the script you already have. Add structure around working behavior instead of throwing it away.
Paths and parameters stop hiding inside the code. The workflow stays recognizable, but the next edit has a safer place to land.
Instead of “refactor this,” you can point an AI assistant at explicit boundaries, conventions, tests, and known GIS constraints.
Validation, logging, staging, and destructive operations become deliberate decisions—not behavior hidden in a 900-line script.
Each case study includes a reusable prompt for applying the pattern to your own ArcPy workflow.
arcpy-project-starter/ ├── .github/ │ └── copilot-instructions.md ├── AGENTS.md ├── CLAUDE.md ├── LICENSE.txt ├── NOTICE.txt ├── pyproject.toml ├── config/ ├── docs/ │ ├── case-studies.md │ ├── customizing.md │ └── getting-started.md ├── scripts/ ├── src/arcpy_project/ │ ├── config.py │ ├── logging_config.py │ └── workflows/ └── tests/
Bring the ArcPy script that has become too important to keep treating like a scratch file. The migration guide helps you preserve known behavior, move one boundary at a time, and stop when the structure is doing its job.
This is not a generic Python boilerplate dump. It is an opinionated ArcGIS Pro-aware starting point for analysts who want to use AI without surrendering judgment over important GIS work.
The free individual license lets you use and modify the kit for unlimited personal, professional, employer, and client projects. Your project-specific work is yours. The reusable Starter Kit itself may not be resold, published, or redistributed for independent reuse.
ArcGIS Pro and ArcPy are not included; ArcPy-dependent examples require an appropriate ArcGIS Pro environment and license.