geemu-skill

GEEMu is a skill folder for AI coding assistants, specifically Claude Code and OpenAI Codex. A skill folder is a set of files you install locally that the assistant reads to gain specialized knowledge and a structured workflow for a particular domain. GEEMu's domain is Google Earth Engine, a cloud-based platform that lets researchers run satellite imagery analysis and remote sensing tasks at scale without managing their own computing infrastructure.

The skill is designed for research-grade tasks: mapping vegetation changes after wildfires, monitoring land cover, computing spectral indices from satellite bands, and exporting large geospatial datasets. Rather than jumping straight to code, GEEMu instructs the assistant to first confirm the user's setup, including the Google Cloud project ID, authentication credentials, Python environment, and whether a network proxy is needed. Then it walks through defining the study area, choosing the right data layers, and thinking through boundary complexity before writing any code.

Inside the folder there are several components. A main instruction file tells the assistant how to approach every task. A references folder contains detailed guidance on proxy configuration, data selection, administrative boundaries, and export strategies. A local knowledge database stored as text files can be searched by keyword to find relevant GEE examples and dataset information. There are also templates for recording run decisions and data layer choices in Markdown, plus complete example workflows for tasks like Sentinel-2 vegetation index calculations, tiled exports for large regions, and fire recovery analysis.

To use it, you clone or download the repository and place the folder in the skills directory that Claude Code or Codex looks for. Because the local knowledge database files are tens of megabytes, the README emphasizes doing a full clone rather than cherry-picking files. Once installed, you tell the assistant to use GEEMu for any Google Earth Engine task, and it follows the structured workflow automatically.

The README is bilingual, written in both English and Chinese.