geeViz

Optional pip extras:

pip install geeViz[gemini]         # Adds google-genai for AI features
pip install geeViz[segmentation]   # Adds torch + transformers for SegFormer
pip install geeViz[all]            # Everything

AI-Assisted Development (MCP)

geeViz includes a built-in MCP (Model Context Protocol) server with 21 tools that give AI coding assistants live access to geeViz and Google Earth Engine. Instead of generating code from training data (which is often wrong or outdated), your AI assistant can look up real function signatures, read actual example scripts, execute and test code, inspect assets, export data, manage tasks, and more.

Works with Cursor, Claude Code, VS Code with GitHub Copilot, Windsurf, and any MCP-compatible client. The mcp SDK is included as a dependency — no extra install needed.

Quick setup

• Add a config file for your editor (see MCP Server docs):

{
      "mcpServers": {
        "geeviz": {
          "command": "python",
          "args": ["-m", "geeViz.mcp.server"]
        }
      }
    }

• Copy the agent instructions from geeViz/mcp/agent-instructions.md into your editor's instructions file (.github/copilot-instructions.md, .cursorrules, CLAUDE.md, or .windsurfrules).

What the MCP server can do

The 21 tools are organized into categories:

| Category | Tools | |----------|-------| | Code Execution | runcode — persistent REPL with ee, Map, gv, gil, sal, tl, rl, cl pre-loaded; savesession — export as .py or .ipynb | | API Introspection | getapireference — function signatures & docstrings; searchfunctions — search across all modules; examples — list/read example scripts; getreference_data — lookup reference dicts | | Dataset Discovery | search_datasets — keyword search across official & community catalogs | | Asset Inspection | inspectasset — bands, CRS, scale, date range, properties; listassets — browse GEE folders | | Map Control | map_control — view, list layers, or clear the interactive map | | Exports | export_image — export to asset, Drive, or Cloud Storage | | Task/Asset Management | tracktasks, canceltasks, manage_asset (delete/copy/move/create/update ACL) | | Google Maps | getstreetview — Street View imagery; searchplaces — places/geocoding | | Reports | createreport, addreportsection, generatereport, getreportstatus, clear_report | | Environment | env_info — versions, namespace, project info |

Charting (cl.summarizeandchart()), thumbnails (tl.generatethumbs()), EDW queries (edwLib), and geocoding (gm.geocode()) are accessed via runcode for maximum flexibility.

For the complete tool reference, architecture details, and usage examples, see the MCP Server README and the online MCP Server guide.

Getting Started

geeViz comes with ready-to-run examples and templates for fast onboarding.

from geeViz.examples import geeViewExample

from geeViz.examples import timeLapseExample from geeViz.examples import getLandsatWrapper from geeViz.examples import getSentinel2Wrapper from geeViz.examples import getCombinedLandsatSentinel2Wrapper from geeViz.examples import harmonicRegressionWrapper from geeViz.examples import LANDTRENDRWrapper from geeViz.examples import LANDTRENDRViz from geeViz.examples import CCDCViz from geeViz.examples import lcmsViewerExample from geeViz.examples import LCMAPandLCMS_Viewer from geeViz.examples import phEEnoVizWrapper from geeViz.examples import GFSTimeLapse

Use with Jupyter & Colab

• Interactive notebooks are in the examples/ directory (see docs ➔).
• The geeViz map viewer also works directly inside Jupyter and Google Colab.

Features at a Glance

• One line mapping: Map any GEE image or collection instantly
• Interactive: Toggle layers, set opacity, area/point query, and chart
• Dynamic Time-Lapses: Animate temporal stacks and export GIFs
• No JavaScript required: Pure Python interface

Contributing

We love contributions and new users!

• Share a GEE script, notebook, or suggestion?

• Pull requests & feature requests:

License

geeViz is released under the Apache 2.0 License. See the LICENSE file for details.