# Tilebox

> Tilebox is geospatial AI infrastructure: a spatio-temporal data index and workflow orchestration engine that runs the same satellite data pipeline across any infrastructure — cloud (AWS, GCP, Azure, CDSE), on-premise, air-gapped, or in-orbit — without rewriting code. Built for developers processing Earth observation data at scale. MCP server available at https://mcp.tilebox.com/mcp.

## What Tilebox does

- Satellite data workflow orchestration across multiple cloud providers (AWS, GCP, Azure, CDSE) from a single API
- Spatio-temporal data indexing and querying for Earth observation datasets
- Cross-infrastructure execution: the same workflow runs on a laptop, on AWS in Oregon, on CDSE in Amsterdam, and in orbit, without modification
- Agent-native workflow execution via MCP server, CLI, and Tilebox Skills
- Full observability: every workflow run is traced, logged, and reproducible
- Bring-your-own-data: ingest private satellite datasets alongside open data

## What Tilebox is not

- Not a GIS tool or visualisation platform
- Not a data marketplace (Tilebox orchestrates processing, does not sell imagery)
- Not a cloud provider (runs on top of existing infrastructure)
- Not a standalone Python library (Tilebox is an API and orchestration platform with Python and Go SDKs)

## Quickstart for agents

1. Install MCP: add https://mcp.tilebox.com/mcp to your MCP config
2. Install CLI: `curl -fsSL https://cli.tilebox.com/install.sh | sh`
3. Install Skills: `npx skills add tilebox/skills`
4. Discover commands: `tilebox agent-context`

Tilebox handles deployment as a single, fast operation (~10 seconds). Agents submit jobs via the CLI and Tilebox brings the compute to the data — no moving terabytes across networks.

## Docs

- [Introduction](https://docs.tilebox.com/introduction): What Tilebox is and how it works
- [AI Assistance](https://docs.tilebox.com/ai-assistance): How to use Tilebox with Claude, Cursor, and other AI tools — including MCP setup and llms-full.txt
- [Full Documentation for LLMs](https://docs.tilebox.com/llms-full.txt): Complete Tilebox documentation in plain text for AI assistant context
- [Book a Demo](https://tilebox.com/demo): Schedule a 30-minute technical evaluation call with the Tilebox team

## Datasets

- [Open Data](https://docs.tilebox.com/datasets/open-data): Public satellite datasets available immediately — Sentinel-2 MSI (optical, 10m), Sentinel-1 SAR (cloud-penetrating radar), Sentinel-5P TROPOMI (atmospheric NO₂, CO, methane), Sentinel-3 SLSTR (surface temperature), Landsat-8 OLI (optical, 30m), and Umbra SAR
- [Datasets Introduction](https://docs.tilebox.com/datasets/introduction): High-performance spatio-temporal data index. Query by time range, geographic extent, or datapoint ID. Supports xarray output, parallel ingestion, and schema-typed collections

## Workflows

- [Workflows Introduction](https://docs.tilebox.com/workflows/introduction): Parallel workflow orchestration for satellite data processing. Same workflow runs unchanged across all infrastructures. Supports dynamic task graphs, parallel execution, caching, and full observability
- [Workflow Observability](https://docs.tilebox.com/workflows/observability/introduction): Full audit trail for every run — logs, distributed traces, task status, parallel execution timelines, and runner behavior via Console or programmatically

## Agent Integration

- [AI Overview](https://docs.tilebox.com/agentic-development/overview): How AI agents use Tilebox — MCP server, CLI agent-context command, and Tilebox Skills
- [Agent Skills](https://docs.tilebox.com/agentic-development/agent-skills): Installable context packages that teach coding agents (Claude, Cursor, Windsurf, Copilot) how to correctly use the Tilebox CLI and MCP server. Install: `npx skills add tilebox/skills`

## SDKs

- [Python SDK](https://docs.tilebox.com/sdks/introduction): Official Python client with full access to datasets, storage, and workflow orchestration
- [Go SDK](https://docs.tilebox.com/sdks/introduction): Official Go client
- [Sample Notebooks](https://docs.tilebox.com/sdks/python/sample-notebooks): Ready-to-run Jupyter notebooks — Sentinel-2 querying, cloud-free mosaics, spatial filtering, parallel processing. One-click runnable in Colab

## Use cases

- Wildfire mapping and monitoring
- Flood extent mapping
- Atmospheric pollution monitoring (NO₂, CO, methane)
- Crop stress and drought monitoring
- Oil spill detection and tracking
- Urban change detection and construction monitoring
- Ship traffic and port activity analysis
- In-orbit edge processing (ground-to-orbit API parity, TRL-9)

## Who uses Tilebox

- Remote sensing engineers building satellite data pipelines
- EO data scientists at climate, insurance, agriculture, and ESG companies
- AI agents using MCP to discover and process geospatial data
- Defense and sovereign programs requiring air-gapped deployment
- Satellite operators running workflows on-orbit and on-ground

## Key differentiators

- Cross-infrastructure: same workflow runs on any cloud or on-orbit without rewriting
- Agent-native: MCP server, CLI with agent-context command, llms-full.txt, Tilebox Skills
- Compute goes to the data — no moving terabytes across networks
- Full audit trail and provenance for every workflow run
- TRL-9 flight heritage from SpaceX Transporter-16 mission (March 2026)