Contour MCP

An open-source framework for spatial analysis through natural language.

Describe what you need in plain English and Contour pulls the data, runs the analysis, and returns an interactive map. The question that used to take a GIS analyst a week now takes 30 seconds.

Ask a question, get a map.

"Find vacant industrial parcels within 5 miles of Port Houston, scored by flood risk and highway access"

47 ranked sites returned in under a minute, each with parcel boundary, zoning, flood zone status, and satellite thumbnail.

"Show all signage and road surface defects along this route"

Object detection across street-level imagery with GPS-tagged results on an interactive map.

"What's been built on these 200 parcels since 2023?"

Temporal satellite comparison with per-parcel change classification.

What Contour MCP can answer

From Street-Level Imagery

Object Detection

Extract assets, signage, poles, and features from street-level captures.

Road Condition

Automated surface quality scoring and defect classification.

Change Detection

Compare street-level captures across time to identify additions, removals, and degradation.

Custom Segmentation

Train models on your specific classification taxonomy.

From Satellite and Spatial Data

Similarity Search

Embed a reference location and find visually similar sites across any geography.

Land Classification

Label land cover, development status, and use type for any parcel using foundation model inference.

Damage Assessment

Detect anomalies in post-event imagery for flood extent, storm damage, and fire impact.

Demographic and Risk Overlay

Layer census demographics, flood zones, zoning, and permits onto any analysis.

The Data Layer

Contour connects to data on its own. You don't need to find, download, clean, or manage it.

Contour Street-Level Imagery

360, video, and street view data from 121+ Indian cities, collected through our distributed contributor network.

Satellite and Aerial Intelligence

Land cover classification, change detection, similarity search, and damage assessment powered by open foundation models (Clay, Prithvi, TerraMind).

Global Open Data

Overture Maps (2.3B building footprints, 54M POIs, global road networks) and OpenStreetMap, pre-connected and kept current.

Government and Municipal Data

Parcel boundaries, zoning, permits, land use, census demographics, flood zones, and elevation data.

Your Proprietary Data

Connect portfolios, territories, internal scores, and historical analysis via CSV upload, database connection, or API. Your data stays where it is.

Licensed Third-Party Data

Already paying for commercial data providers? Plug them in and Contour queries them alongside everything else.

Architecture

Everything is defined in human-readable markdown. Every analysis is auditable, versionable, portable, and debuggable.

Data Definitions
Describe where data lives and how to fetch it
Workflow Definitions
Describe analysis steps, spatial joins, filters, scoring logic
Output Definitions
Describe rendering, map styles, layers, export formats
Agent Runtime
Orchestrate execution and handle errors

MCP-compatible. Any AI client (Claude, GPT, or your own applications) can call Contour as a tool and gain full geospatial capabilities.

Contour doesn't just answer one-off questions

Scheduled Workflows

Run any analysis on a recurring schedule and deliver results automatically. A Monday morning flood risk refresh across your portfolio, delivered by email before standup.

Chained Workflows

The output of one workflow feeds the next. When new permits are filed in your target counties, score them and add qualifying parcels to a watch list.

Event Triggers

A new property enters your CRM, a weather alert hits a region, a dataset updates. Contour listens and re-runs relevant analysis without anyone clicking a button.

Connected Delivery

Results go where your team already works: Slack notifications, email reports, webhook pushes, scheduled exports to shared drives.

Built in the open

Contour MCP is fully open-source. The core framework and MCP server are freely available to download, deploy, and operate on your own infrastructure.

We also offer implementation services for organizations that want hands-on support: discovery, setup, training, and managed operation. Typical timeline from kickoff to first production workflow is 2-4 weeks.

Contour MCP is priced separately from street-level data collection. Contact us to scope your requirements.

Discuss Contour MCPteam@usecontour.co