07 / PUT SHELOB INTO THE WORK

Use Shelob where the work happens.

Query Shelob from the app, an AI agent over MCP, the REST API, or bulk exports for research and model training. The data and labels are the same everywhere.

WHAT IT MEANS

Choose the interface that matches the job, not a different dataset. Workspace, MCP, REST and bulk data return the same resolved places, longitudinal observations, sources and licence records.

SHELOB / connect_shelobILLUSTRATIVE

ASK SHELOB

How do analysts, agents and products use the same spatial evidence without rebuilding the data pipeline?

DATA CONNECTED

WorkspaceShelob skillAuthenticated MCPREST APILicensed data

WHAT COMES BACKInvestigate once. Use the answer everywhere.

Workspace readyClaude + Codex through MCPDeterministic RESTBulk data for frontier AI

02 / WHAT SHELOB DOES

From workspace to agent to production.

Prove it in the workspace

Ask a plain-language question, inspect the joined data on a map and follow every conclusion back to its evidence.

CONNECTS: Place search · maps · trajectories · sources · licence records
WHAT YOU GET: A decision-ready answer before anyone commits to an integration.

Install it in an agent

Add the Shelob skill from a supported marketplace or package, connect the MCP URL and authenticate with the key included in your plan.

CONNECTS: Claude · Codex · IDE agents · Shelob skill · MCP
WHAT YOU GET: Grounded spatial tools available inside the agent you already use.

Ship it with API or data

Call deterministic REST endpoints for production workflows or license attributed datasets for training, evaluation, retrieval and grounding.

CONNECTS: REST · batch queries · exports · bulk data · private overlays
WHAT YOU GET: The same sourced evidence from prototype to product and frontier model.

03 / EXAMPLE ANSWER

How do analysts, agents and products use the same spatial evidence without rebuilding the data pipeline?

DELIVERY PLAN / ONE EVIDENCE LAYERILLUSTRATIVE

Investigate once. Use the answer everywhere.

An analyst investigates six exposed sites in the workspace. A Claude or Codex agent repeats the bounded query through MCP. A product monitors the same portfolio through REST. Every result uses the same place IDs, observation windows, source links and licence records.

04 / PUT IT TO WORK

Start where you work. Keep the same evidence when you scale.

ANALYSTS + TEAMS

Open the workspace and ask

Investigate a site, portfolio or region in plain language. Review the map, trajectory, evidence and data rights before sharing the result.

No integration required · export when readyAsk Shelob about a place ↗

CLAUDE + CODEX + IDE AGENTS

Install the skill and connect MCP

Choose a plan, install the Shelob skill, add the MCP URL and plan key then ask the agent to resolve, compare or verify a place.

Skill package · MCP endpoint · plan credentialsConnect your agent ↗

PRODUCT + DATA TEAMS

Call the same intelligence in production

Use REST for repeatable place, change, evidence and relationship queries. Add private asset data through an isolated overlay when required.

REST endpoints · JSON · batch analysis · exportsSet up the API ↗

FRONTIER AI

License spatial data no single source can provide

Build training, evaluation, retrieval and grounding workflows from joined global spatial data with attribution, provenance and licence records attached.

Coverage and rights scoped with the data teamTalk to our data team ↗

Start with the interface your team, agent or product already uses.

Get early access ↗Connect your agent ↗Frontier AI data ↗