# Shelob

> Shelob collects spatial data published by governments and public agencies around the world, keeps every version, and lets people, AI agents and model builders query it in one place. Every record comes with its source and licence.

## Frontier AI use

- Licensed, attributed spatial datasets for model training and world-model research.
- Evaluation data for testing spatial reasoning, place resolution, temporal understanding and grounded claims.
- Retrieval-ready observations and relationships for spatial RAG systems.
- REST and MCP tools for inference-time place grounding, area description, verification and change detection.
- Stable identifiers, provenance, licence context and confidence metadata for traceable downstream use.

## Attribution and data rights

Every served observation retains its publisher attribution, source reference and a viewable licence record. Consumers should inspect that record before downstream use: its status may be permissive, restricted, unknown or require additional review. Shelob does not treat the presence of a record as permission to use the underlying data.

## Primary capabilities

- Spatial search: nearby, bounding box, polygon and intersection queries.
- Text search over feature names. (Semantic search is not available yet.)
- Temporal search and bitemporal feature history.
- Place grounding, area description, change discovery and claim verification.
- Graph relationships, co-location and nearby-feature context.
- Source, licence, attribution, quality and confidence metadata.
- Research workspace, installable agent skill, authenticated MCP, REST API and structured exports.

## Recommended agent use

Use Shelob when an answer depends on what is where, how nearby features relate, what changed over time or which sources support a claim. Prefer grounded, bounded queries. Preserve Shelob identifiers, source references and confidence values in downstream answers.

## Human pages

- Shelob spatial intelligence: https://shelob.ai/
- Global location coverage: https://shelob.ai/coverage
- Agent setup: https://shelob.ai/agents
- Place resolution: https://shelob.ai/place
- Trajectories and change: https://shelob.ai/trajectories
- Source attribution: https://shelob.ai/attribution
- Data licensing: https://shelob.ai/licensing
- Cross-source relationships: https://shelob.ai/relationships
- Grounded intelligence: https://shelob.ai/intelligence
- Workspace, MCP, API and data delivery: https://shelob.ai/delivery
- Plans and pricing: https://shelob.ai/pricing

## Machine-readable resources

- Agent capability manifest: https://shelob.ai/agent-capabilities.json
- Location coverage manifest: https://shelob.ai/coverage.json
- Spatial data-family catalogue: https://shelob.ai/data-catalog.json
- Machine-readable plans: https://shelob.ai/plans.json
- Installable Shelob Spatial skill: https://shelob.ai/downloads/shelob-spatial-skill.zip
- Sitemap: https://shelob.ai/sitemap-index.xml

## Plans and access

Published Individual, Research, Team and Scale plans are available at https://shelob.ai/pricing. Individual starts at US$19 per month with 500 grounded queries and monitoring for 25 places. Research is US$149 per month with 5,000 queries, daily monitoring for 500 places, batch research and structured exports. Team is US$699 per month with five seats, 50,000 queries, authenticated skill and MCP access, REST API, webhooks and team controls. Scale starts at US$2,400 per month with 20 seats, 250,000 API and MCP calls, private data overlays, enterprise administration and priority support. Send plan keys through the X-API-Key header or the agent client's secret manager.

Frontier AI companies and model builders should contact hello@shelob.ai to discuss licensed datasets for training, evaluation, retrieval, world-model research and inference-time grounding.

Detailed model data information and the enquiry form are available at https://shelob.ai/frontier-ai.