AtlasPI

AtlasPI is a historical geography API designed for AI agents and digital tools that require structured, verifiable data about past entities, events, and boundaries. It addresses the challenge of large language model hallucinations in historical queries by providing empirically validated, academically sourced information, complete with confidence scores and explicit status fields such as confirmed, uncertain, or disputed. The platform covers over 6,500 years of history and includes data on more than 1,000 historical entities, 643 events, 105 rulers, 1,249 archaeological sites, 55 periods, 252 cities, 41 trade routes, and 107 dynasty chains.

The service is delivered as both a REST API and a native MCP (Model Context Protocol) server, allowing seamless integration with AI agents, including those running in Claude Desktop, Claude Code, and any MCP-compatible client. The API offers over 50 endpoints for querying timelines, comparing entities, searching events, and retrieving data near specific coordinates or years. 1 specifications. Every record includes full citation chains with over 2,400 academic sources, enabling users to trace the provenance and reliability of each fact.

AtlasPI is intended for a range of users, including AI agent developers seeking to reduce hallucinations, digital humanities researchers conducting timeline analysis or cross-cultural studies, EdTech platforms integrating interactive historical maps, and game designers generating period-accurate maps and place names. The platform's ethically designed data model ensures that contested territories are represented with all known versions and that both colonial and original names are documented without arbitrating historical disputes.

0 license, with a free public API and MCP server that require no login, API key, or registration. Users can self-host the core system from GitHub. A premium tier is planned for high-volume access, curated datasets, and enterprise support, but the core functionality remains free and open. AtlasPI positions itself as a factual, machine-first resource for historical reasoning, prioritizing transparency and reproducibility in historical data for AI and digital applications.