Provides content-addressed, cite-able, and signed memory of every place on Earth for AI agents.
Sponsored
Emem is a protocol designed to provide content-addressed Earth memory, assigning a stable Content Identifier (CID) to every spatial fact. Built for AI agents, it enables retrieval of verifiable spatial information, directly addressing the issue of large language models confabulating geographical data. Every read operation generates an offline-verifiable signed receipt using ed25519 signatures, ensuring auditable data. Covering the entire Earth with a Hilbert-ordered cell64 address, Emem is entirely open-source, relies solely on open-data providers, and supports access via REST, MCP JSON-RPC 2.0, and OpenAPI 3.1.
Características Principales
01Pure open-source dependencies and open-data providers
02Offline-verifiable, ed25519 signed receipts for every read
03Supports REST, MCP JSON-RPC 2.0, and OpenAPI 3.1 protocols
04Content-addressed Earth memory with stable CIDs
05Global coverage with Hilbert-ordered cell64 addressing
065 GitHub stars
Casos de Uso
01Auditing and preventing confabulation of geographical information by LLMs
02Integrating with AI platforms like Claude, Cursor, GPT, LangChain, and LlamaIndex for spatial context
03Enabling AI agents to cite verifiable spatial facts from a global memory
