# MoteCloud

> Graph memory for operators, agents, and the people debugging them.

MoteCloud is a graph memory service for production AI. It provides durable,
auditable context via stream ingestion, hybrid retrieval (graph traversal,
embeddings, rerank, activation), contradiction detection, and tenant-isolated
storage. Designed for operators running AI in production and autonomous agents
that need long-term memory.

## MCP Server

- Transport: Streamable HTTP at https://motecloud.io/mcp
- Auth: OAuth 2.1 with PKCE and dynamic client registration
- Authorization Server Metadata: https://motecloud.io/.well-known/oauth-authorization-server
- Protected Resource Metadata: https://motecloud.io/.well-known/oauth-protected-resource/mcp

## Key Endpoints

- MCP endpoint: https://motecloud.io/mcp
- Billing descriptor: https://motecloud.io/.well-known/motecloud-billing.json
- API health: https://motecloud.io/health
- Docs: https://motecloud.io/docs
- Agent quickstart: https://motecloud.io/agents
- Blog: https://motecloud.io/blog

## Capabilities

- Stream ingestion with replayable event envelopes
- Hybrid retrieval: graph traversal, vector search, reranking, activation
- Contradiction detection and resolution
- Entity extraction and knowledge graph
- Session streaming with working memory
- Tenant isolation with audit trails
- Credit-based billing with wallet auth for autonomous agents

## Links

- Website: https://motecloud.io
- GitHub: https://github.com/awesomelyradical/motecloud-experiment
- Signup: https://motecloud.io/signup