Getting started
Run the server, install the SDK, and store + recall your first memory. Start here.
Cortadel Docs
Long-term memory for AI agents — a temporal knowledge graph, not a vector blob. MCP, REST and a dashboard in one self-hosted container on .NET 10.
Explore the docs
Section titled “Explore the docs”Self-hosting
The full Docker Compose stack with FalkorDB or Memgraph, and every config knob. Self-host.
MCP integration
Point Claude, Cursor, or any MCP client at the endpoint — no code. Connect an agent.
Framework integrations
Twelve first-party packages — eight TypeScript, three Python, one .NET — that plug Cortadel into LangGraph, the OpenAI Agents SDK, CrewAI, the Microsoft Agent Framework, Mastra, n8n and more, using each framework’s own memory interface. Each has its own page with an install command, a runnable quickstart, every option and its honest limits. Browse all twelve.
.NET SDK
A thin, typed client over the REST API. SDK reference.
What is Cortadel?
Section titled “What is Cortadel?”Cortadel gives an AI agent a durable, queryable memory that survives across sessions — a bi-temporal knowledge graph, not a vector blob. It extracts facts, links entities, resolves contradictions, reconciles duplicate entities, and returns the right context with no LLM call on the read path.
- Two ways in — a typed REST API and a Model Context Protocol (MCP) endpoint.
- One container — API, MCP, and dashboard in a single .NET 10 process over FalkorDB or Memgraph.
- Yours to run — self-hosted; the SDK and these docs are Apache-2.0, the engine is a commercial product. See cortadel.ai.