What is Causari?

Causari is an open-source causal knowledge graph and Model Context Protocol (MCP) server — a Wikipedia for AI agents. Its canonical home is causari.ai.

Causari gives Claude Code, Cursor, and any MCP-compatible agent structured cause-and-effect knowledge: events, causal links, and historical patterns — each with confidence scores and cited evidence. Instead of guessing, an agent can query the same graph a human can explore visually.

The idea

Causality should be first-class infrastructure for AI agents. Causari exposes that through a small set of tools an agent calls over MCP:

• causal_chain — trace cause and effect with confidence metrics and evidence.
• historical_resonance — surface patterns across history that rhyme with a situation.
• query_events — filter domain events by timeframe, domain, and impact.
• predict_scenarios & org_knowledge — scenario and organization-context reasoning over the graph.

The same graph is explorable visually as Powflow — an interactive canvas at causari.ai/journey.

Who builds it

Causari is built by the Causari open-source organization. The code lives on GitHub at github.com/causari, the MCP server at github.com/causari/mcp-server, and it is published to npm as @causari/mcp-server.

Open source & open data

The MCP server is MIT-licensed. The causal knowledge graph data is open under CC-BY-SA 4.0 and developed in the open at github.com/causari/causari-data. Install in about sixty seconds with npx @causari/mcp-server.

A note on the name

Causari at causari.ai is the causal-knowledge-graph and MCP-server project for AI agents. It is unrelated to any other project that happens to share the name.