compass

Overview

Compass builds a temporal knowledge graph of organizational entities, relationships, and documents across systems. It exposes MCP tools that enable AI agents to search the graph, traverse connections, assess impact of changes, and read attached documents. The system features hybrid search combining keyword precision with semantic similarity, impact analysis capabilities, and supports any entity types with custom properties.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• entity_search

  Search for entities in the knowledge graph using hybrid keyword and semantic search.

  Try:Search for ordersTry:Find all customer entities
• entity_context

  Retrieve context around an entity with multi-hop bidirectional traversal.

  Try:Show context for urn:bigquery:ordersTry:Get context for urn:kafka:events with depth 2
• entity_impact

  Analyze impact of an entity change through dependency relationships.

  Try:Analyze impact of urn:kafka:eventsTry:Show what breaks when urn:bigquery:orders changes

Note: Inferred tool names from CLI examples in the usage section. The README mentions 'composable MCP tools' but doesn't provide explicit documentation of the tool names or signatures.

Comparable tools

Installation

# macOS
brew install raystack/tap/compass

# Linux (Debian)
sudo dpkg -i compass_*.deb

# Docker
docker pull raystack/compass:latest

{
  "mcpServers": {
    "compass": {
      "type": "sse",
      "url": "http://localhost:8080/mcp"
    }
  }
}

compass config init
compass server migrate
compass server start

FAQ

What types of entities can Compass model?

Compass supports any entity type with custom properties - hardware, software, operations, people, or any other organizational concept.

How does the MCP integration work?

Compass exposes MCP tools that AI agents can use to search the knowledge graph, traverse relationships, assess impact, and read documents through Server-Sent Events at http://localhost:8080/mcp.

Compare compass with