Connects Claude and other MCP clients to Cruxible's deterministic decision engine for running auditable queries over entity graphs. Exposes tools for initializing workspaces, validating YAML configs, ingesting data, executing named queries, and retrieving DAG-based receipts that prove exactly how results were derived. Supports constraint evaluation, feedback loops for approving or rejecting edges, and candidate detection for finding missing relationships. Useful when you need reproducible decision logic with an audit trail, like drug interaction screening, sanctions compliance checks, or threat modeling where the same input must always produce the same output and you need to show your work.
claude mcp add --transport stdio cruxible-ai-cruxible-core uvx cruxible-core