Which integrations are available for this server?

Enables LangGraph agents to use structured graph queries for dependency analysis, prerequisite chains, and blast radius computations as part of agent workflows.

How do I use ckg-mcp?

1. Click on "Install Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@ckg-mcp Show prerequisites for Prior Authorization in glp1-obesity" That's it! The server will respond to your query, and you can continue using it as needed. Here is a step-by-step guide with screenshots.

ckg-mcp

mcp-name: io.github.Yarmoluk/ckg-mcp

Compact Knowledge Graph MCP server. Pre-structured domain knowledge as a routing layer for agent stacks — 65× more efficient than RAG on structural queries.

License: MIT Python 3.10+ MCP Compatible

Built on the CKG Benchmark — 45 domains, 7,928 queries, fully reproducible results.

What It Does

Drop CKG into your agent stack as an MCP tool. Instead of retrieving text chunks and hoping the LLM infers structure, CKG gives agents pre-compiled dependency paths, prerequisite chains, and concept relationships — directly from a structured graph.

System	BERT F1	Tokens/query	Hallucination Rate
CKG	0.857	274	0%
RAG	0.817	17,900	Variable
GraphRAG	0.825	—	Variable

65× more efficient per token. Higher accuracy than both RAG and Microsoft GraphRAG. Zero hallucinations by construction.

Install

pip install ckg-mcp

Claude Desktop Configuration

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "ckg": {
      "command": "ckg-mcp"
    }
  }
}

Works with Claude Desktop, LangGraph, AutoGen, or any MCP-compatible orchestrator.

Tools

Tool	Description
`list_domains()`	List all 53 available CKG domains
`query_ckg(domain, concept, depth)`	Extract subgraph — prerequisites + dependents up to N hops
`get_prerequisites(domain, concept)`	Full prerequisite chain to root
`search_concepts(domain, query)`	Find concepts by name

Example

# In your agent — via MCP tool call
query_ckg(domain="glp1-obesity", concept="Prior Authorization", depth=3)

# Returns the full causal chain:
## CKG: Prior Authorization (glp1-obesity)

### Prerequisites (what gates this)
  - Payer formulary tier assignment
    - Cost-effectiveness of GLP-1RA therapy
      - GLP-1 receptor agonist drug class
  - Medical necessity criteria

### Builds toward
  - Formulary position
  - Coverage decision

Same interface for codebases:

query_ckg(domain="langchain-core", concept="RunnableSequence", depth=2)
# Returns exact blast radius — 23 dependent modules — before your agent edits anything

Bundled Domains (53 total)

Life Sciences & Clinical glp1-obesity · glp1-muscle-loss · drug-interactions · dementia · icd10-metabolic · modeling-healthcare-data · payer-formulary · cpt-em-coding · hipaa-compliance

Codebase & Software langchain-core · computer-science · circuits · digital-electronics · blockchain · quantum-computing · claude-skills

Mathematics & STEM calculus · algebra-1 · linear-algebra · pre-calc · geometry-course · chemistry · biology · ecology · genetics · bioinformatics · physics · signal-processing · fft-benchmarking

AI & Data machine-learning-textbook · data-science-course · conversational-ai · tracking-ai-course · prompt-class · intro-to-graph · systems-thinking · microsims

Business & Finance economics-course · personal-finance · organizational-analytics · it-management-graph · unicorns

Education & Other statistics-course · ethics-course · theory-of-knowledge · digital-citizenship · asl-book · reading-for-kindergarten · learning-linux · infographics · automating-instructional-design · functions · us-geography · moss

Enterprise domains (regulatory, legal, financial, custom) → graphifymd.com

Why Not RAG?

RAG retrieves text chunks and forces the LLM to infer structure. On multi-hop structural queries — prerequisites, dependency chains, blast radius — that inference fails.

CKG is a pre-compiled routing layer: dependency paths are already in the graph. BFS/DFS traversal, not similarity search. No hallucinations by construction.

Full benchmark: github.com/Yarmoluk/ckg-benchmark