polyg-mcp

The memory system that understands causality.

Ask "why did auth fail?" and get a traced causal chain — not just similar documents.

Quick Start

1. Start FalkorDB

docker run -p 6379:6379 falkordb/falkordb

2. Run polyg-mcp

npx polyg-mcp

That's it. The MCP server is now running.

Connect to Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "polyg": {
      "command": "npx",
      "args": ["polyg-mcp"],
      "env": {
        "OPENAI_API_KEY": "your-key-here",
        "FALKORDB_HOST": "localhost",
        "FALKORDB_PORT": "6379"
      }
    }
  }
}

What Makes polyg-mcp Different?

Most memory systems return similar documents. polyg-mcp traces causal chains:

"Why did the auth service fail?"

JWT_SECRET removed → deployment missing secret → CrashLoopBackOff → 503s → dashboard down
     ↓ 100%              ↓ 100%                      ↓ 95%           ↓ 90%

Four Graphs Working Together

| Graph | What it answers | |-------|-----------------| | Semantic | "What do we know about X?" | | Entity | "What depends on what?" | | Temporal | "What happened when?" | | Causal | "Why did X happen?" |

Environment Variables

| Variable | Required | Default | Description | |----------|----------|---------|-------------| | OPENAI_API_KEY | Yes | - | For embeddings and LLM calls | | FALKORDB_HOST | No | localhost | FalkorDB host | | FALKORDB_PORT | No | 6379 | FalkorDB port | | EMBEDDING_MODEL | No | text-embedding-3-small | OpenAI embedding model | | LLM_MODEL | No | gpt-4o-mini | OpenAI LLM model |

MCP Tools

Retrieval (6 tools)

semantic_search · entity_lookup · temporal_expand · causal_expand · subgraph_merge · linearize_context

Write (7 tools)

remember · add_entity · add_event · add_fact · add_concept · add_causal_link · link_entities

Admin (2 tools)

get_statistics · clear_graph

Links

• GitHub Repository — Full documentation, contributing guide, and source code
• Report Issues

License

MIT © Captain-Jay29