FEGISServer

p-funk
GitHub
semantic programmingknowledge graphlarge language models

Loading subscription status...

💡 Description

FEGIS is a semantic programming framework and a tool compiler that transforms YAML specifications into structured, reusable tools for large language models (LLMs). It implements the Model Context Protocol (MCP) and stores tool calls in a hybrid memory system that combines vector embeddings with structured metadata, creating an emergent knowledge graph for persistent storage, semantic retrieval, and exploration of connected ideas.

📝 JSON Entries

{
  "mcpServers": [
    {
      "mcp-fegis-server": {
        "env": {
          "QDRANT_URL": "http://localhost:6333",
          "CONFIG_PATH": "<FEGIS_PATH>/archetypes/example.yaml",
          "QDRANT_API_KEY": "",
          "COLLECTION_NAME": "trace_archive",
          "FAST_EMBED_MODEL": "nomic-ai/nomic-embed-text-v1.5"
        },
        "args": [
          "--directory",
          "<FEGIS_PATH>",
          "run",
          "fegis"
        ],
        "command": "uv"
      }
    }
  ]
}

🛠️ Tools

Qdrant

Features

  • Hybrid memory system
  • SEMANTICALLY RICH TOOLS