MongoDB-Intelligence-Layer

adrianofratelli-glitch/MongoDB-Intelligence-Layer
★ 1 stars Python AI/LLM Updated today
Proof of concept: MongoDB Atlas as the data and orchestration layer for AI — flexible prompt schemas, live model swap with cost projection, and an autonomous support agent that runs a real tool-use loop through the MongoDB MCP Server (find, $vectorSearch, update on Atlas). React + LeafyGreen · FastAPI · Anthropic API.
View on GitHub → 🔍 Audit Wallet Slippage →

Quick Install

Copy the config for your editor. Some servers may need additional setup — check the README.

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "mongodb-intelligence": {
      "command": "uvx",
      "args": [
        "mongodb-intelligence-layer"
      ]
    }
  }
}

Or install with pip: pip install mongodb-intelligence-layer

README Excerpt

Teams shipping AI features usually end up with the "intelligence" of the system scattered everywhere *except* the database: prompts hardcoded in the app, the model name in an env var, a separate vector store for embeddings, a cache in yet another service, and agent memory nowhere at all. Every prompt tweak, model

Tools (11)

agent_memoryagent_sessionsagent_tracesapp_usersarea_profilescache_configguardrail_denylistguardrail_eventsguardrail_policiessemantic_cachesemantic_fail_mode

Topics

anthropicatlasfastapigenaimongodbragreactvector-search