MCP Catalogs
Home

novyx-mcp

by novyxlabs·30·Score 46

A comprehensive MCP server providing persistent memory and governance for AI agents with 120+ tools.

ai-llmknowledge-graphdeveloper-tools
8
Forks
1
Open issues
this month
Last commit
2d ago
Indexed

Overview

Novyx-MCP is a sophisticated MCP server offering extensive memory management capabilities for AI agents. It provides both local operation with zero configuration SQLite and cloud-based services for enhanced governance features. The server implements a draft-review-merge workflow that allows for staged memory changes before committing to permanent storage, helping maintain consistency across multiple AI agents. With tools for auditing, rollback, threat intelligence, and agent orchestration, it forms a comprehensive framework for managing AI agent workflows.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

you:Multi-agent collaboration with shared memory spaces and fine-grained permissions
you:Audit trails and rollback capabilities for regulated AI applications
you:Policy-as-code implementation for governed AI actions with approval workflows
you:Does this work without an API key?
you:What is the draft-review-merge workflow?

When to choose this

Choose Novyx-MCP when you need memory persistence across AI agent sessions with governance features and audit trails, especially when working with multiple agents that need shared context.

When NOT to choose this

Avoid if you need an open-source solution without cloud dependencies, or if you require features not covered by their tiered pricing (e.g., enterprise-only interventions, pro+ threat intelligence).

Tools this server exposes

12 tools extracted from the README
  • remember

    Store a memory observation with tags, importance, context, and TTL

  • recall

    Semantic search using natural language to find relevant memories

  • draft_memory

    Create a reviewable draft before writing to canonical memory

  • merge_branch

    Merge every open draft in a branch into the canonical memory

  • create_policy

    Create a custom YAML policy with regex rules and severities

  • action_submit

    Submit an action for policy evaluation and approval

  • approve_action

    Approve or deny a pending action that's awaiting human approval

  • create_agent

    Create a new agent with specific capabilities and configuration

  • threat_record

    Log a threat observation for threat intelligence analysis

  • defense_deploy

    Deploy a new automated defense rule to protect against threats

  • eval_run

    Run a full memory health evaluation to identify issues

  • dashboard

    View aggregated stats including usage, pressure, and governance counts

Comparable tools

semantic-memory-mcpmem0-mcpweaviate-mcpvectordb-mcpmemory-server

Installation

pip install novyx-mcp

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "novyx-memory": {
      "command": "python",
      "args": ["-m", "novyx_mcp"],
      "env": {
        "NOVYX_API_KEY": "nram_your_key_here"
      }
    }
  }
}

FAQ

Does this work without an API key?
Yes, the core memory tools work locally with zero configuration using SQLite. The API key is only needed for cloud-based governance and runtime features.
What is the draft-review-merge workflow?
It's a process where agents first create draft memories, then review them using tools like 'memory_branch' and 'draft_diff', and finally merge them to permanent storage. This prevents direct unreviewed writes to memory.

Compare novyx-mcp with

GitHub →

Last updated · Auto-generated from public README + GitHub signals.