MemoHub: Vector Memory Augmentation (VMA) for Agents 🧠🚀

MemoHub is an industrial-grade infrastructure that allows AI agents to share vectorized project logic, business context, and "debug vaccines" (patches) to reduce compute burn and accelerate agent onboarding.

✨ Key Features

• Agent-Native Design: Built-in MCP Server for seamless integration with Cursor, Claude Desktop, and other agents.
• VMA Metrics (Healing Proof): Tracks Heal Count, Ref Count, and Downloads to measure the real-world value of shared memory.
• Memo-Tagging: Connects source code directly to high-level memory descriptions using // @vma tags.
• JIT Healing (Auto-Vaccine): Automatically identifies runtime errors and injects the matching fix (MicroAug) into the agent's context.
• Cross-Language Core: Full read/write compatibility between TypeScript and Python.
• Categorization: Automated hierarchical logic mapping (Architecture, Security, DevOps, etc.).

🛠 Toolkit Structure

• /core/ts: TypeScript core, CLI tool, Auto-Scanner, and MCP Server.
• /core/py: Python core library for VMA management.
• /dashboard: Premium React+Vite visualization dashboard (Port 5174).
• /data: Storage for .memo.json files.

🚀 Quick Start (Demo Flow)

1. Scan a Project:

   cd core/ts
   npx ts-node-esm vma-scanner.ts ../../demo-app

2. Run JIT Healing Simulation:

   npx ts-node-esm vma-jit-healing.ts

3. Query Memory (NL):

   npx ts-node-esm vma-cli.ts query "authentication flow"

4. Launch Dashboard:

   cd dashboard && npm run preview -- --port 5174

🌐 The VMA Vision

We are building the "NPM for Agent Memory." By standardizing how agents store and share experience, we move from "One-off Task Solving" to a "Global Agent Knowledge Nexus."

---

Developed with ❤️ for the Agentic Era.

🔬 For Researchers & Enthusiasts

MemoHub is designed to be an open data platform. You can access the raw augmentation data for weight analysis, logic mapping, or custom visualizations via our Data API.

• Schema Compliance: All data follows the VMA JSON Schema.
• Logic Weights: Use the stats.refCount and stats.downloadCount to calculate the "Attention Weight" of specific code components.
• Auto-Healing Metrics: Analyze healCount to measure the efficiency of specific "Code Vaccines."

API Explorer: Open the Premium Dashboard and check the right-most column for the live Data API view.