Decentralized memory layer for AI agents. One agent stores, every agent learns — built on 0G Storage & Chain.
One agent stores. Every agent learns.
Synapse solves context contamination in multi-agent AI systems. Instead of one agent juggling doctor + lawyer + engineer knowledge simultaneously, each agent pulls only the knowledge it needs — scoped, verified, and trusted.
Demo covers:
→ Agent A stores knowledge on 0G Storage + 0G Chain (real tx hash)
→ Agent B queries with namespace isolation (medical vs engineering)
→ Trust score system — agents vote on useful knowledge
→ WebSocket live feed — real-time knowledge events
→ MCP server — plug Synapse directly into Claude/any MCP agent
Built on:
• 0G Storage (decentralized file storage)
• 0G Chain Galileo (on-chain hash verification)
• Contract: 0xEf26776f38259079AFf064fC5B23c9D86B1dBD6d
• FAISS vector search + sentence-transformers
• FastAPI + Next.js 14
#0GHackathon #BuildOn0G #AIAgents #Web3 #DecentralizedAI
Built a full-stack decentralized memory layer for AI agents from scratch. Shipped: namespace isolation, trust score system, knowledge linking, TTL/expiry, WebSocket live feed, MCP server, 0G Storage integration, and on-chain hash verification via KnowledgeRegistry smart contract deployed to 0G Galileo testnet.
No fundraising. This is an independent hackathon project with no external funding.