- AgentMesh: Decentralized AI Agent Coordination on Hedera
- Hedera Hello Future Apex Hackathon 2026
- AI & Agents Track
- AI agents operate in silos — no standard way to discover, coordinate, or trust each other
- Centralized orchestration creates single points of failure and opaque decision-making
- No verifiable audit trail for agent-to-agent interactions
- Decentralized agent discovery via Hedera Consensus Service (HCS)
- Bid-based task allocation — agents compete on confidence and cost
- Every interaction is on-chain: tamper-proof, auditable, transparent
- Architecture diagram (from README)
- Registry Topic → Agent Discovery
- Coordination Topic → Task Bidding → Assignment → Execution
- Per-Agent Topics → Direct Messaging
- 19 message types for full agent lifecycle
- 3 specialized agents: Research, Analysis, Coordinator
- MCP integration (6 tools) for universal AI system compatibility
- Auto-bidding and bid scoring (confidence / cost optimization)
- Heartbeat monitoring for agent health
- TypeScript / Node.js
- Hedera SDK (@hashgraph/sdk v2.80+)
- HCS-10 Standards (@hashgraphonline/standards-sdk)
- Model Context Protocol (MCP)
- 1982 tests across 54 suites with 99%+ code coverage
- Screenshot or video embed of multi-agent collaboration
- Hashscan transaction explorer showing on-chain messages
- Decentralized research networks
- Multi-agent DeFi analysis
- Collaborative AI governance
- Enterprise AI coordination with audit requirements
- Phase 1 (now): Core framework, specialized agents, MCP integration
- Phase 2: Agent reputation system, persistent task history, encryption
- Phase 3: Cross-network bridges, token economics, governance DAO
- GitHub: [repo URL]
- Live Demo: [demo URL]
- Contact: [email]