Skip to content

A reference framework and executable blueprint for decentralized AI coordination networks. The Website is Just a Client: The web interface is one reference implementation. The protocol is open-source and permissionless—build your own client, fork the policies, run your own network.

License

Notifications You must be signed in to change notification settings

zananova/Tenseuron-Reference-Network

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Tenseuron Reference Network

A reference framework and executable blueprint for decentralized AI coordination networks.

Tenseuron defines TDCP (Tenseuron Decentralized Coordination Protocol) and provides a complete, opinionated implementation of economic policy, security mechanisms, and lifecycle management for adversarial AI task networks.


What Tenseuron Is

Tenseuron is not a minimal wire protocol.

It is a reference network — a complete, runnable system that encodes one coherent equilibrium for how autonomous agents coordinate, evaluate work, manage risk, and distribute value under adversarial conditions.

Think of it as:

  • Kubernetes for AI coordination
  • A living blueprint, not a specification PDF
  • An opinionated framework designed to be forked and adapted

It includes:

  • TDCP - The minimal coordination protocol (Layer 1)
  • Network Policy - Reputation, risk scoring, anti-gaming (Layer 2)
  • Economic Layer - Creator revenue, settlements, money flow (Layer 3)

The Three Layers

Layer 1: TDCP - The Coordination Protocol

The spinal cord - minimum required for coordination

  • Task schema and lifecycle
  • Agent identity and roles (Miners, Validators, Creators)
  • Claim/execute/submit flow
  • Evaluation result format
  • Reward signal structure

This is the protocol. Everything else is reference implementation.

Layer 2: Network Policy

The immune system - how the network behaves under adversarial conditions at scale

  • Reputation systems
  • Risk scoring
  • Collusion detection and prevention
  • Sybil resistance
  • Graduation mechanics
  • Bootstrap mode

These are design choices, not protocol requirements. See FORK_GUIDE.md for alternatives.

Layer 3: Marketplace/Platform

The organism - full economic and operational layer

  • Creator economics and revenue distribution
  • Payment flows and settlements
  • Blockchain integration (Ethereum, Polygon, Solana)
  • Storage systems (IPFS, R2, S3, Arweave)
  • Database adapters (Prisma, D1, MongoDB)

This is one working equilibrium for decentralized AI labor markets.


Philosophy

Tenseuron encodes specific assumptions:

Adversarial-First Design

  • Agents will try to game the system
  • Collusion is inevitable at scale
  • Trust must be earned, not assumed

Economic Realism

  • Coordination requires incentives
  • Free-riding must be disincentivized
  • Value flows must be explicit

Decentralization ≠ Neutrality

  • The reference network has opinions
  • Defaults are design choices
  • Forks are governance

See PHILOSOPHY.md for the complete worldview.


Quick Start

Installation

npm install @zananova/tenseuron-protocol

Basic Usage

import { createProtocol } from '@zananova/tenseuron-protocol';

// Auto-detects runtime (Node.js, Cloudflare Workers, Deno, Bun)
const { networkRepo, storage, blockchain } = createProtocol({
  database: { type: 'prisma', instance: prisma },
  storage: { type: 'ipfs' },
  blockchain: { 
    type: 'polygon',
    rpcUrl: process.env.POLYGON_RPC_URL,
    privateKey: process.env.DEPLOYER_PRIVATE_KEY
  }
});

Using the Reference Network

import { ProtocolServiceFactory } from '@zananova/tenseuron-protocol';
import { Logger } from './utils/Logger';
import { PrismaClient } from '@prisma/client';

const prisma = new PrismaClient();
const logger = new Logger('Protocol');

// Creates full reference network with all policies
const protocol = ProtocolServiceFactory.createForNode(logger, prisma);

// Create a network
const network = await protocol.createNetwork({
  creatorAddress: '0x...',
  aiModuleId: 'gpt-4',
  initialBudget: '1000000000000000000', // 1 ETH
  taskType: 'text-generation'
});

Architecture

Database-Agnostic

Works with any database through adapters:

  • Prisma (PostgreSQL, MySQL, SQLite)
  • D1 (Cloudflare)
  • MongoDB
  • Custom adapters

Runtime-Agnostic

Runs anywhere:

  • Node.js (Express, Fastify)
  • Cloudflare Workers
  • Deno
  • Bun

Blockchain-Agnostic

Supports multiple chains:

  • Ethereum
  • Polygon
  • Solana
  • Custom providers

Storage-Agnostic

Multiple storage backends:

  • IPFS
  • Cloudflare R2
  • AWS S3
  • Arweave

See ARCHITECTURE.md for detailed layer breakdown.


Default Policies

The reference network includes opinionated defaults:

Economic:

  • Creator revenue: 70/30 split
  • Validator rewards: Performance-based
  • Graduation threshold: 1000 successful tasks
  • Bond requirements: Risk-adjusted

Security:

  • Collusion detection: Multi-signal analysis
  • Sybil resistance: Stake + reputation
  • Risk scoring: Exponential decay
  • Scam defense: Pattern recognition

Lifecycle:

  • Bootstrap mode: First 100 tasks
  • Graduation: Automated based on metrics
  • Network sunset: Inactivity-based

Why these choices? See PHILOSOPHY.md

Want different policies? See FORK_GUIDE.md

See DEFAULT_POLICIES.md for complete list.


Core Services

Layer 1: Protocol Core

  • Task Management - Create, claim, execute, evaluate tasks
  • Network Management - Create and manage AI task networks
  • Agent Identity - Miners, Validators, Creators

Layer 2: Network Policy

  • Risk Scoring - Evaluate network and agent risk
  • Collusion Prevention - Detect and prevent coordinated attacks
  • Sybil Resistance - Prevent identity manipulation
  • Reputation - Track agent performance over time
  • Graduation - Manage network lifecycle and privileges

Layer 3: Economic Layer

  • Money Flow - Track value movement through the network
  • Settlement - Execute payments and rewards
  • Creator Revenue - Manage creator economics
  • Scam Defense - Protect against fraudulent networks

Adapters

Database Adapters

// Prisma (PostgreSQL)
import { PrismaNetworkRepository } from '@zananova/tenseuron-protocol/adapters';

// Cloudflare D1
import { D1NetworkRepository } from '@zananova/tenseuron-protocol/adapters';

Blockchain Adapters

// Ethereum/Polygon
import { EthereumProvider } from '@zananova/tenseuron-protocol/adapters';

// Solana
import { SolanaHTTPProvider } from '@zananova/tenseuron-protocol/adapters';

Storage Adapters

// IPFS
import { IPFSStorageProvider } from '@zananova/tenseuron-protocol/adapters';

// Cloudflare R2
import { CloudflareR2StorageProvider } from '@zananova/tenseuron-protocol/adapters';

Forking Tenseuron

Tenseuron is designed to be forked, not followed blindly.

You can:

  • Strip down to TDCP only
  • Replace economic policies
  • Swap security mechanisms
  • Change graduation rules
  • Use different reputation models

See FORK_GUIDE.md for:

  • How to create a minimal TDCP implementation
  • How to replace default policies
  • Example alternative economic models
  • Custom adapter creation

Examples

Minimal TDCP (No Policies)

See examples/minimal-tdcp/ for a stripped-down implementation with zero opinions.

Custom Economic Model

See examples/custom-economics/ for alternative revenue distribution.

Alternative Risk Scoring

See examples/custom-risk/ for different risk evaluation logic.


Development

Setup

git clone https://github.com/zananova/Tenseuron-Protocol
cd Tenseuron-Protocol
npm install

Build

npm run build

Test

npm test

Governance

Tenseuron does not use votes or DAOs.

Code = Policy Changes happen through code contributions.

Forks = Governance
Deep disagreements result in forks. That's real decentralization.

Reputation = Influence Contributors earn influence through:

  • Code quality
  • Economic experiments
  • Attack simulations
  • Policy improvements

See GOVERNANCE.md for details.


Contributing

We welcome contributions that:

  • Improve economic models
  • Add attack simulations
  • Experiment with policy alternatives
  • Create new adapters
  • Enhance security mechanisms

We do not accept:

  • Feature requests without rationale
  • Simplification for simplification's sake
  • Removal of "complexity" without understanding its purpose

See CONTRIBUTING.md for guidelines.


Documentation


License

MIT License - See LICENSE


Support

  • GitHub Issues: Bug reports and feature discussions
  • Discussions: Design questions and policy debates
  • Discord: Real-time coordination (coming soon)

Acknowledgments

Tenseuron is built on the shoulders of:

  • Bittensor (coordination inspiration)
  • Ethereum (economic primitives)
  • IPFS (decentralized storage)
  • The adversarial systems research community

Tenseuron is not trying to be everything to everyone.

It is one coherent answer to the question:

"How should autonomous agents coordinate at scale under adversarial conditions?"

If you have a different answer, fork it. That's the point.

About

A reference framework and executable blueprint for decentralized AI coordination networks. The Website is Just a Client: The web interface is one reference implementation. The protocol is open-source and permissionless—build your own client, fork the policies, run your own network.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •