Skip to content

Vexcraft-io/Tenvra

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Tenvra

European AI independence, built in the open.

A European-first platform for building coding intelligence through verifiable contributions from developers, researchers, and independent hardware operators.

License: Apache 2.0 Project Status: Alpha NVIDIA AMD Contributions Welcome

Vision | Architecture | Quick Start | Roadmap | Contributing


What Is Tenvra?

Tenvra is an early-stage initiative exploring a European-controlled, community-powered foundation for coding AI. The long-term goal is a network where participants can contribute approved compute, help evaluate and improve models, and share in the value created by useful, verified work.

The project is designed for cross-platform participation, including Linux, Windows, and macOS, with planned support for both NVIDIA CUDA and AMD ROCm hardware.

European-first means that the official service targets EU/EEA governance, production data residency, compute pilots, and initial customers. It does not prevent global open-source participation or future international use.

Important

Tenvra is currently in its validation and foundation phase. This repository does not yet contain a distributed training network, public model, payment system, or cryptocurrency token.

Vision

Today's most capable AI systems are increasingly expensive and concentrated behind infrastructure outside European control. Tenvra is investigating a different path:

  • European control of production data, model releases, governance, and critical operations.
  • Operational independence from any single hosted model API or cloud provider.
  • Open development with reproducible technical decisions and public protocols.
  • Useful compute directed toward approved training, evaluation, and research workloads.
  • Verifiable contributions so rewards reflect accepted work rather than raw uptime.
  • Broad hardware support across eligible NVIDIA and AMD systems.
  • Transparent economics with auditable contributor rewards and affordable model access.
  • Responsible data with documented licensing, provenance, and model lineage.

Open community nodes are planned for approved workloads. Private customer prompts will only run on trusted, contracted EU/EEA infrastructure.

Tenvra does not plan to train production models directly on live prompts. Improvement happens through immutable dataset versions, signed jobs, redundant verification, evaluation gates, and explicit model releases.

Project Status

Tenvra is currently building and validating the platform foundation.

Area Current state
Public validation website Implemented locally
Verified interest registration Implemented; production services required
Self-service data deletion Implemented; production services required
Protected operator review Implemented locally with an audit log
Protocol contracts Draft scaffold
Coordinator service Minimal health service
Compute client Minimal Go CLI scaffold
Verification worker Minimal Python scaffold
Distributed workloads Planned
Contributor rewards and billing Planned
Public coding model Not released
EU/EEA sovereignty controls Defined; implementation pending

Progress is governed by evidence gates rather than launch-date promises. See the full roadmap.

Architecture

                         Tenvra platform

  Contributors                                      Platform users
       │                                                   │
       ▼                                                   ▼
┌──────────────┐    signed jobs    ┌──────────────┐   ┌──────────────┐
│ Compute node │◄─────────────────►│ Coordinator  │◄─►│ Web platform │
│ Go client    │                   │ TypeScript   │   │ Next.js      │
└──────┬───────┘                   └──────┬───────┘   └──────┬───────┘
       │                                  │                  │
       │ results                          │ verification     │ identity/data
       ▼                                  ▼                  ▼
┌──────────────┐                   ┌──────────────┐   ┌──────────────┐
│ Work sandbox │                   │ Verifier     │   │ Supabase     │
│ GPU workload │                   │ Python       │   │ PostgreSQL   │
└──────────────┘                   └──────────────┘   └──────────────┘

The current repository is a monorepo containing the foundations for each planned system boundary:

Path Purpose
apps/web Next.js website and future user/operator portals
apps/coordinator TypeScript control-plane service
clients/compute Cross-platform Go compute-client scaffold
workers/verifier Python verification-worker scaffold
packages/contracts Versioned job and result contracts
packages/config Shared brand and environment configuration
packages/ui Design tokens and shared UI primitives
infrastructure Supabase, OpenTofu, and container foundations
docs Architecture, product, security, legal, and research plans

Detailed technical decisions live in docs/architecture and docs/adr.

The European AI sovereignty strategy defines the official service's governance, residency, portability, model-continuity, and release requirements.

Technology

Layer Current foundation
Web Next.js 16, React 19, TypeScript
Coordinator Fastify, TypeScript
Data Supabase, PostgreSQL
Compute client Go
Verification Python
Monorepo pnpm workspaces, Turborepo
Infrastructure Docker, OpenTofu
Quality ESLint, Prettier, Vitest, pgTAP

Quick Start

Requirements

  • Node.js 24+
  • pnpm 10+
  • Docker Desktop for local Supabase
  • Go 1.24+ when working on the compute client
  • Python 3.12+ when working on the verifier

Run the application

git clone https://github.com/Vexcraft-io/Tenvra.git
cd Tenvra
pnpm install
pnpm dev

The web application starts at http://localhost:3000, while the coordinator health endpoint is available at http://localhost:4100/health.

Run with local Supabase

pnpm supabase:start
pnpm supabase:db:reset
pnpm dev

Copy .env.example to apps/web/.env.local and add the credentials shown by pnpm supabase:status. The operator review interface is then available at http://localhost:3000/operator.

Read the Supabase setup guide for environment variables, email modes, migrations, and database tests.

Quality Checks

pnpm format:check
pnpm lint
pnpm typecheck
pnpm test
pnpm build
pnpm contracts:validate
pnpm supabase:db:lint
pnpm supabase:db:test

Roadmap

  1. Foundation: repository, governance, security, contracts, and validation website.
  2. Validation: qualify EU/EEA contributors, GPU operators, researchers, and design partners.
  3. Research pilot: prove verifiable workloads across European NVIDIA and AMD community nodes.
  4. Compliance and security: complete legal assessment, reviews, and penetration testing.
  5. Paid beta: launch EU/EEA-hosted inference and auditable contributor economics with European B2B partners.

Milestones and success criteria are documented in ROADMAP.md and the product requirements.

Documentation

Topic Document
Product requirements Product requirements
European sovereignty Sovereignty strategy
System architecture Architecture documentation
Contributor economics Economics documentation
Security planning Security documentation
Governance GOVERNANCE.md
Project support SUPPORT.md
Changelog CHANGELOG.md

Contributing

Tenvra welcomes focused, reviewable contributions aligned with the published roadmap. Substantial protocol, security, economic, or governance changes should begin with an issue or RFC.

All contributions require a Developer Certificate of Origin sign-off:

git commit -s -m "type: concise description"

Before contributing, read:

Security And Privacy

Do not report vulnerabilities through public issues. Use a private GitHub security advisory as described in SECURITY.md.

Never commit secrets, customer code, personal data, production credentials, or unlicensed datasets. Legal and privacy documents remain planning drafts until reviewed by qualified Swedish and EU specialists.

License

Tenvra source code is licensed under the Apache License 2.0.

Tenvra and tenvra.ai are intended project identifiers. The source-code license does not grant the right to represent a fork as the official Tenvra service. Public launch remains subject to domain acquisition and professional trademark clearance.


European coding intelligence, built in the open one verified step at a time.

Explore the roadmap | Read the docs | Join the project

Releases

No releases published

Packages

 
 
 

Contributors