A European-first platform for building coding intelligence through verifiable contributions from developers, researchers, and independent hardware operators.
Vision | Architecture | Quick Start | Roadmap | Contributing
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.
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.
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.
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.
| 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 |
- 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
git clone https://github.com/Vexcraft-io/Tenvra.git
cd Tenvra
pnpm install
pnpm devThe web application starts at http://localhost:3000, while the coordinator health endpoint is available at http://localhost:4100/health.
pnpm supabase:start
pnpm supabase:db:reset
pnpm devCopy .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.
pnpm format:check
pnpm lint
pnpm typecheck
pnpm test
pnpm build
pnpm contracts:validate
pnpm supabase:db:lint
pnpm supabase:db:test- Foundation: repository, governance, security, contracts, and validation website.
- Validation: qualify EU/EEA contributors, GPU operators, researchers, and design partners.
- Research pilot: prove verifiable workloads across European NVIDIA and AMD community nodes.
- Compliance and security: complete legal assessment, reviews, and penetration testing.
- 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.
| 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 |
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:
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.
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.