NOVA Secure File-Sharing

NOVA is a privacy-first, decentralized file-sharing primitive, empowering user-owned AI at scale with encryted data persistence. NOVA enables secure storage and sharing of sensitive data (e.g., datasets for AI agent fine-tuning) without centralized intermediaries, leveraging group key management, IPFS, NEAR smart contracts, and verifiable TEEs via Shade Agents.

NOVA fills critical gaps in NEAR’s ecosystem β€”no native encrypted persistence for TEEs, Intents, or Shade Agentsβ€” while inheriting NEAR’s strengths like sharding for scalability, low-cost transactions (~0.01 NEAR/gas), and AI-native tools (e.g., NEAR AI CLI). Whether you're building AI social platforms, DeFi apps, or autonomous agent workflows, NOVA provides a secure, verifiable data layer.

Why Use NOVA?

  • Privacy-First: Encrypt files with group keys managed off-chain in TEEs, ensuring only authorized users or AI agents access dataβ€”keys never exposed on-chain.

  • Decentralized: Store files on IPFS, log metadata on NEAR’s immutable ledger, and manage access via smart contracts. No central servers.

  • AI-Ready: Seamlessly integrates with NEAR’s TEEs, Intents, and Shade Agents, enabling secure data for AI training and execution.

  • Developer-Friendly: Free-to-integrate SDKs (Rust crate and JS package) with pay-per-action fees baked into the contract, blending into your dApp’s backend.

Key Features

  • Group Creation & Management: Owners (NEAR AccountIds) create groups via smart contracts, supporting collaborative AI training with multi-group membership. Anyone can create groups (per future updateβ€”currently owner-gated for MVP stability).

  • Access Control: Smart contracts maintain a mapping table for members and attestations, ensuring only authorized users access files via ephemeral tokens. Vital for user-owned AI privacy.

  • Secure Storage: Files are encrypted with group keys and pinned to IPFS, optimized for AI dApps (e.g., datasets for fine-tuning).

  • Access Workflow: Authorized users claim nonce-based tokens (ed25519-signed payloads) from on-chain, then fetch keys from TEEs for local decryption, ensuring verifiable access.

  • Revocation & Key Rotation: Remove members and rotate keys in TEEs with lazy re-encryption to minimize latency/gas costs for large groups.

  • Integrity & Trackability: Log signed transactions (with file hashes) on-chain for non-corruption guarantees, leveraging NEAR’s ledger for verifiability.

Group Key Security

Keys are managed off-chain in verifiable TEEs via Shade Agents. Never published on-chain, NOVA file-sharing ensures unbreakable privacy against blockchain fetches.

NOVA's keys are generated, stored, and distributed exclusively within Trusted Execution Environments (TEEs) using Shade Agents. This eliminates any on-chain exposure:

  • Off-Chain Key Management: Keys are derived and encrypted in TEE-secure SQLite databases, accessible only by verified Shade workers (multi-instance with identical code hashes for redundancy and shared access).

  • No On-Chain Keys: The smart contract stores only group metadata, attestations (checksums/code hashes), and used noncesβ€”no keys or decryptable data. RPC queries (e.g., view_state) reveal nothing sensitive.

  • Secure Distribution: Users request ephemeral nonce-based access tokens from the contract (gated by on-chain membership). Tokens incorporate a timestamp-derived SHA256 nonce (preventing replay attacks) and are verified in-TEE before key release, ensuring single-use validity without shared secrets.

  • Verification & Attestation: Every key operation returns a TEE checksum (via agentInfo), proving execution in genuine hardware with unmodified codeβ€”no tampering possible.

  • Rotation & Revocation: On member removal, keys rotate in-TEE (new derivation), invalidating prior access without re-encryption overhead.

  • Attack Resistance: Even targeted attacks (e.g., indexing interactions or RPC dumps) can't extract keys: they're never on-chain. High-value targets (e.g., AI datasets) remain secure against nation-state or sophisticated threats.

NOVA's architecture combined with Shade/TEEs confidentiality provides bullet-proof security for your data: verifiable, private, and resilient, aligning with NEAR's user-owned AI vision.

NOVA x NEAR

NOVA complements NEAR’s AI-focused tools:

  • TEEs: Secures data at rest/transit for confidential compute (e.g., private AI inference in Phala enclaves).

  • Intents: Gates solver access to encrypted payloads, enabling private, AI-driven fulfillment (e.g., cross-chain swaps).

  • Shade Agents: Persists off-chain data for autonomous workers, resolving the "oracle problem" with verified inputs (e.g., prediction markets).

Integration Options

Choose the integration that best fits your use case:

πŸ€– MCP Server - AI Assistant Integration

For AI-assisted workflows using Claude or other MCP-compatible assistants, integrate the publicly deployed https://nova-mcp.fastmcp.app/mcp as a custom connector in your mcp client.

You can also interact with NOVA directly from its multi-user interface at https://nova-sdk.com

Best for: Natural language file operations, AI agent workflows, conversational interfaces

Documentation: /mcp-server | GitBook


πŸ“¦ JavaScript SDK - Web & Node.js

For web applications, backend services, and Node.js environments.

Best for: Web dApps, API servers, browser applications, TypeScript projects

Documentation: /nova-sdk-js | GitBook


πŸ¦€ Rust SDK - System & Smart Contract Integration

For high-performance applications, blockchain integration, and system-level development.

Best for: Smart contracts, CLI tools, high-performance services, native applications

Documentation: /nova-sdk-rs | GitBook


Quick Start Examples

Chat in natural language at https://nova-sdk.com

JavaScript SDK

Rust SDK

Architecture

  1. Client-Side Encryption: Files encrypted locally before upload

  2. IPFS Storage: Encrypted files stored on decentralized IPFS

  3. NEAR Blockchain: Access control groups and transaction logs

  4. Key Management: Group keys managed off-chain in verifiable TEEs via Shade Agentsβ€”never exposed on-chain, distributed via nonce-based access tokens with attestation proofs.

Use Cases

πŸ€– AI & Machine Learning

  • Dataset Sharing: Securely share training data between researchers

  • Model Fine-Tuning: Store and access sensitive data for AI agent training

  • TEE Integration: Provide encrypted inputs/outputs to confidential compute environments

🏒 Enterprise & Collaboration

  • Document Sharing: Secure file sharing within organizations

  • Access Revocation: Remove member access and rotate keys automatically

  • Audit Trails: Immutable transaction logs on NEAR blockchain

πŸ” Privacy-Preserving Applications

  • Healthcare Records: HIPAA-compliant data sharing

  • Financial Data: Secure transmission of sensitive financial information

  • Identity Documents: User-controlled identity verification data

NEAR Token Requirements

Operations require small NEAR token deposits for storage:

  • Register group: ~0.1 NEAR

  • Add member: ~0.0005 NEAR

  • Revoke member: ~0.0005 NEAR

  • Claim token: ~0.001 NEAR

  • Record transaction: ~0.002 NEAR

Ensure your NEAR account has sufficient balance before operations.

Documentation

Comprehensive documentation is available on GitBook:

πŸ“š NOVA Documentation

Security Considerations

⚠️ Important Security Notes:

  1. Private Keys - Never publish NEAR private keys to version control

  2. Key Storage - Keys managed in TEEs; never handle plaintext in code

  3. IPFS Privacy - IPFS content is addressable by CID; encryption is essential

  4. Access Control - Always verify user authorization before operations

  5. Key Rotation - Revoked members cannot decrypt content uploaded after revocation

  6. Client-Side Encryption - Files are encrypted locally, never exposing plaintext to IPFS

  7. Token Ephemerality - Nonces and timestamps prevent replay; refresh tokens frequently

Future Roadmap

Potential Enhancements

  • AI Metadata Extraction: Automate metadata extraction with AI for optimized IPFS indexing.

  • Dataset Monetization: Add pricing at file upload so file owners can monetize their datasets.

  • Per-user rights: So far all group members can upload files in the group. This could be controllable with per-member rights to be set at add member or later updated.

Contributing

Contributions are welcome! We accept contributions for:

  • Bug fixes and improvements

  • New SDK features

  • Documentation enhancements

  • Integration examples

  • Test coverage

How to Contribute

  1. Fork the repository

  2. Create a feature branch (git checkout -b feature/amazing-feature)

  3. Add tests for new functionality

  4. Ensure all tests pass

  5. Commit your changes (git commit -m 'Add amazing feature')

  6. Push to the branch (git push origin feature/amazing-feature)

  7. Open a Pull Request

See individual SDK directories for specific testing instructions.

Resources

NOVA Resources

NEAR Resources

Storage Resources

AI Integration

Support

Need help? We're here for you:

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

Built with ❀️ for the NEAR ecosystem, leveraging:

  • NEAR Protocol for decentralized access control

  • IPFS/Pinata for decentralized storage

  • Shade Agents & TEEs for verifiable key management

  • Model Context Protocol for AI integration


Ready to build privacy-first dApps? Choose your integration: MCP | JavaScript | Rust

Last updated