English | Türkçe | Español | 简体中文 | Français | Deutsch | 日本語 | Português (BR) | Русский | 한국어 | العربية | हिन्दी
Holochain-based, agent-centric AI audit infrastructure with EU AI Act compliance.
HoloXtrace provides a decentralized, tamper-proof audit system for AI systems using Holochain's P2P DHT architecture. It eliminates gas fees and provides millisecond latency while maintaining immutability and transparency.
- Agent Identity Management: DeepKey-based cryptographic identities
- DID Resolution: did:holo decentralized identifier support
- SPIFFE/SPIRE Integration: Dynamic workload identity for secure service authentication
- Membrane Proof: Sybil-resistant access control with proof-of-work and SPIRE attestation
Key Features:
- Cryptographic agent verification
- Key rotation support
- SPIFFE ID attestation
- EU AI Act compliance metadata
- Immutable Audit Trail: Source-chain based event logging
- Hash-chain Integrity: Cryptographically linked audit events
- EU AI Act Compliance: Decision metadata, human oversight tracking
- Compliance Reporting: Automated compliance analysis
Key Features:
- Tamper-proof event logging
- Chain integrity verification
- AI system activity tracking
- Anomaly detection
- C2PA Manifest Storage: Detached manifests in DHT
- Metadata Stripping Protection: Immutable provenance data
- Chain of Custody: Complete asset history tracking
- AI Generation Metadata: Training data provenance, model versioning
Key Features:
- C2PA standard compliance
- Manifest verification
- Custody chain tracking
- Metadata stripping detection
Holochain's P2P DHT architecture eliminates transaction fees entirely.
Direct peer-to-peer communication provides near-instant operations.
Source-chain cryptography ensures tamper-proof records.
Built-in support for transparency, accountability, and human oversight requirements.
Membrane proof with SPIRE attestation prevents unauthorized access.
C2PA manifests stored in DHT prevent metadata stripping attacks.
- Rust 1.70+
- Holochain 0.4+ tools (for DNA packaging)
- wasm32-unknown-unknown target
Note: This implementation uses HDK 0.4 APIs which are stable and production-ready. HDK 0.6+ migration would require significant refactoring of entry type definitions (see KNOWN_ISSUES.md for details).
# Install Rust target
rustup target add wasm32-unknown-unknown
# Check that code compiles
cargo check
# Build WASM
cargo build --target wasm32-unknown-unknown --release
# Package DNA (requires Holochain 0.6 tools)
hc dna pack dnas/holotrace/workdirSee BUILD.md for detailed build instructions.
Note: This implementation uses HDK 0.6.0 (latest stable, November 2025) with full API compatibility.
use holotrace::identity_coordinator::*;
let identity = AgentIdentity {
agent_key: my_agent_key,
did_identifier: "did:holo:12345...".to_string(),
spiffe_id: Some("spiffe://example.org/workload".to_string()),
verification_methods: vec![...],
created_at: sys_time()?,
compliance_metadata: ComplianceMetadata {
risk_class: RiskClass::Limited,
purpose: "AI content moderation".to_string(),
transparency_level: "High".to_string(),
},
};
let hash = register_identity(identity)?;use holotrace::audit_coordinator::*;
let event = AuditEvent {
event_id: "event-001".to_string(),
event_type: AuditEventType::Inference,
agent: my_agent_key,
previous_event_hash: Some(previous_hash),
payload: serde_json::to_string(&inference_data)?,
timestamp: sys_time()?,
ai_system_id: Some("ai-model-v1".to_string()),
decision_metadata: Some(DecisionMetadata {
input_hash: "...".to_string(),
output_hash: "...".to_string(),
model_version: "1.0.0".to_string(),
confidence: Some(0.95),
human_reviewed: false,
explanation: Some("High confidence prediction".to_string()),
}),
};
let hash = create_audit_event(event)?;use holotrace::provenance_coordinator::*;
let manifest = C2PAManifest {
manifest_id: "manifest-001".to_string(),
asset_hash: compute_hash(&asset_data),
manifest_data: c2pa_encoded_data,
creator: my_agent_key,
created_at: sys_time()?,
signature: sign_manifest(&manifest_data)?,
custody_chain: vec![...],
};
let hash = store_c2pa_manifest(manifest)?;Holotrace is designed to help organizations comply with the EU AI Act requirements:
- Immutable audit trails
- Complete decision logging
- Human oversight tracking
- Agent-centric architecture
- Cryptographic signatures
- Chain of custody
- Risk classification metadata
- Automated compliance reporting
- Anomaly detection
- Model versioning
- Training data provenance
- Generation parameters
Dynamic workload identity provides:
- Automatic credential rotation
- Mutual TLS authentication
- Zero-trust security model
Membrane proof mechanism:
- Challenge-response authentication
- Optional proof-of-work
- SPIRE attestation validation
- Source-chain validation
- Hash-chain verification
- Digital signatures
- Latency: < 10ms for local operations, < 100ms for DHT operations
- Throughput: Scales with network size
- Storage: Distributed across all nodes
- No Blockchain: Zero gas fees, no mining
┌─────────────────────────────────────────────────┐
│ Holotrace DNA (Holochain) │
├─────────────────────────────────────────────────┤
│ ┌────────────┐ ┌───────┐ ┌─────────────┐ │
│ │ Identity │ │ Audit │ │ Provenance │ │
│ │ Zome │ │ Zome │ │ Zome │ │
│ │ │ │ │ │ │ │
│ │ DeepKey │ │ Hash │ │ C2PA │ │
│ │ did:holo │ │ Chain │ │ Manifests │ │
│ │ SPIFFE/ │ │ EU AI │ │ Custody │ │
│ │ SPIRE │ │ Act │ │ Chain │ │
│ └────────────┘ └───────┘ └─────────────┘ │
├─────────────────────────────────────────────────┤
│ Holochain DHT Layer │
│ (P2P, Zero Gas Fees, Millisecond Latency) │
└─────────────────────────────────────────────────┘
Contributions are welcome! Please ensure:
- All tests pass
- Code follows Rust best practices
- EU AI Act compliance is maintained
- Security considerations are addressed
See LICENSE file for details.
All documentation is available in multiple languages. Click on your preferred language in each document.
DEPLOYMENT_GUIDE.md- Deployment and usage guideARCHITECTURE.md- Detailed technical architectureBUILD.md- Build and testing instructionsTEST_GUIDE.md- Comprehensive testing guideEXAMPLES.md- Complete usage examples
EU_AI_ACT_COMPLIANCE.md- EU AI Act compliance guideSPIFFE_INTEGRATION.md- SPIFFE/SPIRE integration guideQUICK_REFERENCE.md- Quick reference guide
IMPLEMENTATION_SUMMARY.md- What has been implementedKNOWN_ISSUES.md- Known issues and limitationsFINAL_SUMMARY.md- Project summary
🌐 = Available in 12 languages: English (primary), Türkçe, Español, 简体中文, Français, Deutsch, 日本語, Português (BR), Русский, 한국어, العربية, हिन्दी