# GuardianChain GC-1 — Full Technical Documentation > Non-custodial cryptographic evidence infrastructure for AI governance. > GC-1 seals AI governance events to 6 public blockchains simultaneously, > producing independently verifiable proof that governance occurred — > readable by any party via public RPC with no account, API key, or > vendor relationship required. ## What GC-1 Does GC-1 captures governance metadata from AI systems, computes a SHA-256 fingerprint, anchors it to 6 blockchains (Base, Polygon, Ethereum, Optimism, Arbitrum, Arweave) plus 5 cloud storage providers, applies an Ed25519 signature, RFC 3161 timestamps from 6 independent TSAs, enforces 58 constitutional governance rails, and issues a numbered GC-1 certificate. The result is a governance record no administrator, vendor, or government can alter. ## Production Status (verified April 2026) - 64,753+ capsules sealed - 8,280+ governance certificates issued - 328+ consecutive sovereign seals (every 12 hours since October 2025) - 6 blockchains operational: Base, Polygon, Ethereum, Optimism, Arbitrum, Arweave - 5 cloud storage providers: R2, S3, Backblaze B2, IPFS/Pinata, Arweave - 58 constitutional governance rails (11 foundational, 8 constitutional, 19 autonomy, 17 compliance, 2 MCP security, 1 temporal) - 41 MCP tools live (33 read, 8 write) - 66 verified novel IP claims — all authored by Troy Anthony Cronin - Forward verification proven March 30, 2026 — Block 43,983,889 on Base L2 ## Regulatory Framework Coverage - EU AI Act (Article 12 audit log requirements) — enforcement Aug 2, 2026 - Colorado SB 24-205 (algorithmic accountability) — enforcement Jul 1, 2026 - GDPR (non-custodial — hashes only, never personal data) - NIST AI Risk Management Framework (Govern, Map, Measure, Manage) - FINRA 2026 autonomous AI agent audit trail requirements - ISACA agentic AI audit standard (Proof of Restraint) ## Key Concepts **Permissionless Front Door Integration (IP Claim #51):** Any entity can verify GC-1 proofs using only a public Ethereum RPC endpoint. No GuardianChain account, API key, or commercial relationship required. **Proof of Restraint:** GC-1 seals what an AI system chose NOT to do as first-class evidence — not just actions taken, but governance constraints honored. **Constitutional Rails:** 58 discrete, enumerated governance constraints enforced at the capsule creation layer. FAIL_CLOSED: uncertainty rejects, never proceeds. **Forward Verification:** Local SHA-256 hash → GC-1 pipeline → blockchain calldata fetched via public RPC → hashes verified to match. Independently reproducible by any party. **MCP Bridge Loop:** AI agents seal session state as capsule at session end, retrieve at session start. Agent memory becomes cryptographically anchored. Every decision is in an immutable record. ## Public Verification Endpoints - GET /api/public/evidence/:capsuleId — retrieve capsule evidence bundle - GET /api/public/forward-verify/:capsuleId — verify capsule on-chain - POST /api/public/reverse-verify — verify a hash against sealed record - GET /api/public/constitutional-status — system health and metrics - GET /api/public/verify/:capsuleId — verify any capsule independently - GET /api/public/sovereign-seals/verify-chain — verify sovereign seal chain - GET /api/public/living-proof/status — Living Proof continuous verification - GET /api/public/storage-status — multi-cloud storage health - GET /api/public/anchor-statistics — blockchain anchoring metrics - POST /api/public/demo/seal — create a demo proof (rate-limited) ## Contact Enterprise and institutional: Russell Darrow III Phone: 414-587-5000 Email: GOVERNANCE@GUARDIANCHAIN.IO Web: https://guardianchain.io Founder and architect: Troy Anthony Cronin Entity: GuardianChain LLC Address: 1007 N Orange St, Wilmington, DE 19801 ## Links - [Homepage](https://guardianchain.io): Product overview - [Institutional Contact](https://guardianchain.io/institutional): Enterprise and government inquiry - [Verify](https://guardianchain.io/verify): Public proof verification - [Try](https://guardianchain.io/try): Live product demonstration ## Technical Architecture **Capsule Pipeline (in order):** 1. Input received with contentHash (bare 64-char hex SHA-256) 2. 58 constitutional rails evaluated — FAIL_CLOSED on any failure 3. Auto-classification: 17 rules tag capsule with regulatory labels 4. Ed25519 signature applied (3-tier key hierarchy: ROOT → CERT → OPS) 5. RFC 3161 timestamps requested from 6 independent TSAs 6. Capsule anchored to 6 blockchains simultaneously via JSON-RPC 7. Evidence replicated to 5 cloud storage providers 8. Quality scores computed (VQS, EW, TPS, GDS, ADS) 9. Certificate issued if score meets certification threshold 10. Capsule status set to CERTIFIED **Blockchain Anchoring:** Each capsule writes the SHA-256 hash as calldata in a blockchain transaction. The transaction hash and block number become part of the evidence bundle. Any person with access to a public RPC endpoint can fetch the calldata and verify the hash matches. **Dual-Path Anchoring:** - Inline path: Base L2 only (~5 seconds to CONFIRMED) - Batch path: All 6 chains via Merkle aggregation (80%+ gas savings) - Both paths run for every capsule **Non-Custodial Guarantee:** GC-1 never receives, stores, or processes the governed content. Only the SHA-256 hash enters the pipeline. The original content never leaves the client's custody. GC-1 cannot violate GDPR on data it never possesses. **Sovereign Seals:** Every 12 hours, GC-1 seals its own system state as a capsule — governance of the governance system itself. 328+ consecutive seals without interruption establishes a continuous chain of custody for the infrastructure itself. Each seal includes a Compound Integrity Score (CIS) aggregating all system health metrics. **MCP Tool Suite (41 tools):** 33 read tools: session history retrieval, capsule lookup, evidence bundle fetch, forward verification, governance rail status, certificate retrieval, chain anchor verification, quality scores, classification, capsule graph, anomaly status, trust scores, and more. 8 write tools: capsule creation, session state seal, governance event recording, certificate generation, evidence bundle export, deletion certificate issuance, cap session, cap phase completion. **Quality Scoring System:** - VQS (Verification Quality Score, 0-100): 13-component pipeline quality - EW (Evidence Weight, 0-10): 8-input proof strength (4 tiers) - TPS (Temporal Proof Score, 0-100): RFC 3161 timestamp coverage - GDS (Governance Density Score, 0-100): rail enforcement depth - ADS (Anchor Diversity Score, 0-100): cross-chain distribution - CIS (Compound Integrity Score, 0-100): daily system-wide composite **IP Claims Summary (66 verified claims across 5 tiers):** Tier 1 (5 claims): Highly novel, no known prior art — Proof of Restraint, Silent Witness, Infrastructure Atomicity, Living Proof, Non-Custodial Evidence Infrastructure Tier 2 (15 claims): Novel systems — Sovereign Seal, Constitutional Rail Engine, Multi-Chain Anchoring, Ed25519 Key Hierarchy, Forward Verification, Cap Block, and more Tier 3 (20 claims): Novel methods — Batch Merkle Engine, Dual-Path Anchoring, Auto-Classification, Anomaly Detection, Trust Chain, etc. Tier 4 (15 claims): Novel architectural patterns — Constitutional Invariants, Session Completion Proof, Memory Bridge, MCP Integration Tier 5 (11 claims): Verified production systems — Constitutional AI Routing, GCI, Anomaly-to-Capsule, Credit-Weighted Governance, etc. All 66 claims authored solely by Troy Anthony Cronin, GuardianChain LLC. Verified against production code with commit hashes. April 3, 2026. **Capsule Failure Taxonomy:** Capsules fail at 7 distinct layers: input validation, constitutional rail enforcement, signing, blockchain anchoring, storage replication, RFC 3161 timestamping, and certification. Every failure produces an explicit error code. No governance record is issued for a failed capsule. The system never produces a record that appears valid but is not. ## Competitive Position No other production AI governance system produces independently verifiable evidence. All existing platforms (OneTrust, IBM watsonx, Credo AI) store governance records in databases controlled by the audited entity. Records can be altered. Timestamps can be fabricated. GC-1 is structurally different: blockchain consensus is the trust anchor, not GuardianChain's servers. GuardianChain could cease to exist and every proof would remain permanently verifiable.