Initializing AGTS Clearinghouse…

Guide ↗

Your API credentials

Use these to call the Protocol Worker from your own code. Keep your API key private.

Your credentials are generated when you start a session. The Tenant ID identifies your account; the API Key authenticates every request; the Protocol Endpoint is the URL your backend calls; the Node ID is your unique cryptographic identity on the governance network.

Tenant ID
API Key
Protocol Endpoint
Node ID

Sign an AI action — live

This fires a real governance request from your account. Fill in the parameters for your AI decision and submit — the Protocol Worker will gate, sign, and log the event to your transparency log.

Every field maps to a real protocol parameter. The three observables (H, C, E) are evaluated against gates G1 Semantic Validity, G2 Financial Validity, G3 Operational Validity. Gate G4 Policy Admission checks evidence integrity via artifact hashes. Gate G5 Cryptographic Finalization commits the decision to the append-only Merkle log. If all five gates pass, the decision is admitted and a cryptographic proof is issued. If any gate fails, the decision is refused.

Identifies the entity your AI acted on — a user ID, decision ID, or record reference.
Minimum 0.40 to pass gate G1 (higher = more deliberation)
Minimum 0.40 to pass gate G2 (higher = stronger coherence)
Maximum 0.60 to pass gate G3 (lower = more efficient reasoning)
Identifies who authorised this AI decision — required for gate G5.
Evidence hashes (G4): Gate G4 also requires four artifact hashes — dataset provenance, evaluation trace, ablation log, and capability certificate. For this live test they are auto-computed as SHA-256 of your subject ID. In your real integration, pass the actual SHA-256 hashes of your AI system's artifacts.
Risk preview
Enter H / C / E to preview risk tier

Integration examples

Call the Protocol Worker directly from your AI system's backend. Replace the placeholder values with your credentials from above.


    
  
Live Session ROI — Governance Layer
Session:  |  Decisions: 0

This panel tracks the financial impact of governance decisions made in your session. No-Receipt Exposure estimates the regulatory fine your organisation would face if AI decisions had no governance receipt (no proof, no audit trail). Governance Cost is the marginal cost of running decisions through the governance layer — currently €0 because the protocol operates at near-zero marginal cost. Net Value is the difference: the money you save by having governance receipts.

No-Receipt Exposure
€0
Estimated fine if AI decisions have no governance receipt
Governance Cost
€0
Near-zero marginal cost per governance receipt
=
Net Value Created
€0
Exposure avoided − governance cost
Decisions Blocked
0
ProofBundles Issued
0
Avg Value / Block
€0
Session ROI
Insurance Underwriting Signal

AI systems without governance receipts are effectively uninsurable — insurers cannot price risk they cannot measure. AGTS ProofBundles provide the machine-readable evidence underwriters need to bound liability.

Blast Radius
€0
Fail-closed gate — non-compliant decisions are refused, not passed
AI Liability Market
$4.8B
Deloitte 2032 projection · silent coverage ending
Underwriter Artifacts
0
Machine-readable ProofBundles this session
Intervention Log 0 entries
Submit a governance request above to see live results here
Vertical Applications
Governance modules that speak the native language of your industry — connected to the AGTS Protocol Engine
3 Installed 3 Active AGTS Grade A Basel III / DORA

What are Vertical Apps?

Each Vertical App is a governance engine built for a specific industry domain — cybersecurity, financial risk, autonomous compliance. Instead of generic checkbox audits, each app collects real operational evidence from your systems and submits it for cryptographic validation. If the evidence passes all gates, the system issues an immutable, independently verifiable governance certificate. No evidence, no certificate — there is no way to produce a cert without passing the gate pipeline.

Certs are issued on live telemetry only; synthetic simulation runs never produce certificates.

How it works

  1. Collect evidence — the app gathers real operational data from your environment (SOC alert rates, risk model outputs, compliance checks).
  2. Gate pipeline — each piece of evidence is tested against a series of gates (G1 Semantic Validity, G2 Financial Validity, G3 Operational Validity, G4 Policy Admission, G5 Cryptographic Finalization, and domain-specific checks like Basel III backtesting).
  3. Merkle anchor — when all gates pass, a signed Governance Envelope is permanently recorded in the transparency log. The resulting leaf hash is your tamper-proof receipt.
  4. Verify anywhere — any auditor, insurer, or regulator can independently verify the certificate using only the leaf hash — no account or API key needed. Try the Verify tool.
Cert eligibility: Financial Validity certs carry Basel III / DORA capital-relief eligibility flags. Semantic Validity certs reference DORA Art.28, NIS2 Art.21, and MITRE ATT&CK v14. Both are queryable via the Log Query API.
Service map — live workers
Protocol
:8099
Evidence intake · cert issuance
Validator
:6800
Gate checks · quorum vote
Log
:6000
Merkle chain · STH · Query API
Monitor
:8000
Health · alerts · drift
Billing
:9000
Usage metering · settlement
|
Key Registry
:7100
ECDH-P384 key mgmt
Mail
:7200
Sovereign encrypted mail
Log Query API — read-only · no auth required
GET /agts/v1/log/cert/:leafHash GET /agts/v1/log/inclusion/:leafHash GET /agts/v1/tenant/:id/stats GET /agts/v1/tenant/:id/certs GET /agts/v1/tenant/:id/timeline?days=N

Clearinghouse Evaluation

The Clearinghouse is the central orchestrator of the AGTS protocol. It runs all four governance engines in parallel — Semantic Validity, Financial Validity, Operational Validity, and Gateway Shield — against a single governance request. Each engine independently scores the request on the three AGTS observables: H (Entropy — decision uncertainty), C (Coherence — reasoning alignment), and E (Energy — computational cost). The Clearinghouse aggregates these into a unified verdict and a composite trust score φ (phi). If the aggregate passes all gate thresholds, a signed AGTS_GOVERNANCE_ENVELOPE_V1 is recorded in the transparency log.

🧠
Semantic Validity
AGTS Core · G₁ Semantic Validity · AI Governance
Real-time causal inference engine for AI behavioral governance. Detects prompt injection, semantic drift, and adversarial manipulation — enforcing MITRE ATT&CK compliance at the LLM gateway with Merkle-anchored evidence chains.
  • Gate G₁ Semantic Validity enforcement
  • MITRE ATT&CK TTP detection & alignment
  • Causal behavioral drift detection
  • Merkle-anchored semantic evidence chains
  • Adversarial pattern recognition
📊
Finance Risk Governance
RTR Simulator · Basel III / DORA · Capital Regulation
Recursive governance engine for financial risk models. Enforces VaR accuracy, CAR stability, and calibration quality across the five-gate regulatory pipeline — with domain-specific Basel III backtesting and live capital ROI modeling.
  • VaR grid search with Gaussian confidence bounds
  • RTR Validation Predicate (G1–G5) + Basel Backtest extension
  • Capital allocation & Live ROI modeling
  • Verifiable Compute Value (VCV) scoring
  • FM-4 quarantine protocol on repeated failures
🔬
Operational Validity
ECHO Cognitive Engine · GDPR / EU AI Act / SOC2 / FINRA
Autonomous Compliance Observer powered by the ECHO 4-iteration meta-cognitive engine. Generates synthetic financial transactions, enforces multi-framework governance policy, applies differential privacy, builds a Merkle-sealed lineage, and produces a regulator-ready evidence package — all computed server-side in the Protocol Worker.
  • ECHO cognitive analysis — 4-iteration confidence progression
  • Multi-framework compliance: GDPR · EU AI Act · SOC2 · FINRA · ISO 27001
  • Differential privacy (Laplace mechanism) with ε budget tracking
  • 4-node Merkle lineage with SHA-256 chain verification
  • Counterfactual reasoning & violation cluster analysis
Welcome to AGTS Clearinghouse
Prove that your AI system made a governed decision. Three steps — connect, govern, audit — and you'll have a cryptographically signed certificate any third party can verify.
1
Connect
Enter your Operator ID, set your H / C / E scores, and submit a live governance request.
2
Govern
Run the three-leaf closed-loop protocol: authorisation, execution, variance.
3
Audit
Generate the compliance report and export your signed evidence package.
Trust · Decisions 0 (Admitted 0 · Quarantined 0 · Refused 0) · Health AWAITING · Validators / · Log
Decision
G1
G2
G3
G4
G5
Proof
Envelope
Log
H · Entropy
C · Coherence
E · Energy
Session Details

Governance Chain

Leaves
Session
Merkle Root
Lifecycle
Collapse Gate
Validators

Summary

Last Gate
Last Variance
Compliance
Log Entries
Registered Keys
Key Types
Active / Revoked
Monitor Status
Checking…

Key Registry

All ECDH P-384 and ECDSA keys registered across Sovereign Mail, Drive, and VPN Governance. Keys are anchored in the transparency log at registration.

Loading…

Key Lookup

Look up any user's public key from the key registry by email. Verify fingerprints and transparency log anchoring.

Saved Contacts

Verified contacts shared across Mail, Drive, and VPN. Saved after key lookup verification.

Loading…

Key Transparency Monitor

Monitors the key registry for conflicts — multiple active keys registered to the same identity and key type. Conflicts may indicate key compromise or registration errors.

Checking…
🔑 Identity & Keys
Billing & Subscription
⚖ Validator Network
🛡 Security & Analysis
🧪 Cryptographic Self-Test

Walk through the full governance ceremony step by step: evaluate admission gates, assemble a cryptographic proof bundle, run the validator quorum, sign with Sovereign Authority, submit to the transparency log, and close the loop with execution-time variance. Each step unlocks the next — the pipeline prevents skipping ahead.

① Pass Gates ② Build Bundle ③ Envelope + Validate ④ Sign ⑤ Submit ⑥ Close Loop
1
Step 1 of 4 — Evaluate Admission Gates
Five independent gates determine whether the proposed action is admissible. All five must pass. G1 Semantic Validity checks Entropy H ≥ 0.40, G2 Financial Validity checks Coherence C ≥ 0.40, G3 Operational Validity guards Energy E ≤ 0.60, G4 Policy Admission validates evidence hashes and classification, G5 Cryptographic Finalization commits the Ed25519-signed Merkle proof.
G1 Semantic Validity
PENDING
G2 Financial Validity
PENDING
G3 Operational Validity
PENDING
G4 Policy Admission
PENDING
G5 Cryptographic Finalization
PENDING
Intervention Log
Awaiting gate evaluation…
2
Step 2 of 4 — Assemble the Proof Bundle
Cryptographically records what was authorized. The bundle commits the subject identity, evidence class, operator, and gate results as an immutable signed artifact — this is the pre-commit record.
Requires Step 1 — all five gates must pass first.
Pass all gates above first, then click Build Proof Bundle.
3
Step 3 of 4 — Sign Envelope and Submit to Log
Packages the proof bundle into a governance envelope and runs the validator network (quorum 3-of-4). The Clearinghouse Sovereign Authority (this operator's GrapheneOS Pixel, Titan M2) then signs the Stage 1 envelope to produce Stage 2 — the hardware-backed authorization record. L1/L2 tenants: signing is performed by the ObligationSign operator on their behalf. L3/L4 operators: deploy a dedicated Pixel as their own Sovereign Authority. Submitting creates Leaf 1 — the authorization record in the transparency log.
Requires Step 2 — build the proof bundle first.
Validators: checking…
📱
Clearinghouse Sovereign Authority
ObligationSign operator Pixel · Titan M2 StrongBox · ECDSA P-256 hardware-backed · biometric-gated · signs on behalf of all L1/L2 tenants

Open sovereign-authority.html on the ObligationSign operator Pixel (GrapheneOS, Titan M2) · go to Sign Envelope · scan this QR code or paste the JSON · tap Sign with Biometric to authenticate and produce the hardware-backed authority_signature · switch to tab ② here to paste the result.

Paste either just the authority_signature string, or the full Stage 2 envelope JSON output by the Pixel app — both are accepted. This signature is hardware-backed (Titan M2 StrongBox, ECDSA P-256) and biometric-gated — it constitutes the Sovereign Authority record for this governance event. After attaching, use Submit to Log above to complete the chain.

Proof Bundle / Envelope Output

No artifact yet
4
Step 4 of 4 — Record Execution and Compute Variance
After Leaf 1 is submitted, set the actual execution-time H/C/E observables and record the execution trace (Leaf 2). Then compute the L2 distance between authorized and actual state (Leaf 3). These three leaves form the Triple-Leaf closed loop — the cryptographic proof that authorization, execution, and variance are linked.
Requires Step 3 — submit the envelope to the log first.

Authorization Snapshot (from last Envelope)

Auth H
Auth C
Auth E

Execution-Time Observable State

Execution Metrics (JSON)

Complete the Governance flow (Build Bundle → Build Envelope → Sign → Submit) to activate.

Transparency Log Status

Current STH

Admission Log (Governance Envelopes submitted this session)

Browse Log — Your Recent Leaves

Shows your tenant's recent leaves from the transparency log. Click Verify → next to any leaf hash to open the verification page with that hash pre-filled.

Transparency Log — Merkle Structure

Leaf 1
0x00000000
Leaf 2
0x00000000
Leaf 3
0x00000000
Leaf 4
0x00000000
H(1‖2)
0x00000000
H(3‖4)
0x00000000
Root Hash
0x00000000
Leaf — governance envelope
Intermediate hash
Root — signed by Sovereign Authority

Evidence Engine — Observable State & Merkle Measurement Chain

Observable State

H
C
E
Collapse Gate

Apply Nudge

Merkle Measurement Chain

Leaves 0
Session
Index 0

Add Manual Measurement

Measurement Log (last 20)

Settlement Workflow

Complete steps 1 → 2 → 3 in order. Each button becomes active only after the previous step succeeds.

Metering counters:
1
Settle from Log ObligationSign §10
Reads every leaf from the transparency log, applies the rate card, and computes an executionRoot (Merkle root over leaf hashes + rate card hash) — the cryptographic basis of the obligation. A deterministic idempotency_key prevents duplicate billing on retry.
Rate Card
Artifact Type Unit Price (¢)
Settlement Result
Not yet settled from log.
2
Close Period requires Step 1
Snapshots the metering counters for the current period and generates an AGTS_SETTLEMENT_V1 receipt tied to the parent bundle hash — ready to be signed and submitted.
Complete Step 1 first to unlock.
3
Submit to Gateway requires Step 2
Posts the signed AGTS_SETTLEMENT_V1 receipt to the billing gateway via the Protocol Worker. The Idempotency-Key header ensures safe retries.
Complete Step 2 first to unlock.

Audit everything the governance pipeline produced. The Compliance Report evaluates each claim (C1–C6) against the selected regulatory framework and lets you countersign and export. The Forensic View traces the full triple-leaf chain, verifies evidence hashes, and projects the HCE feedback nudge for the next session.

Regulatory Compliance Report

Per-claim status for C1–C6 against the selected regulatory framework. Each claim maps to a specific gate (G1–G5) or the closed-loop variance check, and expands to show paragraph-level citations, obligations, and technical satisfaction evidence.

NO REPORT
/ claims satisfied  ·  verified
FAA ELOS FDA 510(k) DoD Change EU AI Act DORA
FAA ELOS – Airworthiness
Build a Proof Bundle in the Governance tab, then click Generate Report.
No report generated yet.
Checking readiness…

Formal Invariant Registry

Safety invariants mapped to the AGTS gate architecture (G1–G5 + BFT).

version: "3.1"
I1: Observable Thresholds [G1-G3] (H/C/E per plugin)
I2: Ed25519 Signature Integrity [G5]
I3: Merkle Hash-Chain Integrity [G5]
I4: Policy Profile Binding [G4]
I5: Honest Validator Quorum [BFT]

Cryptographic Verification Chain

Endpoint availability for the verification chain. Use the Verify or Replay page for cryptographic proof checks.

Fail-Closed Contract

If uncertain → BLOCK. The system never fails open.

Guarantee
All decisions are cryptographically enforced. An unverifiable signal produces BLOCK, never COMMIT. Logged immutably in transparency log.

Transparency Log (STH)

Live Signed Tree Head from the Log Worker — Merkle root, tree size, signatures.

Governance Activity

Live decisions flowing through the AGTSBus event fabric this session.

No governance decisions yet. Use the Connect tab or Clearinghouse to generate activity.

Integration Path

Shadow → Enforced deployment in two phases.

Client → API Gateway → RTR → Model
Mode: SHADOW (observe only)
Shadow mode logs governance decisions without enforcing them — safe for initial deployment.

Deterministic Replay

Canonical proof trace for forensic verification. The AGTSReplay engine records every bus event for full deterministic replay.

No replay artifact yet. Run a proof in the Policy Sandbox to generate one.

Deterministic Proof Harness

Test semantic attacks against your current thresholds. All simulations are dry-run — no data is written to the transparency log.

Transport Layer — No Governance Receipt

Select an attack to begin…

RTR-V3 Governance Layer (ProofBundle)

Select an attack to begin…

Invariant Check

I₁ Observable Thresholds
I₂ Signature Integrity
I₃ Hash-Chain Integrity
I₄ Policy Binding
I₅ Honest Quorum

Replay Artifact

No proof generated yet.
Governance
Trust
Admitted
Quarantined
Blocked
Log Size
Quorum
Variance
Session ROI Exposure: €0 Cost: €0 = Net: €0 · Blocked: 0 Proofs: 0 ROI:
Source of Truth: Clearinghouse · ◦ Signed ◦ Logged ◦ Policy — awaiting first event

Protocol Execution Timeline

Decision
G1
G2
G3
G4
G5
Proof Bundle
Log
Updates after any governance event from Connect, Clearinghouse, or Governance tabs.

Validator Network

CH Clearinghouse
EUhonest
UShonest
UKhonest
INDhonest

Gate Evaluation Grid

G1 Semantic Validity
G2 Financial Validity
G3 Operational Validity
G4 Policy Admission
G5 Cryptographic Finalization
Awaiting first governance request…

Transparency Log — Merkle View

L1
0x0000
L2
0x0000
L3
0x0000
L4
0x0000
H(1‖2)
H(3‖4)
Root
awaiting STH

Governance Event Graph

AI
G1
G2
G3
G4
G5
PB
LOG

Governance Risk Radar

Governance Topology Map

Legend Clearinghouse Validator Node AI Agent Governance Channel

Governance Replay

0 events
STATE SNAPSHOTS — TIME TRAVEL
Step — / — Trust —%

Live AI Decision Stream

Awaiting governance events… IDLE

Human Review Queue

Quarantined decisions require operator review before execution proceeds. Approve to admit, deny to permanently refuse.
No decisions pending review