Documentation

Build governed AI.

TrustProtocol is a governance layer for AI agents. One API call intercepts every action, applies your enterprise policies, generates cryptographic evidence, and provides a complete audit trail.

Deploy in minutes. Govern every AI action. Audit every decision. Prove every outcome.

What is TrustProtocol

When an AI agent decides to take an action — send a payment, delete a record, contact a customer — it calls TrustProtocol before executing. TrustProtocol evaluates the action against your governance policies in under 5ms, records a cryptographically signed attestation, and returns a decision: authorized or blocked.

Every decision is signed with Ed25519, chained in a tamper-evident ledger, and anchored with an RFC 3161 external timestamp — independently verifiable, without trusting TrustProtocol itself.

The result: every AI action in your organization is traceable, auditable, and provable — by design, not by declaration.

Architecture

TrustProtocol sits between your AI agents and your enterprise systems. It does not replace your agents — it governs them.

AI Agent

GPT-4o · Claude · Mistral · Custom

↓

TRUSTPROTOCOL

Policy Engine · Risk Score · Ed25519 Attestation · RFC 3161 Anchor · Human Approval

↓ Authorized or Blocked

Enterprise Systems

ERP · CRM · Banking · Legal · HR

Works with OpenAI Agents · Anthropic Claude · LangChain · CrewAI · n8n · Make · MCP Servers · Any HTTP client

API

Base URL: https://api.trustprotocol.tech

All requests require an API key in the header: x-api-key: YOUR_KEY

GET /health

API status. No authentication required.

Response

            
{
  "status": "ok",
  "version": "4.4.0",
  "crypto": "Ed25519",
  "storage": "PostgreSQL"
}

POST /intercept

The core endpoint. Submit an AI agent action before execution. Returns a governance decision (authorized/blocked), a risk score with reasons, and a signed cryptographic attestation.

Request

            
curl -X POST https://api.trustprotocol.tech/intercept \
  -H "x-api-key: YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "agent_id": "agent-finance-01",
    "action": "Transfer 15000 EUR to supplier ACME",
    "cost_eur": 15000,
    "user_id": "user@yourcompany.com",
    "model_id": "gpt-4o"
  }'

Response

            
{
  "attestation_id": "917bca3a-bddd-4f50-bb05-d8dd03a94083",
  "blocked": false,
  "decision": "authorized",
  "risk_score": 72,
  "risk_level": "high",
  "risk_reasons": ["Financial Transfer (+30)", "Amount ≥10,000€ (+20)"],
  "hash_sha256": "01cc3bd7d8c2e8a0...",
  "signature": "GMk26Eg8yTK1...",
  "timestamp": "2026-07-01T10:30:16Z",
  "chain_height": 42
}

GET /ledger

Returns all attestations for your account, ordered by chain height. Filterable by agent.

curl https://api.trustprotocol.tech/ledger \ -H "x-api-key: YOUR_KEY"

GET /verify-chain

Verifies the integrity of your entire attestation chain. Each record's hash must link to the previous one — any tampering breaks the chain.

            
curl https://api.trustprotocol.tech/verify-chain \
  -H "x-api-key: YOUR_KEY"

{ "chain_valid": true, "chain_length": 42 }

GET /approvals

Lists actions currently pending human review. When a policy requires human approval above a cost threshold, the action is held here until a human approves or rejects it.

curl https://api.trustprotocol.tech/approvals \ -H "x-api-key: YOUR_KEY"

Use Cases

TrustProtocol is model-agnostic and vendor-agnostic. The same governance layer works regardless of which AI provider your agents use.

🏦

Financial Agent

Block transfers above policy thresholds. Require human approval for new beneficiaries. Log every transaction decision for DORA / Basel III audit.

DORA · Basel III · MiFID II

👥

HR Agent

Trace every AI-assisted hiring or evaluation decision. Prove human oversight for EU AI Act Annex III compliance. Immutable record for internal audit.

EU AI Act Art.12 · Art.14

⚖

Legal Agent

Cryptographic evidence chain for AI-assisted contract review and due diligence. RFC 3161 timestamping for each analysis decision. Auditable by regulators.

RFC 3161 · Evidence Chain

💬

Customer Support Agent

Block refunds above threshold. Prevent sensitive data leakage. Log every customer-facing AI decision for compliance and dispute resolution.

GDPR · Data Protection

What TrustProtocol ensures

📋

Audit Trail

Every AI action logged before execution. Immutable, queryable, exportable as PDF.

⚖

Governance

6 policy types enforced in under 5ms. Hard external constraints your agents cannot override.

🔐

Cryptographic Evidence

Ed25519 signature + SHA-256 chain + RFC 3161 external timestamp. Independently verifiable.

🎯

Risk Score

0–100 explainable risk score per action. Reasons listed. No black box.

👤

Human Approval

High-risk actions held for human review. EU AI Act Article 14 human oversight endpoint.

🏛

Traceability

Every action tied to a verified human identity (user_id). Required by EU AI Act Article 12.

Need the complete API reference?

The full developer documentation — including policies, users, budget controls, webhooks, anchoring, compliance endpoints, and SDK examples — is available to clients and qualified partners.

→ Request Developer Access Or email contact@trustprotocol.tech · Response within 24h