AgentPass -- Secure MCP for Regulated Finance. MCPS + Agent Identity for the Agent Economy. IETF Draft
NEW: Read the AgentPass Case Study — Securing the $5 Trillion Agentic Economy
OPENAPI INITIATIVE · APPROVED EXTENSION
Our x-agent-trust extension is officially registered in the OpenAPI Extensions Registry
First vendor extension specifically designed for APIs serving autonomous AI agents · View live spec · JWKS endpoint · See it in the demo

The security layer for AI agent infrastructure.

Identity, integrity, policy, detection, response — the full runtime control plane for autonomous AI agents. Validated where it counts: our MCPS analyzer is merged into cisco-ai-defense/mcp-scanner (PR #160).

MCPS -- Secure MCP built for regulated finance. Learn more →

OFAC + HMT Sanctions (75K entries)
Mobile SDKs (iOS, Python, Node)
Cryptographically Signed Payments
L0-L4 Behavioural Trust Scoring
Challenge-Response Identity
Hash-Chained Audit Trail
Merged into:
cisco-ai-defense/mcp-scanner ↗
PR #160
SOC 2 Control Mapping
14 Trust Service Criteria mapped to AI agent operations. Submitted to AICPA, CSA, ISACA, and NIST. View mapping
McKinsey, October 2025
$3T–$5T in global agentic commerce value by 2030. Up to $1T of US retail revenue directed by AI agents. 75% of NRF 2026 retailers implementing agentic commerce.
The agentic commerce opportunity · someone has to verify every one of those payments.
iOS SDK for Mobile Agent Payments -- Live Now
Live Demo -- AgentPass iOS SDK
The Agent Runtime Control Plane

The passport is the front door. Behind it, a full runtime control plane.

Identity to response — five layers that govern every AI agent at runtime. Not a black box: built on open standards, with dated patents and IETF drafts behind every layer.

01 · Identity

Who is this agent?

Verifiable agent passports. verifiable agent certificates, challenge-response identity, L0–L4 trust levels, registry resolution.

02 · Integrity

Is the message real?

Every message and tool call signed. Per-call replay protection and tool-hash pinning. MCPS — an IETF Internet-Draft.

03 · Policy

What is it allowed to do?

Authority earned, never assumed. Trust-gated scopes & spend limits, per-agent and fleet policy. Deploy SaaS or fully self-hosted.

04 · Detection

Is it behaving?

Continuous behavioural monitoring — magnitude, velocity, recipient and timing anomalies surface drift and shadow agents; trust adjusts automatically.

05 · Response

Stop it — provably.

Instant kill switch & quarantine, instant certificate revocation, hash-chained tamper-evident audit. Every action reversible and SIEM-ready.

Our Security Pillars
Five layers, built on 30 years of data breach prevention across the industry. Every one anchored in a public record: IETF Internet-Drafts (MCPS, ATTP, AEBA), UK patents with priority dates, the OpenAPI Extensions Registry, an OWASP MCP Top 10 contribution, and citation in academic security research.

Standards & Compliance

Built on open standards. 10 IETF Internet-Drafts including ATTP (Agent Trust Transport Protocol) and MCPS. OWASP MCP Top 10 contributor. OpenAPI Extensions Registry entry. Submitted to EBA, FCA, and PCI SSC.

OWASP MCP Security Cheat Sheet
Contributor -- Section 7: Message Integrity & Replay Protection
IETF Internet-Draft
draft-sharif-agent-payment-trust-00
IETF Internet-Draft
draft-sharif-mcps-secure-mcp
UK Patents (UKIPO)
Multiple patents filed covering agent trust, payment security, and cryptographic signing
FCA Regulatory Sandbox
Application submitted for autonomous agent payment oversight
PCI DSS v4.0.1 Mapping
Complete compliance mapping submitted to PCI SSC
EBA Position Paper
PSD2 position paper on AI agent payments submitted
SOC 2 Agent Controls Mapping
14-control mapping of Trust Service Criteria to AI agent operations. Submitted to AICPA, CSA, ISACA, and NIST.
Cited in Academic Research
MCPS listed as Defense Mechanism D5 in “A Formal Security Framework for MCP-Based AI Agents” (arXiv:2604.05969). Cited alongside Anthropic, Google, Microsoft, and NIST.
SDKs
🐍 Python (PyPI)📦 Node.js (npm)🍎 Swift (iOS)

Signed Payments

Every transaction cryptographically signed. Non-repudiable receipts proving which agent authorised what.

Trust Scoring

5-dimension behavioural trust score (0-100). Agents earn spending authority through proven behaviour.

Spend Limits

Per-transaction and daily limits enforced by trust level. Agents cannot exceed their authority.

Replay Protection

Unique nonce per transaction. Captured payment requests cannot be re-sent.

Audit Trail

Hash-chained tamper-evident log. JSON + RFC 5424 syslog. SIEM-ready.

Anomaly Detection

Magnitude, velocity, recipient, and timing anomalies detected. Trust automatically adjusts.

OFAC + HMT Sanctions

75,784 sanctions entries screened on every payment. UK HMT (57K) + US OFAC SDN (18K). Sanctioned recipients blocked in real time.

Mobile Payments

Native iOS SDK with Keychain-secured ECDSA keys. Python and Node.js SDKs for server-side. Agents pay from any platform.

Agent Registry

DNS for agents. Register, resolve, and search agent identities. Anti-squatting protection. AgentSign-certified.

Deploy Your Way

Cloud or on-premise. Your compliance, your infrastructure.

SaaS

Managed by us. Sign up, get an API key, start verifying agents in minutes. Zero infrastructure.

  • ✓ Free sandbox with $10K test balance
  • ✓ agentpass.co.uk API
  • ✓ Automatic sanctions updates
  • ✓ No ops required
New

Self-Hosted

Deploy in your own infrastructure. Docker image with everything included. Your data never leaves your network.

  • ✓ Single Docker container
  • ✓ Sanctions data baked in
  • ✓ License key activation
  • ✓ Full regulatory control
Contact us for access

Compliance

AgentPass is integrated with FIPS 140-2 Level 3 compliant key management.

AgentPass Self-Hosted

Docker container with license key. Deploy in minutes. Your infrastructure, your control.

Starter
10
agents
  • Built-in CA
  • Trust levels L0-L4
  • Scope enforcement
  • Sanctions screening
  • Agent dashboard
  • Signed audit trail
Get Starter
Enterprise
unlimited agents
  • Everything in Pro
  • Unlimited agents
  • KMS integration
  • AEBA monitoring
  • Dedicated support
  • Custom trust models
Contact Us

Trust Levels

LevelScorePer TransactionDaily LimitUse Case
L00-19$0$0No financial access
L120-39$10$50Micro-payments
L240-59$100$500Standard transactions
L360-79$1,000$5,000Enterprise purchasing
L480-100$50,000$200,000Full access (audited)