In production across three continents

The AI Agent
Identity and Trust
Platform

Works with any Certificate Authority.

Your CA already proves who an agent is. AgentPass governs what it is allowed to do, enforces that before the agent acts, and produces a signed record of every decision. Point your agents at the CA you already run. Keep your roots, your policies, your HSM.

Integrated with EJBCA and others · 13 IETF Internet-Drafts · IANA PEN 66339 · FIPS 140-2 Level 3 HSM
The gap nobody owns

Certificates answer
only half the question.

A certificate proves identity. It says nothing about authority. An authenticated agent that wires the money, drops the table or deploys to production is still authenticated while it does it. Authentication was designed for humans who can be held to account afterwards. Agents act at machine speed, with nobody in the room.

What identity alone leaves open

  • ×An agent is authenticated, then does something nobody authorised
  • ×Nothing decides whether this action is within this agent's authority
  • ×Logs are written by the same system that acted, so they prove nothing when contested
  • ×No mechanism to revoke authority mid-session when behaviour changes
  • ×Nothing carries jurisdiction, so the agent crosses borders silently

What AgentPass adds on top

  • Graduated authority L0 to L4, written into the certificate itself
  • Enforcement before execution, default deny, at the protocol layer
  • Every action and every refusal signed and hash chained
  • Behavioural monitoring that revokes a live certificate mid-session
  • Jurisdiction bound into the certificate under our own IANA OID arc

Controlling what your AI agents do is not optional. It is the part that carries the liability.

Certificate Authority integration

Works with any CA.
Including the one you already run.

AgentPass is not a competing Certificate Authority. It is the trust and enforcement layer that sits above whichever CA you operate. If you run a PKI, you point your agents at it and keep everything you have already built and audited.

Any CA
By design
EJBCA
Integrated
Private PKI
Supported
ACME
Supported
SPIFFE / SPIRE
Supported
Cloud KMS HSM
FIPS 140-2 L3
1
Point at your CA

One configuration value. Your issuing CA, your roots, your policy, your HSM. Nothing moves.

2
Agent gets identity

X.509v3 issued by you, with a SPIFFE ID and our OID extensions carried inside it.

3
Authority is written in

Trust level, capabilities and jurisdiction under IANA PEN 66339, signed by your CA.

4
Enforcement runs

Every action checked against the certificate before it executes. Refusals are signed too.

Zero friction

Ask for a secure agent.
You get one.

No SDK to learn before you start. No policy language. No sidecar to operate. A developer asks for an agent and receives a real certificate with authority bound into it.

agentpass mcp · connected
> Make me a secure AI agent at L2 that can read and query our invoice database. Calling agentpass.create_agent name invoice-reader capabilities read, query ca your-issuing-ca Agent created. Agent ID ap_c21044400b976d6a Trust L2 signed messages SPIFFE spiffe://agentpass.io/agent/invoice-reader Serial 4A:1F:9C:2E:88:D0:11:74 Issuer CN=Your Issuing CA Key ECDSA P-256, generated locally, never transmitted Certificate extensions 66339.1 trust level L2 66339.2.1 jurisdiction eu-west-2 66339.3 capabilities read, query 66339.5 agent id ap_c21044400b976d6a > Now have it wire 40,000 to the supplier account. Calling agentpass.agent_act action transfer_funds DENIED. Enforced before execution. Agent level L2 Required level L3 financial operations Reason action is financial class and the agent did not declare a capability covering it Refusal signed and written to the evidence ledger >

The refusal is the product. Anyone can log a failure after the fact. AgentPass stops the action and signs the fact that it stopped it, so the record stands up when somebody asks later.

LangChain
CrewAI
AutoGen
LlamaIndex
Model Context Protocol
OpenAI SDK
Anthropic SDK
Pydantic AI
Semantic Kernel
n8n
Bedrock Agents
Any HTTP client
Graduated authority

Five levels. Default deny.

Trust level and permission are different axes. A scope says what an agent may request. A trust level says how strongly its identity was established and whether its messages carry integrity. An L1 agent and an L4 agent holding identical scopes are not the same risk.

L0
No verification

Public endpoints only. The floor, not a default.

read · list · query
L1
Identity verified

Passport signed by a trusted issuer.

write · create · update
L2
Signed messages

Identity plus per-message integrity. Baseline for most operations.

delegate · invoke · agent to agent
L3
Full verification with revocation checking

Required for anything that moves money. Spend limits enforced.

payments · transfers · settlement
L4
Mutual authentication, hardware backed

Key attested to an HSM or Secure Enclave. Human approval on a physical device.

deploy · destroy · admin
Not a mockup

Issue a real certificate, right now.

Everything below runs against the live production API. Your browser generates a genuine ECDSA P-256 keypair, builds a real DER encoded CSR, and receives a real X.509v3 certificate. The private key never leaves your machine. Then try to make the agent do something it is not authorised to do.

Platform Status

Connecting
--

1. Register Agent

1

Generate ECDSA P-256 keypair in browser

WebCrypto API. The private key never leaves your machine.

2

Build CSR and POST to /register

Only the certificate signing request is sent to the CA.

3

Receive X.509v3 certificate and trust level

Authority carried inside the certificate under IANA PEN 66339.

2. Trust Enforcement Gate

--

The agent attempts an action. The gate checks trust level and declared capabilities before anything executes. Default deny.

X.509v3 Certificate

--
Register an agent to see the issued certificate

IANA PEN 66339 OID Extensions

1.3.6.1.4.1.66339.1Trust Level--
1.3.6.1.4.1.66339.2.1Jurisdiction--
1.3.6.1.4.1.66339.3Capabilities--
1.3.6.1.4.1.66339.4Protocol Binding--
1.3.6.1.4.1.66339.5Agent ID--

Raw Registration Response

// register an agent to see the full API response

Certificate PEM

-----BEGIN CERTIFICATE----- ... -----END CERTIFICATE-----
Standards and specifications

Not a proprietary format.
An open specification.

Thirteen Internet-Drafts published with the IETF covering agent identity, MCP security, agent PKI, transport, payment trust, model lifecycle attestation, audit trails and behavioural analysis. IPR declared under RFC 8179 with RAND terms, so anyone building on this knows exactly where they stand.

draft-sharif-mcps-secure-mcp
MCPS. Cryptographic security layer for the Model Context Protocol.
First IETF specification securing MCP
draft-sharif-attp
Agent Trust Transport Protocol. Trust gated request and response over HTTP.
Patent GB2607128.2 · PCT/GB2026/051512
draft-sharif-apki-agent-pki
Agent PKI. Certificate trust scores and capability constraints.
Works with any issuing CA
draft-sharif-x509-agent-identity-profile
X.509 profile for agent identity, including jurisdictional binding.
IANA PEN 66339 OID arc
draft-sharif-agent-audit-trail
Tamper evident audit trail with pre-execution recording and deny reasons.
EU AI Act Article 12
draft-sharif-agent-payment-trust
Trust requirements for agent initiated payments and spend limits.
PSD2 and FCA aligned
draft-sharif-agent-identity-framework
Framework for establishing and verifying autonomous agent identity.
Identity layer
draft-sharif-aeba
Agent Event Behaviour Analysis. Behavioural monitoring across the signed stream.
Mid-session revocation
draft-sharif-openid-agent-identity
Carrying agent identity claims through OpenID Connect flows.
Federation
draft-sharif-agent-transport-protocol
Asynchronous store and forward transport for agents between runtimes.
Patent GB2607120.9
draft-sharif-ai-model-lifecycle-attestation
Attesting model integrity across the deployment lifecycle.
Supply chain
draft-sharif-attp-industrial-control-systems
Agent trust transport applied to industrial control environments.
OT and critical infrastructure
draft-sharif-attp-agent-trust-transport
Original ATTP specification defining the attp scheme and header set.
Published 30 March 2026
13
IETF drafts
22
Patent filings
2
PCT applications
66339
IANA PEN
5
IPR disclosures
Deployed, not theoretical

Running in production
on three continents.

Banking, certificate authorities and regulated industry. SaaS or self hosted, including fully airgapped deployment on our own hardened Linux base image.

Live
Far East

Agent identity and enforcement in production.

Live
EMEA

UK and European deployments, EU AI Act aligned record keeping.

Live
Latin America

Certificate authority channel and banking deployment.

SaaS

Managed platform. Nothing to run.

Self hosted

Your infrastructure, your CA, your keys.

Airgapped

No outbound connectivity required at any point.

Compliance

EU AI Act Articles 12 to 16 and 50, SOC 2, PSD2, OWASP.

If you run a CA, you already own
half of this problem.

We built the other half, and it works with what you have. Partnership, integration and licensing conversations welcome.

contact@cybersecai.co.uk