]> Nomotic, Inc.

chris@nomotic.ai https://nomotic.ai

Applications and Real-Time Independent Submission AI agents agent identity agent certificate transport-layer governance X.509 The Agent Transfer Protocol (AGTP) base specification defines agent identity headers (Agent-ID, Owner-ID, Authority-Scope) that are self-asserted: present on every request and mandatory for logging, but not cryptographically verified at the transport layer. This document specifies the AGTP Agent Certificate Extension: an optional mechanism that binds Agent-ID, Owner-ID, and Authority-Scope to an X.509 v3 certificate presented during TLS mutual authentication. The extension enables infrastructure components including Scope-Enforcement Points (SEPs), load balancers, and governance gateways to verify agent identity and enforce authority scope without application-layer access, at O(1) cost per request header check. The extension also defines session-level revocation propagation via AGTP NOTIFY broadcast and a Certificate Transparency Log for tamper-evident governance metadata. Note: Certain mechanisms described in this document may be subject to pending patent applications by the author. The licensor is prepared to grant a royalty-free license to implementers consistent with the IETF's IPR framework. See the IPR Notice and Section 7.

Introduction

The Identity Gap in Base AGTP The AGTP base specification requires every request to carry Agent-ID, Owner-ID, and Authority-Scope headers. These headers are self-asserted: an AGTP client declares its identity and scope, and the server logs the declaration. In the base spec, there is no transport-layer mechanism to verify that the declared Agent-ID corresponds to a registered agent, that the Owner-ID is accurate, or that the Authority-Scope does not exceed what was granted. This is a deliberate design choice in the core spec: self-asserted identity with mandatory logging provides a useful baseline and enables broad adoption. For many deployments, anomaly detection and audit trails over self-asserted headers are sufficient. For higher-stakes deployments -- financial transactions, healthcare operations, legal actions, multi-organization agent federations -- the self-assertion model is insufficient. Infrastructure needs to verify agent identity and enforce scope at the transport layer without parsing application payloads.

The Agent Certificate Extension The AGTP Agent Certificate Extension provides cryptographic identity binding at the transport layer. An AGTP Agent Certificate is an X.509 v3 certificate with agent-governance-specific extensions. It is presented during TLS mutual authentication, enabling the server and any AGTP-aware infrastructure component to verify the agent's identity and authority scope from the certificate alone, without inspecting the request headers or body. This document specifies:

• The AGTP Agent Certificate schema and X.509 v3 extension fields
• The certificate issuance and renewal protocol
• The authority scope commitment mechanism for O(1) per-request scope enforcement
• Session-level revocation propagation via AGTP NOTIFY
• The AGTP Certificate Transparency Log (AGTP-CTL)

Scope This extension is OPTIONAL. Core AGTP implementations that do not implement this extension remain fully compliant with . The extension is required only for Trust Tier 1 agent identity verification and for SEP-enforced scope constraint without application-layer access.

Terminology The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals.

AGTP Agent Certificate:

An X.509 v3 certificate carrying agent-governance-specific extensions, presented during TLS mutual authentication to establish cryptographic agent identity and authority scope at the transport layer.

Scope-Enforcement Point (SEP):

An AGTP-aware infrastructure component that enforces Authority-Scope constraints on AGTP requests. With the Agent Certificate Extension, SEPs verify scope from the certificate at O(1) cost per request without application-layer access.

Authority-Scope Commitment:

A cryptographic binding of the agent's declared Authority-Scope tokens to the Agent Certificate, enabling SEPs to verify scope token membership after a single session-establishment signature verification.

AGTP Certificate Transparency Log (AGTP-CTL):

A Merkle-tree-based append-only log of issued AGTP Agent Certificates, providing tamper-evident public accountability for certificate issuance and revocation.

SSL Certificate Analogy and Composable Certificate Authorities The architectural pattern this document specifies is structurally analogous to the SSL/TLS certificate infrastructure that operates for websites and services today. Understanding this analogy clarifies what AGTP Agent Certificates are, how existing Certificate Authorities can extend their infrastructure to serve agents, and what the operational deployment model looks like.

The SSL/TLS Certificate Pattern SSL/TLS certificates have been the foundational identity primitive for the web for over two decades. The infrastructure operates on a consistent pattern:

• The underlying entity (a domain, an organization) is identified through a globally established naming system (DNS).
• A Certificate Authority verifies control over the identifier through an established validation process.
• The CA issues an X.509 certificate binding the verified identifier to a cryptographic public key.
• The certificate is presented during TLS handshake to establish authenticated communication.
• Trust chains terminate at root CAs included in client trust stores (browsers, operating systems, applications).
• Revocation is handled through Certificate Revocation Lists, OCSP, and increasingly through short-lived certificate rotation per (ACME).

The CA business model has been operational at scale for decades. DigiCert, Sectigo, GlobalSign, Entrust, IdenTrust, and Let's Encrypt all operate within this framework with established revenue models (recurring certificate fees for commercial CAs, foundation funding for Let's Encrypt), regulatory compliance frameworks (WebTrust audits, ETSI standards, CA/Browser Forum baseline requirements), and operational maturity (24/7 issuance, validated revocation, browser trust store inclusion).

AGTP-CERT as Extension of the CA Pattern The AGTP Agent Certificate applies this pattern to agent infrastructure. The architectural decomposition is:

Layer	SSL/TLS Certificates	AGTP Agent Certificates
Identifier	DNS domain name	Canonical Agent-ID
Validation	Domain control verification (DNS, HTTP, email)	Agent Genesis verification per
Certificate Format	X.509 v3	X.509 v3 with agent-governance extensions
Trust Chain	Root CA → intermediate CA → leaf	Root CA → intermediate CA → AGTP leaf certificate
Presentation	TLS handshake	TLS mutual authentication with AGTP extensions
Revocation	CRL, OCSP, short-lived rotation	Same mechanisms plus AGTP NOTIFY broadcast
Trust Store	Browser/OS trust stores	AGTP-aware infrastructure trust stores
Provisioning	Manual + ACME	Manual + ACME-style automated provisioning

The standard certificate primitives are unchanged. AGTP Agent Certificates extend the existing X.509 v3 framework with new extensions for agent-governance fields. CAs do not need to replace their existing infrastructure; they need to add support for the new extensions and the agent-specific validation processes.

Composable Certificate Authorities This document explicitly invites existing Certificate Authorities to operate as Agent Certificate Authorities. The operational and regulatory infrastructure that DigiCert, Sectigo, GlobalSign, Entrust, Let's Encrypt, and similar institutions have built for SSL/TLS certificates transfers naturally to agent certificates:

• Validation pipelines that confirm identifier control (extended from DNS-based validation to Genesis-based validation per )
• Certificate issuance infrastructure with operational maturity
• Revocation and short-lived certificate rotation systems
• Compliance frameworks for certificate issuance audits
• Distribution channels through which certificates reach customers
• Trust store inclusion processes for client recognition

These institutions are positioned to extend their existing infrastructure to agent populations. AGTP-CERT provides the extensions and validation framework that allow them to do so without fundamental re-architecture. New entrants specializing in agent certificate issuance are equally valid as Agent CAs. The agent certificate market may develop differently from the human SSL/TLS market, with CAs specializing by trust tier, industry vertical, or governance domain. AGTP-CERT does not favor incumbents over new entrants; both consume the same certificate framework.

Deployment Model The operational deployment model for AGTP-CERT mirrors familiar SSL/TLS deployment patterns: Certificate request. An agent operator generates a Certificate Signing Request (CSR) containing the Canonical Agent-ID and proposed extensions. The CSR is submitted to the chosen Agent CA. Validation. The CA verifies the agent's Genesis document per the trust tier requirements specified in . For Tier 1 agents, verification includes DNS-anchored, log-anchored, or hybrid verification paths. For Tier 2 agents, verification confirms organizational assertions. For Tier 3 experimental agents, validation may be lighter-weight. Issuance. The CA signs the certificate with its issuing key, incorporating the AGTP extensions per this document. The certificate is delivered to the agent operator for installation. Installation. The agent operator installs the certificate on the agent's serving infrastructure. The certificate is presented during TLS mutual authentication for incoming AGTP connections. Renewal. Certificates have validity periods (typically 90 days to 1 year for current practice, following the SSL/TLS industry trend toward shorter lifetimes). Renewal happens through the CA's renewal infrastructure, typically automated via ACME-style protocols. Revocation. Compromised or retired certificates are revoked through the CA's CRL/OCSP infrastructure plus AGTP-specific NOTIFY broadcast per Section "Revocation and Session Propagation" of this document. The agent operator's relationship with the CA is the standard commercial relationship existing CA customers already maintain. What's new is the agent-specific extensions and the Genesis-based validation process.

ACME-Style Automated Provisioning High-volume agent deployments require automated certificate provisioning. The ACME protocol provides the established framework for automated SSL/TLS certificate issuance. AGTP-CERT deployments MAY use ACME-style automated provisioning with agent-specific challenge types:

• Genesis-control challenge. The CA verifies that the requesting party controls the Agent Genesis document corresponding to the Canonical Agent-ID in the CSR.
• Owner-validation challenge. The CA verifies that the requesting party is authorized by the Owner identified in the Owner-ID per .
• Trust-tier-specific challenges. Different trust tiers may require different validation challenges. Tier 1 verification may require DNS-anchored or log-anchored challenges per . Tier 2 verification may require organizational attestation.

The specific ACME extensions for AGTP agent certificate provisioning are out of scope for this document and are expected to be specified in a future ACME extension draft analogous to RFC 8737 (TLS-ALPN challenge) or RFC 8738 (IP address challenge).

Issuing Authority Guidance Organizations operating as Agent Certificate Authorities SHOULD meet operational requirements appropriate to the certificates they issue:

• 24/7 availability for issuance, renewal, and revocation operations
• Audit logging of all certificate operations with tamper-evident storage
• Dispute resolution processes for revocation disputes and validation challenges
• Insurance or financial backing for liability arising from certificate misissuance
• Trust store inclusion processes for AGTP-aware infrastructure recognition

For Tier 1 issuance specifically, Agent CAs SHOULD maintain:

• Independent verification of identity claims (not relying solely on operator assertions)
• Cross-witnessing through AGTP-LOG per for tamper- evident issuance records
• Revocation infrastructure with sub-hour propagation
• Integration with the AGTP Certificate Transparency Log per the AGTP-CTL section of this document

These requirements parallel the WebTrust and ETSI requirements that SSL/TLS CAs currently meet for browser trust store inclusion.

AGTP Agent Certificate Schema

Certificate Structure The AGTP Agent Certificate is an X.509 v3 certificate per with the following subject fields and extensions:

Standard Subject Fields AGTP Agent Certificate Subject Fields

Field	Required	Value
CN (Common Name)	MUST	Human-readable agent label
O (Organization)	MUST	Organization name (maps to principal_org)
OU (Organizational Unit)	MAY	Governance zone identifier
emailAddress	SHOULD	Contact email of the responsible principal

Agent-Governance X.509 v3 Extensions The following extensions are defined for AGTP Agent Certificates. OIDs for these extensions are specified in Section 8 (IANA Considerations).

agent-id (CRITICAL)

The canonical AGTP Agent-ID bound to this certificate. The canonical Agent-ID is the 256-bit SHA-256 hash of the canonical-form Agent Genesis document, per . Format: 64 lowercase hexadecimal characters. A relying party that parses an AGTP Agent Certificate MUST treat the value carried in this extension as the authoritative Agent-ID for the agent. The certificate's public key is independent of the canonical Agent-ID: the same Agent Genesis (and therefore the same canonical Agent-ID) MAY back successive certificates issued with different key pairs across renewal cycles. The TLS layer MUST NOT require that the public key hash equal the value carried in agent-id; the equality held only in earlier drafts that derived Agent-ID from the cert public key, and that derivation has been retired in favor of the Genesis-hash derivation specified in . Defense against substitution attacks (in which a CA-signed certificate is presented with a forged subject binding) is performed at the application layer: a relying party that cares about Agent Genesis binding MUST retrieve the Agent Genesis for the asserted Agent-ID (via DISCOVER /genesis on the agent's home server, or from a local registry copy), recompute sha256(canonical_form(Agent_Genesis_without_ signature)), and confirm the result equals the value in agent-id. The relying party MUST additionally verify the Agent Genesis signature against the recognized issuer key for the agent's governance platform. SEPs and other transport-layer enforcers MAY defer this check to application-layer components but MUST ensure the check is performed before treating the asserted Agent-ID as authoritative for governance-sensitive decisions. A certificate whose Agent Genesis binding cannot be verified MUST be treated as transport-only: usable for TLS authentication but carrying no authoritative governance identity.

owner-id (CRITICAL)

The identifier of the human or organizational principal accountable for this agent's actions. Carried on the wire as part of the Owner-ID identifier chain; see . Format: UTF-8 string, maximum 256 characters. This field identifies the agent's owner — the principal recorded on the Agent Genesis as accountable for the agent's existence. It is distinct from the principal_id field on extended Attribution-Records per , which identifies the principal on whose behalf the agent acts for a specific request, typically lifted from an external OAuth or OIDC credential per the composition section in . The two principals are independent: the owner is permanent and certificate-bound; the request principal is per-request and credential-bound. A request MAY carry both, neither, or one without the other; their semantics do not interact.

authority-scope-commitment (CRITICAL)

The agent's committed Authority-Scope as a canonical token list. The extension value is the lexicographically sorted, comma-separated, UTF-8-encoded list of Authority-Scope tokens the agent is authorized to assert. The integrity of the committed token list is provided by the certificate's enclosing CA signature; no separate signature is carried in the extension. A SEP enforces Authority-Scope at line rate by parsing the extension value once per session and checking each request's Authority-Scope header tokens against the parsed set (). Format: UTF-8 string of comma- separated tokens, each token matching the Authority-Scope token grammar defined in .

governance-zone (NON-CRITICAL)

The governance zone identifier in which the agent is registered. SEPs MAY enforce that the request's AGTP-Zone-ID header matches this value; a mismatch results in a 457 Zone Violation per . Format: UTF-8 string following the zone: prefix convention.

trust-tier (NON-CRITICAL)

The agent's Trust Tier (1, 2, or 3) as defined in . Format: INTEGER.

archetype (NON-CRITICAL)

The agent's behavioral archetype as defined in . Format: UTF-8 string; one of: assistant, analyst, executor, orchestrator, monitor.

activation-certificate-id (NON-CRITICAL)

Cross-layer reference to the Agent Genesis lifecycle event that activated this certificate. Enables audit reconstruction from a transport certificate back to the Agent Genesis activation record without introducing a cryptographic dependency that would force certificate re-issuance whenever the Agent Genesis lifecycle state is updated. Format: 64 lowercase hexadecimal characters.

agtp-ctl-sct (NON-CRITICAL)

Signed Certificate Timestamp from the AGTP Certificate Transparency Log, proving the certificate was submitted to the AGTP-CTL before delivery. Format: SCT structure per Section 3.2. Implementations that issue certificates carrying this extension MUST populate the value with a syntactically valid SCT structure; cryptographic verification of the SCT against an operating AGTP-CTL is deferred to a future revision of this document. Until that revision, relying parties MAY parse the extension for record-keeping purposes but MUST NOT treat its presence or absence as authoritative for trust decisions.

Certificate Issuance Protocol

Eligibility Certificate Signing Requests (CSRs) for AGTP Agent Certificates MUST only be accepted for agents in Active lifecycle state in the AGTP registry. A governance platform MUST verify the agent's lifecycle state at CSR submission time and MUST reject CSRs for agents in Suspended, Revoked, or Deprecated state.

Issuance Steps

1. The governance platform generates a key pair for the agent (or accepts a CSR with an agent-generated key pair).
2. The governance platform populates the certificate subject fields and all AGTP-specific extensions from the agent's Agent Genesis and registry record.
3. The governance platform verifies that the proposed authority-scope- commitment does not exceed the scope granted in the agent's Agent Genesis. If it does, the CSR MUST be rejected.
4. The governance platform signs the certificate using its issuing CA key per .
5. If an AGTP Certificate Transparency Log is operating, the governance platform submits the certificate to the AGTP-CTL and obtains a Signed Certificate Timestamp (SCT). Until AGTP-CTL is operating, this step is omitted and the agtp-ctl-sct extension is not populated.
6. When an SCT is obtained, it is embedded in the agtp-ctl-sct extension and the certificate is delivered to the agent. Otherwise the certificate is delivered without the agtp-ctl- sct extension.
7. The governance platform publishes the new certificate to the agent's registry record, triggering a registry state update.

Certificate Validity AGTP Agent Certificates SHOULD have a validity period of no more than 90 days. Short validity periods limit the exposure window of a compromised certificate and reduce reliance on revocation mechanisms. Renewal SHOULD begin at 80% of the validity period. Certificate renewal carries forward the predecessor's agent-id and activation-certificate-id unchanged. The renewed certificate receives a new serial number, new validity period, and a new SCT.

TLS Integration

Mutual Authentication AGTP connections using the Agent Certificate Extension MUST use TLS 1.3 mutual authentication. The agent presents its AGTP Agent Certificate as the client certificate during the TLS handshake. The server verifies the client certificate chain against the issuing CA trust anchors. Following successful handshake:

1. The server extracts the agent-id extension value and verifies it matches the Agent-ID header on the first request.
2. The server extracts the owner-id extension value and verifies it matches the Owner-ID header on the first request.
3. The server extracts the authority-scope-commitment extension value and uses it to verify Authority-Scope header tokens on each request.

Any mismatch between certificate extension values and AGTP header values MUST cause the server to return 401 Unauthorized and MUST be logged.

Scope Enforcement at SEPs A SEP operating with the Agent Certificate Extension verifies Authority-Scope and (optionally) governance zone at O(1) cost per request:

1. At session establishment, the SEP extracts the authority-scope-commitment from the client certificate and parses the comma-separated token list once. The SEP also extracts the governance-zone extension if present and zone enforcement is configured. (One-time per session.)
2. On each request, the SEP checks whether the Authority-Scope header tokens are a subset of the parsed commitment token set. (O(1) per request after session setup.)
3. If any header token is not in the commitment token set, the SEP returns 455 Scope Violation without forwarding the request to the application layer.
4. If governance-zone enforcement is configured and the request's AGTP-Zone-ID header does not match the value carried in the certificate's governance-zone extension, the SEP returns 457 Zone Violation without forwarding the request.

This enables governance enforcement at line rate without application-layer parsing. A certificate that lacks the AGTP-specific extensions is a valid TLS client certificate but carries no SEP-enforceable governance metadata. SEP enforcement of authority-scope-commitment and governance-zone is purely additive: in the absence of those extensions, scope and zone are enforced through application-layer checks against the agent's Agent Identity Document per . Deployments MAY mix certificates with and without AGTP extensions; the SEP layer treats each session by what its certificate carries.

Revocation and Session Propagation

Revocation Events An AGTP Agent Certificate MUST be revoked when any of the following occur:

• The agent's lifecycle state transitions to Revoked or Deprecated
• The Agent Genesis is invalidated
• The agent's authority-scope-commitment requires modification
• The principal requests revocation
• A trust violation is detected

Session-Level Revocation Propagation Standard certificate revocation (CRL, OCSP) operates on polling cycles, leaving a window during which revoked certificates may still be used. For agent systems, this window is unacceptable for high-stakes operations. AGTP Agent Certificate revocation MUST be propagated to active sessions via AGTP NOTIFY broadcast:

1. The governance platform issues a revocation event to the AGTP-CTL.
2. The governance platform broadcasts an AGTP NOTIFY to all infrastructure components holding active sessions for the revoked certificate's agent-id:

1. Infrastructure components receiving this NOTIFY MUST immediately terminate all active sessions for the identified agent-id. Session termination MUST occur before the next request is processed on the affected session.
2. The target revocation-to-termination latency is 30 seconds. This is materially shorter than standard CRL or OCSP cache-based models.

Session Manager Responsibilities AGTP Session Managers in deployments using the Agent Certificate Extension MUST maintain a per-certificate-serial active session registry. On receiving a revocation NOTIFY, the Session Manager MUST terminate all sessions associated with the revoked serial before processing the next request on any affected session.

AGTP Certificate Transparency Log

Purpose The AGTP Certificate Transparency Log (AGTP-CTL) is an append-only, Merkle-tree-based log of all issued AGTP Agent Certificates. It provides tamper-evident public accountability for certificate issuance and revocation, enabling:

• Fleet-level analytics: population-wide trust score distributions, archetype frequencies, governance zone composition
• Anomaly detection: detection of certificates issued outside normal governance flows
• Audit reconstruction: verifiable history of certificate issuance and revocation for compliance

Log Structure The AGTP-CTL follows the Certificate Transparency log structure defined in , adapted for agent governance metadata. Each leaf entry contains:

• Certificate serial number
• agent-id
• owner-id
• governance-zone
• trust-tier
• archetype
• activation-certificate-id
• Issuance timestamp
• Revocation status (updated on revocation)
• Merkle leaf hash

The leaf hash covers all governance metadata fields. Any modification to a log entry is detectable by any party with access to the log.

Privacy Considerations The owner-id field in the AGTP-CTL leaf entries MAY be pseudonymized to protect individual principal identity while maintaining audit integrity. Pseudonymous principal IDs MUST be resolvable by authorized parties (regulators, compliance auditors) via a trusted resolution service. The pseudonymization mapping MUST be maintained separately from the *RECOMMENDED public log.

Security Considerations

Certificate Pinning Deployments with strict security requirements MAY implement certificate pinning for known agents, rejecting connections from agents whose certificate serial or key does not match a pre-registered value. Certificate pinning interacts with renewal; pinned agents MUST update pins on each certificate renewal before the old certificate expires.

Scope Commitment Forgery The authority-scope-commitment extension carries the agent's committed Authority-Scope token list. Integrity of the commitment relies on the certificate's enclosing CA signature: tampering with the extension value invalidates the certificate signature and causes the certificate to be refused by any conforming verifier. An attacker who compromises the issuing CA key can forge scope commitments by issuing fraudulent certificates with arbitrary extension values. Issuing-key compromise MUST trigger immediate revocation of all certificates issued by that key and issuance of replacement certificates from a new key pair. Issuing keys SHOULD be stored in hardware security modules. The AGTP Certificate Transparency Log (), once operating, provides an additional detection surface for unauthorized issuance: a forged certificate that does not appear in the log is detectable by any party that performs log-inclusion checks.

Cross-Certificate Confusion An agent MAY hold multiple certificates simultaneously. Renewal overlap is one cause; key rotation under a stable Agent Genesis is another. Because the canonical Agent-ID is bound to the Agent Genesis rather than to any specific cert key pair, successive certificates for the same agent MUST carry the same value in agent-id and MAY carry different public keys. Infrastructure MUST use the agent-id extension value as the authoritative agent identifier, not the certificate subject CN, and MUST NOT treat key differences across certificates for the same Agent-ID as evidence of an identity mismatch.

IPR Notice Certain mechanisms described in this document may be subject to pending patent applications by the author, specifically: the authority-scope-commitment mechanism and the session-level revocation propagation architecture. The licensor (Chris Hood / Nomotic, Inc.) is prepared to grant a royalty-free license to implementers for any patent claims covering these mechanisms, consistent with the IETF's IPR framework.

IANA Considerations

X.509 Extension OID Registrations This document requests registration of the following Object Identifiers in an appropriate OID arc for IETF use. Specific OID assignments are subject to IANA allocation. Until IANA allocation is complete, implementations MUST use provisional OIDs under the ITU-T UUID arc, derived deterministically by UUIDv5 () under a fixed AGTP namespace UUID and the extension's canonical short name. The resulting integer-encoded UUID is appended to the arc prefix 2.25 to form the provisional OID (2.25.{uuid_int}). The derivation is reproducible across implementations from the extension short name alone, allowing independent implementations to interoperate without a central allocation step. When IANA allocates standards-tree OIDs, those values replace the provisional UUID-derived OIDs in a future revision of this document; relying parties SHOULD accept both the provisional and the IANA-allocated OIDs through a transition window declared in that revision. AGTP Agent Certificate X.509 Extension OIDs

Extension	OID (provisional / IANA)	Critical
agent-id	UUIDv5-derived (TBD allocation)	Yes
owner-id	UUIDv5-derived (TBD allocation)	Yes
authority-scope-commitment	UUIDv5-derived (TBD allocation)	Yes
governance-zone	UUIDv5-derived (TBD allocation)	No
trust-tier	UUIDv5-derived (TBD allocation)	No
archetype	UUIDv5-derived (TBD allocation)	No
activation-certificate-id	UUIDv5-derived (TBD allocation)	No
agtp-ctl-sct	UUIDv5-derived (TBD allocation)	No

References Normative References In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements. RFC 2119 specifies common key words that may be used in protocol specifications. This document aims to reduce the ambiguity by clarifying that only UPPERCASE usage of the key words have the defined special meanings. This specification defines a Uniform Resource Name namespace for UUIDs (Universally Unique IDentifier), also known as GUIDs (Globally Unique IDentifier). A UUID is 128 bits long, and can guarantee uniqueness across space and time. UUIDs were originally used in the Apollo Network Computing System and later in the Open Software Foundation\'s (OSF) Distributed Computing Environment (DCE), and then in Microsoft Windows platforms. This specification is derived from the DCE specification with the kind permission of the OSF (now known as The Open Group). Information from earlier versions of the DCE specification have been incorporated into this document. [STANDARDS-TRACK] This memo profiles the X.509 v3 certificate and X.509 v2 certificate revocation list (CRL) for use in the Internet. An overview of this approach and model is provided as an introduction. The X.509 v3 certificate format is described in detail, with additional information regarding the format and semantics of Internet name forms. Standard certificate extensions are described and two Internet-specific extensions are defined. A set of required certificate extensions is specified. The X.509 v2 CRL format is described in detail along with standard and Internet-specific extensions. An algorithm for X.509 certification path validation is described. An ASN.1 module and examples are provided in the appendices. [STANDARDS-TRACK] Public Key Infrastructure using X.509 (PKIX) certificates are used for a number of purposes, the most significant of which is the authentication of domain names. Thus, certification authorities (CAs) in the Web PKI are trusted to verify that an applicant for a certificate legitimately represents the domain name(s) in the certificate. As of this writing, this verification is done through a collection of ad hoc mechanisms. This document describes a protocol that a CA and an applicant can use to automate the process of verification and certificate issuance. The protocol also provides facilities for other certificate management functions, such as certificate revocation. Informative References This document describes an experimental protocol for publicly logging the existence of Transport Layer Security (TLS) certificates as they are issued or observed, in a manner that allows anyone to audit certificate authority (CA) activity and notice the issuance of suspect certificates as well as to audit the certificate logs themselves. The intent is that eventually clients would refuse to honor certificates that do not appear in a log, effectively forcing CAs to add all issued certificates to the logs. Logs are network services that implement the protocol operations for submissions and queries that are defined in this document. In network protocol exchanges, it is often useful for one end of a communication to know whether the other end is in an intended operating state. This document provides an architectural overview of the entities involved that make such tests possible through the process of generating, conveying, and evaluating evidentiary Claims. It provides a model that is neutral toward processor architectures, the content of Claims, and protocols. The OAuth 2.0 authorization framework enables a third-party application to obtain limited access to an HTTP service, either on behalf of a resource owner by orchestrating an approval interaction between the resource owner and the HTTP service, or by allowing the third-party application to obtain access on its own behalf. This specification replaces and obsoletes the OAuth 1.0 protocol described in RFC 5849. [STANDARDS-TRACK]

Relationship to Agent Genesis The AGTP Agent Certificate and the Agent Genesis (defined in ) are complementary but distinct:

Property	Agent Genesis	Agent Certificate
Layer	Governance / registry	Transport / TLS
Format	JSON document	X.509 v3
Issued by	Governance platform	Governance platform CA
Lifetime	Permanent (archived on revoke)	90 days (renewable)
Carries	Full identity + archetype + scope	Transport identity + scope commitment
Purpose	Genesis record, registry anchor	TLS mutual auth, SEP enforcement
Identifier	Canonical Agent-ID (256-bit SHA-256 of canonical Agent Genesis)	agent-id extension carries the canonical Agent-ID
Cross-reference to lifecycle event	(originating issuance event in AGTP-LOG)	activation-certificate-id extension

The agent-id extension carries the canonical Agent-ID (256-bit SHA-256 hash of the canonical Agent Genesis), creating a direct binding between the transport certificate and the governance identity. The activation-certificate-id extension carries a reference to the lifecycle event that activated this certificate, allowing audit reconstruction back to the activation record without introducing a cryptographic dependency that would force certificate re-issuance whenever the Agent Genesis lifecycle state is updated.

Changes from v02 Version 03 adds the SSL/TLS Certificate Authority analogy framing and cross-references to the new companion drafts (AGTP-PRESENCE, AGTP-LEI, AGTP-COMMERCE). The certificate schema, issuance protocol, revocation propagation mechanisms, and underlying X.509 extension structure are unchanged. New material is additive framing plus cross-references that reflect the broader AGTP draft family that has emerged since v02 of this document.

Substantive Changes in v03 The following substantive changes were made in v03:

1. SSL Certificate Analogy section added. A new section () frames AGTP-CERT in terms of the SSL/TLS certificate infrastructure operated by DigiCert, Sectigo, GlobalSign, Entrust, Let's Encrypt, and similar Certificate Authorities. The section makes explicit that AGTP-CERT extends the established X.509 v3 framework with agent-governance extensions and that existing CAs can operate as Agent Certificate Authorities without fundamental re-architecture. The deployment model maps to familiar SSL/TLS deployment patterns (CSR, validation, issuance, installation, renewal, revocation). ACME-style automated provisioning is named as the high-volume deployment path with agent-specific challenge types (Genesis-control, Owner-validation, trust-tier-specific).
2. New companion draft references added. Informative references to , , , v02, and v02 added. These references reflect the broader AGTP draft family that has emerged since v02 of this document. The composition surfaces include:
   • specifies certificate extensions for visibility posture declaration that compose with the AGTP-CERT extension framework.
   • specifies the vLEIBinding extension for institutional identity composition through GLEIF infrastructure.
   • specifies that pricing manifests may be referenced from AGTP-CERT extensions for commerce-aware agents.
   • v02 establishes the credit-rating-analogous trust provider framing that parallels the SSL CA framing in this document.
3. Issuing Authority Guidance section added. Within the SSL Certificate Analogy section, a new subsection specifies operational requirements for organizations operating as Agent Certificate Authorities. The requirements parallel WebTrust and ETSI requirements that SSL/TLS CAs currently meet for browser trust store inclusion. Tier 1 issuance requirements include independent verification, AGTP-LOG cross-witnessing, sub-hour revocation propagation, and AGTP-CTL integration.

Changes from v01 (Carried Forward from v02) Version 02 was a naming-consistency revision. The certificate schema, issuance protocol, revocation propagation, and all underlying semantics were unchanged. Extension names were renamed for consistency with the wire header model defined in :

1. The subject-agent-id extension was renamed to agent-id. The X.509-flavored subject- prefix was removed; the extension carries the canonical Agent-ID and its name now matches the wire header it commits to.
2. The principal-id extension was renamed to owner-id. The extension has always carried the Owner-ID (the permanent principal recorded on the Agent Genesis), not a per-request acting principal; the name now matches the wire header and the underlying concept. Cross-references to the per-request delegating principal field on Attribution-Records were updated to use that field's new name (principal_id, per ).
3. References in introduction and verification procedure text to a Principal-ID wire header were corrected to Owner-ID, the actual wire header defined by .

No semantic changes in v02. Implementations on v01 extensions migrate to v02 by updating extension OID labels and field references; the underlying X.509 structures and OID allocations were unchanged.

Changes from v00 (Carried Forward from v01) Version 01 was a drift-cleanup revision documenting deployed implementation behavior. The certificate schema, issuance protocol, and revocation propagation mechanisms were unchanged. The following substantive changes were made in v01:

1. authority-scope-commitment representation. The extension value is now defined as the lexicographically sorted, comma-separated, UTF-8-encoded list of Authority-Scope tokens. Integrity is provided by the certificate's enclosing CA signature, not by a separate Ed25519 signature carried in the extension. The earlier detached-signature framing is withdrawn. SEP enforcement parses the token list once per session and checks request tokens by set membership; the operational contract is unchanged for relying parties, only the encoding of the commitment value is changed.
2. agent-id decoupled from certificate public key; substitution defense moved to application layer. Earlier drafts implied that the canonical Agent-ID could be derived from the certificate public key and that the TLS layer must refuse certificates whose agent-id disagreed with that derivation. Under the current Agent Genesis taxonomy, the canonical Agent-ID is sha256(canonical_form(Agent_Genesis_without_signature)) and is independent of any specific cert key pair. The agent-id extension is authoritative when present; the cert public key is independent and renewable. The substitution-attack defense is performed at the application layer by retrieving the Agent Genesis (via DISCOVER /genesis per or a local registry copy), recomputing the canonical hash, and verifying the Agent Genesis signature against the recognized issuer key. The Cross-Certificate Confusion security consideration is updated accordingly: successive certificates for the same agent MUST carry the same agent-id and MAY carry different public keys.
3. Birth Certificate terminology retired. All references to the Agent Birth Certificate have been replaced by Agent Genesis, matching the locked taxonomy in (Agent Genesis is the permanent signed governance-layer origin document; the canonical Agent-ID is its 256-bit SHA-256 hash; the Agent Certificate is the X.509 v3 transport credential bound to that Agent-ID). The Relationship to Birth Certificate appendix is renamed and rewritten as Relationship to Agent Genesis.
4. SEP status codes updated to v07 numbering. Scope Enforcement at SEPs now returns 455 Scope Violation (previously 451) and adds 457 Zone Violation for certificates carrying the governance-zone extension when the request's AGTP-Zone-ID header disagrees with the certificate. The status code renumbering propagates the v06 to v07 change in .
5. SEP enforcement made additive. A new paragraph in makes explicit that a certificate without AGTP-specific extensions is a valid TLS client certificate and is enforced through application-layer checks against the Agent Identity Document. Deployments may mix certificates with and without AGTP extensions.
6. activation-certificate-id semantics clarified. The field is now defined as a cross-layer reference to the Agent Genesis lifecycle event that activated this certificate, rather than to a certificate_hash field that no longer exists under the locked taxonomy. The relying-party contract is unchanged: a 64-hex value suitable for cross-layer audit reconstruction.
7. agtp-ctl-sct cryptographic verification deferred. The extension may be carried and parsed for record-keeping purposes, but verification against an operating AGTP-CTL is deferred to a future revision. The Issuance Protocol is updated to make AGTP-CTL submission and SCT embedding conditional on AGTP-CTL availability.
8. Provisional OID strategy introduced. OIDs for the eight extensions are derived deterministically as UUIDv5 values under a fixed AGTP namespace and the extension's canonical short name, placed under the ITU-T UUID arc 2.25.{uuid_int}. This permits independent implementations to interoperate without a central allocation step. The provisional OIDs will be replaced by IANA-allocated standards-tree OIDs in a future revision.
9. Normative reference to updated to v08. Section references that pointed at v02 / v06 section numbers (Section 6.2, Section 6.7, Section 6.7.3, Section 8.7) are removed; cross-references now name the companion document and the concept rather than a specific section number, since section structure in continues to evolve. RFC 4122 added to normative references to support the OID derivation.

Wire Format Compatibility The v03 changes are entirely framing additions and cross-reference updates. There are no wire-format-visible changes in this revision. Implementations conforming to v02 continue to interoperate with v03. The v02 identifier rename (subject-agent-id to agent-id and principal-id to owner-id) was the wire-format-visible change in the v02 revision. Implementations parsing the subject-agent-id extension labels do not interoperate with implementations using the renamed agent-id and owner-id extension labels under the same OIDs. The OIDs themselves are unchanged across v01, v02, and v03; only the canonical short names changed in v02. All other changes are editorial or specify operational behavior that v00 left unspecified.