# ============================================================================= # STANDARDS ANCHOR REGISTRY - trademark-like unique proper names # ============================================================================= # Implements the Pathways `external_name` registry kind (PATHWAYS_REFERENCE # Level-F+ grammar) as concrete data. Each anchor is BOTH: # (1) a unique, independently discernible proper name to track across every # indexed context and Pathway dimension (a "trademark-like" name), and # (2) an authenticated entry point for engagement with that standard/community. # # Slug grammar: tm: (tm: = tracked proper name / mark) # Every mention of an anchored name anywhere in this bundle should cite its slug # so mentions, and the individual/entity associations with them, are trackable # and deep-linkable (see hypergraph-domain/techniques/ANCHOR_NAME_LEDGER.v1.md). # # `authenticated_entry_point` DIDs are placeholders for engagement endpoints not # yet provisioned; `steward` and `engagement_url` are researched public refs. # `stops_short` paraphrases the Gap Analysis finding for that effort. # `poc_domains` = which of the six SDGs the anchor primarily affiliates with. # ============================================================================= registry_version: "1.0" registry_kind: external_name category: Onc # lineage marker; this registry is domain-agnostic mark_authority: "djat-poc-hypergraph-20260716 (open invitation registry)" # The construct everything else anchors to. core_anchor: slug: tm:poc-cv-standard proper_name: "Proof-of-Control Cryptographic Verification Standard" short_form: "the Standard" steward: "AAI Society (Proof-of-Control Lab)" stewards_named: ["Ken Huang (co-chair)", "Tricia Wang (co-chair)"] engagement_url: "https://djat-poc-hub.pages.dev/" authenticated_entry_point: "did:web:poc.aai-society.org" section_root: "std:poc-cv-0.1.0-rev" poc_domains: [security, privacy, portability, authorization, identity, provenance] note: > The construct all six Shared Domain Graphs connect to. Defines Stage 3 (cryptographically verifiable) of the Proof-of-Control spectrum. anchors: # ── Governance / risk-management frameworks ──────────────────────────── - slug: tm:nist-ai-rmf proper_name: "NIST AI Risk Management Framework (AI RMF 1.0 / NIST AI 100-1)" aliases: ["NIST AI RMF", "AI RMF"] steward: "National Institute of Standards and Technology (NIST)" engagement_url: "https://www.nist.gov/itl/ai-risk-management-framework" authenticated_entry_point: "did:web:airc.nist.gov" kind: governance-framework answers: "How should organizations govern AI risk?" stops_short: "Guidance for people and process; produces no evidence at execution." poc_domains: [security, authorization, provenance] gap_row: nist_ai_rmf - slug: tm:iso-iec-42001 proper_name: "ISO/IEC 42001:2023 - AI Management System (AIMS)" aliases: ["ISO 42001", "ISO/IEC 42001"] steward: "ISO/IEC JTC 1/SC 42" engagement_url: "https://www.iso.org/standard/81230.html" authenticated_entry_point: "did:web:iso.org" kind: management-system-standard answers: "Is there a management system for AI?" stops_short: "Attests process maturity, not system behavior." poc_domains: [authorization, provenance, security] gap_row: iso_iec_42001 - slug: tm:csa-aicm proper_name: "CSA AI Controls Matrix (AICM v1.1)" aliases: ["CSA AI Controls Matrix", "AICM"] steward: "Cloud Security Alliance (CSA)" engagement_url: "https://cloudsecurityalliance.org/artifacts/ai-controls-matrix-v1-1" authenticated_entry_point: "did:web:cloudsecurityalliance.org" kind: control-catalog answers: "Which controls should an AI deployment have?" stops_short: > Control objectives; evidence they operated remains operator-produced. (Working-group co-chair authored it - said with precision.) detail: "247 control objectives across 18 domains; maps to ISO 42001, ISO 27001, BSI AIC4, EU AI Act." poc_domains: [security, authorization, privacy] gap_row: csa_aicm - slug: tm:ieee-7000 proper_name: "IEEE 7000-series Standards (ethical/values-based system design)" aliases: ["IEEE 7000", "IEEE 7000-series"] steward: "IEEE Standards Association" engagement_url: "https://standards.ieee.org/initiatives/autonomous-intelligence-systems/" authenticated_entry_point: "did:web:standards.ieee.org" kind: design-process-standard answers: "Was the system designed responsibly?" stops_short: "Design-time process standards, not runtime evidence." poc_domains: [provenance, identity] gap_row: ieee_7000 - slug: tm:soc2 proper_name: "SOC 2 (Trust Services Criteria)" aliases: ["SOC 2", "SOC2"] steward: "AICPA & CIMA" engagement_url: "https://www.aicpa-cima.com/topic/audit-assurance/audit-and-assurance-greater-than-soc-2/" authenticated_entry_point: "did:web:aicpa-cima.com" kind: attestation-report answers: "Did the organization implement its controls?" stops_short: > Auditor attestation, sampled at the organization level; an agent can execute an unauthorized action while its vendor remains fully compliant. detail: > 2017 Trust Services Criteria (revised points of focus 2022): Security (mandatory), Availability, Processing Integrity, Confidentiality, Privacy. AICPA update: exposure draft expected Fall 2026, final 2027. poc_domains: [security, privacy, authorization] gap_row: soc2 - slug: tm:eu-ai-act proper_name: "EU AI Act (Regulation (EU) 2024/1689)" aliases: ["EU AI Act"] steward: "European Commission / EU AI Office" engagement_url: "https://artificialintelligenceact.eu/" authenticated_entry_point: "did:web:digital-strategy.ec.europa.eu" kind: regulation answers: "Does the system meet agreed criteria?" stops_short: > Accredited assessors with privileged access certify conformance; no evidence anyone else can check of what the system actually did. detail: > Full high-risk enforcement Aug 2, 2026. Art. 12 tamper-evident logging (>=6mo), Art. 14 human oversight + interruption, Art. 15 cybersecurity resilience; Recitals 99-100 extend the compliance boundary to every agent in a chain. poc_domains: [authorization, provenance, security, privacy] gap_row: eu_ai_act # ── Zero-trust and runtime foundations ───────────────────────────────── - slug: tm:nist-sp-800-207 proper_name: "NIST SP 800-207 - Zero Trust Architecture" aliases: ["Zero-trust architecture", "NIST SP 800-207", "ZTA"] steward: "National Institute of Standards and Technology (NIST)" engagement_url: "https://csrc.nist.gov/pubs/sp/800/207/final" authenticated_entry_point: "did:web:csrc.nist.gov" kind: architecture-paradigm answers: "Should this request be trusted?" stops_short: "Governs access at the moment of each request; silent on what the agent did once access was granted." poc_domains: [authorization, identity, security] gap_row: zero_trust - slug: tm:anthropic-zt-agents proper_name: "Anthropic - Zero Trust for AI Agents (May 2026)" aliases: ["Anthropic Zero Trust for AI Agents"] steward: "Anthropic" engagement_url: "https://www.anthropic.com/" authenticated_entry_point: "did:web:anthropic.com" kind: vendor-guidance answers: "Are the agent's controls set correctly?" stops_short: > Preventive at provisioning; explicitly guidance rather than assurance, producing no evidence an outside party can check that controls held. poc_domains: [authorization, identity, security] gap_row: anthropic_zt - slug: tm:confidential-computing-tee proper_name: "Confidential Computing / Trusted Execution Environments (TEEs)" aliases: ["Confidential computing", "TEEs", "TEE"] steward: "Confidential Computing Consortium (Linux Foundation)" engagement_url: "https://confidentialcomputing.io/" authenticated_entry_point: "did:web:confidentialcomputing.io" kind: hardware-mechanism answers: "Was the execution environment protected?" stops_short: > Attests the enclave was untampered, not who authorized the agent or what it did; one mechanism the PoC evidence layer can build on. poc_domains: [security, provenance, identity] gap_row: tee # ── Vendor toolkits and protocols ────────────────────────────────────── - slug: tm:ms-agent-governance-toolkit proper_name: "Microsoft Agent Governance Toolkit" steward: "Microsoft" engagement_url: "https://www.microsoft.com/en-us/security/business/ai-machine-learning/" authenticated_entry_point: "did:web:microsoft.com" kind: vendor-toolkit answers: "How does one vendor's agent stay governed?" stops_short: "Single-vendor and proprietary; no shared evidence definition, no evidence crossing vendor boundaries with the agent." poc_domains: [authorization, identity] gap_row: vendor_toolkits - slug: tm:mastercard-verifiable-intent proper_name: "Mastercard Verifiable Intent / Agent Pay" steward: "Mastercard" engagement_url: "https://www.mastercard.com/news/" authenticated_entry_point: "did:web:mastercard.com" kind: vendor-protocol answers: "How does one vendor's agent stay governed?" stops_short: "Single-vendor and proprietary; no shared, cross-boundary evidence definition." poc_domains: [authorization, identity] gap_row: vendor_toolkits - slug: tm:ping-kya proper_name: "Ping Identity - Know Your Agent (KYA)" aliases: ["Ping Identity", "KYA"] steward: "Ping Identity" engagement_url: "https://www.pingidentity.com/" authenticated_entry_point: "did:web:pingidentity.com" kind: vendor-protocol answers: "How does one vendor's agent stay governed?" stops_short: "Single-vendor and proprietary; no shared, cross-boundary evidence definition." poc_domains: [identity, authorization] gap_row: vendor_toolkits - slug: tm:agent-observability proper_name: "Agent Observability Tooling (operator logs)" steward: "(various vendors)" engagement_url: "https://genai.owasp.org/" authenticated_entry_point: null kind: observability answers: "What happened, per the operator's own logs?" stops_short: "Proprietary and operator-controlled; the record can be altered by the party it describes." poc_domains: [provenance, security] gap_row: observability # ── Orchestration & provenance substrate ─────────────────────────────── - slug: tm:pathways proper_name: "Pathways - Reference Implementation Specification (RIS v1.2.0)" aliases: ["Pathways", "Pathways RIS", "Pathways v1.2.0", "Pathways Reference Implementation Specification"] steward: "the open Pathways project (Originator)" stewards_named: ["DJ Thomson (technique-originator)"] engagement_url: "https://djat-poc-hub.pages.dev/" authenticated_entry_point: "did:placeholder:dj-thomson-pathways" kind: reference-implementation-spec answers: "How is agentic orchestration encoded, provenanced, and exchanged as verifiable pathways?" stops_short: > Orchestration and provenance encoding via PathwayRuns and Aqua trees; PoC Stage-3 conformance is declared against the Standard, not implied by Pathways alone. detail: > Normative RIS v1.2.0 (*Governance and Boundary Classes*): pathway templates, PathwayRuns, gate profiles, boundary classes (Z1-Z4), governance descriptors, trusted peering, and collaborative pathway exchange. Reference implementation for the adopted PoC Provenance domain; deeply integrated with FINOS CALM in this bundle. Lineage: djat_x_poc canon/PATHWAYS_REFERENCE_v1.2.0.md. poc_domains: [provenance, portability, authorization, identity, security, privacy] role_in_bundle: shared-orchestration-and-encoding-language # ── Security verification standards / certifications ──────────────────── - slug: tm:aiuc-1 proper_name: "AIUC-1 (AI Underwriting Certification / standard)" aliases: ["AIUC-1"] steward: "AIUC" engagement_url: "https://www.aiuc-1.com/" authenticated_entry_point: "did:web:aiuc-1.com" kind: certification answers: "Can an AI product be certified for underwriting/insurance?" stops_short: "Certification and crosswalks; not itself a cryptographic runtime-evidence layer." detail: "Crosswalks bidirectionally to OWASP Agentic Top 10; oriented to AI insurance/underwriting." poc_domains: [security, authorization, provenance] gap_row: aiuc_1 insurance_relevant: true - slug: tm:owasp-agentic-top10 proper_name: "OWASP Top 10 for Agentic Applications (2026)" aliases: ["OWASP Top 10 for Agentic AI", "OWASP Agentic AI Top 10"] steward: "OWASP GenAI Security Project" engagement_url: "https://genai.owasp.org/resource/owasp-top-10-for-agentic-applications-for-2026/" authenticated_entry_point: "did:web:owasp.org" kind: threat-catalog answers: "What are the most critical agentic AI security risks?" stops_short: "Names the threats (ASI01-ASI10); does not itself produce evidence a control held at runtime." threats: ["ASI01 Agent Goal Hijack", "ASI02 Tool Misuse", "ASI03 Identity & Privilege Abuse", "ASI04 Agentic Supply Chain", "ASI05 Unexpected Code Execution", "ASI06 Memory & Context Poisoning", "ASI07 Insecure Inter-Agent Comms", "ASI08 Cascading Failures", "ASI09 Human-Agent Trust Exploitation", "ASI10 Rogue Agents"] poc_domains: [security, identity, authorization] gap_row: owasp_agentic - slug: tm:owasp-aivss proper_name: "OWASP AI Security Verification Standard (AISVS v1.0)" aliases: ["AIVSS", "AISVS", "OWASP AIVSS"] steward: "OWASP Foundation" stewards_named: ["Jim Manico (contributor)"] engagement_url: "https://owasp.org/www-project-artificial-intelligence-security-verification-standard-aisvs-docs/" authenticated_entry_point: "did:web:owasp.org" kind: verification-standard answers: "Which testable security requirements has an AI system met?" stops_short: "A catalogue of testable requirements; verification is performed and attested by the tester, not openly checkable by any third party." detail: "v1.0 released 2026-06-24; 3 levels; categories incl. Access Control & Identity, Autonomous Orchestration, MCP Security, Monitoring/Logging." poc_domains: [security, identity, authorization] gap_row: owasp_aivss - slug: tm:csa-vanta-agent-trust-controls proper_name: "CSA / Vanta - Agent Trust Controls" aliases: ["CSA Vanta Agent Trust Controls", "Agent Trust Controls"] steward: "Cloud Security Alliance + Vanta" engagement_url: "https://www.trustcontrols.ai/" authenticated_entry_point: "did:web:trustcontrols.ai" kind: control-mapping-initiative answers: "Which agent trust controls apply, and how do they map?" stops_short: "Open initiative to map to new controls; mapping, not runtime evidence." poc_domains: [security, authorization, provenance] gap_row: csa_vanta - slug: tm:csa-atf proper_name: "CSA Agentic Trust Framework (ATF)" aliases: ["Agentic Trust Framework", "ATF"] steward: "Cloud Security Alliance" engagement_url: "https://github.com/massivescale-ai/agentic-trust-framework" authenticated_entry_point: "did:web:cloudsecurityalliance.org" kind: zero-trust-governance answers: "How is a zero-trust governance model applied to autonomous agents?" stops_short: "Operationalizes OWASP Agentic threats within zero trust; governance architecture, not open runtime evidence." detail: "Published 2026-02-02; five elements incl. incident response; maps to AICM, OWASP Agentic, NIST 800-207, ISO 42001." poc_domains: [security, authorization, identity] gap_row: csa_atf # ── Discovery / interoperability ─────────────────────────────────────── - slug: tm:agent-resource-discovery proper_name: "Agentic Resource Discovery (ARD) Specification" aliases: ["Agent Resource Discovery Spec", "ARD"] steward: "Google and industry partners" engagement_url: "https://developers.googleblog.com/announcing-the-agentic-resource-discovery-specification/" authenticated_entry_point: "did:web:google.com" kind: discovery-spec answers: "How do agents discover and verify tools/services across boundaries?" stops_short: "Discovery + domain-based verification of resources; not evidence of what an agent did after connecting." detail: "ai-catalog.json + registries; domain-based ownership/verification; compatible with MCP and OpenAPI." poc_domains: [portability, identity] gap_row: agent_resource_discovery # ── Healthcare pilot alignment ───────────────────────────────────────── - slug: tm:haarf proper_name: "HAARF - Healthcare AI Agents Regulatory Framework (1.0)" aliases: ["HAARF"] steward: "Task Force for AI Agents in Healthcare" engagement_url: "https://github.com/Task-force-for-AI-agents-in-Healthcare/haarf" authenticated_entry_point: "did:web:github.com:Task-force-for-AI-agents-in-Healthcare" kind: sector-regulatory-framework answers: "How are autonomous medical AI agents verified across their lifecycle?" stops_short: "Synthesizes 279 requirements across 8 categories; the PoC health pilot maps evidence directly to HAARF rather than competing with it." detail: "8 categories incl. C2 Model Passport & Traceability, C3 Cybersecurity, C5 Agent Registration & Identity, C6 Autonomy Governance, C8 Tool-Use Security." poc_domains: [provenance, identity, authorization, security] gap_row: haarf mappings_ref: "https://github.com/Task-force-for-AI-agents-in-Healthcare/haarf/tree/main/HAARF/1.0/mappings" # ── The collaboration bundle's own invitation targets (FINOS) ────────── - slug: tm:finos proper_name: "FINOS (Fintech Open Source Foundation)" steward: "FINOS (Linux Foundation)" stewards_named: ["Karl Moll (Technical Project Advocate)"] engagement_url: "https://finos.org" authenticated_entry_point: "did:web:finos.org" kind: open-source-foundation answers: "Who convenes cross-industry open collaboration for financial-services AI?" poc_domains: [security, authorization, provenance] invitation_target: true - slug: tm:finos-calm proper_name: "FINOS CALM - Common Architecture Language Model" aliases: ["CALM", "Common Architecture Language Model"] steward: "FINOS Architecture as Code (AasC) community" stewards_named: ["Karl Moll (Technical Project Advocate)"] engagement_url: "https://calm.finos.org/" authenticated_entry_point: "did:web:calm.finos.org" kind: modeling-language answers: "How is system architecture defined as machine-readable code?" detail: > JSON meta-schema. Nodes (unique-id, node-type, name, description), Relationships (interacts | connects | deployed-in | composed-of), Controls (pattern-keyed, requirement-url + config), Flows (sequenced transitions over relationships). Adopted here as the shared modeling language, integrated with Pathway encodings. poc_domains: [security, authorization, provenance, portability, identity, privacy] invitation_target: true role_in_bundle: shared-modeling-language - slug: tm:finos-aigf proper_name: "FINOS AI Governance Framework (AIGF)" aliases: ["AIGF", "AI Governance Framework"] steward: "FINOS" engagement_url: "https://air-governance-framework.finos.org/" authenticated_entry_point: "did:web:finos.org" kind: governance-framework answers: "What are the risks and mitigations for AI in financial services?" detail: "Catalog of AI risks + mitigations; converging with CALM (reference architectures) and CCC (controls)." poc_domains: [security, authorization, privacy, provenance] invitation_target: true - slug: tm:finos-ccc proper_name: "FINOS Common Cloud Controls (CCC)" aliases: ["CCC", "Common Cloud Controls"] steward: "FINOS" engagement_url: "https://www.finos.org/common-cloud-controls-project" authenticated_entry_point: "did:web:finos.org" kind: control-catalog answers: "What are the technology-neutral controls for compliant cloud deployments?" detail: "Controls catalog in the AIGF x CCC x CALM convergence Karl Moll convenes." poc_domains: [security, authorization] invitation_target: true # ── Complementary precedents referenced by the PoC standard ──────────── - slug: tm:c2pa proper_name: "C2PA - Coalition for Content Provenance and Authenticity" aliases: ["C2PA"] steward: "C2PA (Joint Development Foundation / Linux Foundation)" engagement_url: "https://c2pa.org/" authenticated_entry_point: "did:web:c2pa.org" kind: provenance-standard answers: "How is content provenance cryptographically attested?" stops_short: "Content provenance, not agent-action provenance; a precedent the PoC provenance domain builds on." poc_domains: [provenance] gap_row: null - slug: tm:maestro proper_name: "MAESTRO - 7 Layers of Control (agentic threat modeling)" aliases: ["MAESTRO", "MAESTRO 7 Layers of Control"] steward: "Cloud Security Alliance (referenced)" engagement_url: "https://cloudsecurityalliance.org/" authenticated_entry_point: "did:web:cloudsecurityalliance.org" kind: threat-modeling answers: "How is an agentic system threat-modeled across seven layers?" detail: "Proposed for the PoC conformance-tier assessment methodology (threat modeling)." poc_domains: [security, authorization] gap_row: null # --------------------------------------------------------------------------- # Cross-cutting note: the AAIS domain classification (Jim's table in the Gap # Analysis) maps each PoC domain to an architectural mechanism and external # alignment targets. That table is encoded in # cross-domain/data/gap-analysis-matrix.v1.json under "aais_domain_map". # ---------------------------------------------------------------------------