THE COMPANION MARKETPLACE & AGENT ECONOMY

Foundational economic framework for the Human Sovereignty Operating System

Instrument: The Companion Marketplace & Agent Economy

Function: Canonical economic framework for trust-based commerce, agent registration, certification, permission scopes, revenue models, and marketplace governance across the Human Sovereignty Operating System — enabling Companions, KAAI Agents, Families, Organizations, Banks, Telcos, Governments, Developers, and Service Providers to transact under human sovereignty, authorization chains, and accountability architecture

Version: 1.0 (Founding Framework)

Status: Subordinate to the Human Sovereignty Charter and all prior founding instruments; governed by the Companion Charter, Life Operating System, Human Digital Twin Architecture, Life Graph Architecture, Trust Vault Architecture, Device Trust Mesh, Family Trust Network, Organization Graph Enterprise Companion Framework, and KAAI Standard

Core constraint: Human sovereignty over all marketplace transactions and agent grants. No platform, institution, agent, or marketplace operator holds standing root authority over a human's economic estate or agent authorization graph by default.

Preamble

Companion Marketplace ecosystem for Humans, Companions, Agents, Families, Organizations, Banks, Telcos, Governments, Developers, Service Providers. Must support Human Sovereignty, Companion Governance, Agent Accountability, Trust-Based Commerce, Global Scale. App Store of the Agent Era. Economic layer of Companion ecosystem.

The digital age promised convenience and delivered extraction. Humans accumulated subscriptions across hundreds of services, authorizations in opaque permission dialogs, financial relationships in institutional silos, and agent-like automations that act without accountability, portability, or revocable grant chains. Each marketplace optimized for platform revenue — attention, engagement, data extraction — not for human sovereignty. Each agent deployment optimized for velocity — API keys in environment variables, OAuth tokens without scope decay — not for trust.

Commerce without human authorization is theft with better UX. Agents without accountability are liability with better latency. Trust without scoring is folklore. The Human Sovereignty Charter establishes rights — ownership, control, portability, inspectability, deletion, revocation, inheritance. Rights require economic architecture that implements them when agents multiply, transactions cross borders, families govern children, enterprises deploy fleets, banks move money, and governments issue permits.

This document defines the Companion Marketplace & Agent Economy — the foundational economic framework through which Keyra Companions, KAAI-authorized agents, Families, Organizations, Banks, Telecommunications carriers, Governments, Developers, and Service Providers discover, certify, authorize, transact, and settle under explicit constitutional governance.

The Companion Marketplace connects Humans, Companions, Agents, Families, Organizations, Banks, Telcos, Governments, Developers, and Service Providers into a unified trust-based commerce fabric. It is the economic layer of the Companion ecosystem — the App Store of the Agent Era, subordinate to human sovereignty rather than platform attention economics.

The Marketplace is designed to scale from one authorized agent to one hundred billion agents — each operating under human-granted permission scopes, trust scores, certification levels, and audit chains — without redesigning the constitutional invariants that subordinate all commerce to human authority.

Preamble — Historical Context

The history of digital marketplaces is a history of platform capture. Application stores authenticated developers to platforms, not agents to humans. Cloud marketplaces sold compute to enterprises without agent accountability semantics. Plugin ecosystems extended applications without portable authorization. Early agent marketplaces listed API wrappers without certification, trust scoring, or revocation infrastructure. Each solved distribution. None solved sovereign commerce — transactions where the human remains root grantor, inspector, and revoker.

When artificial intelligence became persistent — Companions that negotiate, agents that purchase, twins that represent financial intent — the application marketplace model collapsed. An agent cannot purchase responsibly without scoped financial grants. A family cannot protect children without child-safe agent certification. A bank cannot settle without authorization chain attestation. A government cannot issue permits through agents without audit and revocation. The Companion Marketplace closes the economic trust gap.

Preamble — Relationship to Founding Instruments

This Framework is subordinate to the Human Sovereignty Charter. Where marketplace technical requirements appear to conflict with human sovereignty, human sovereignty prevails and marketplace implementations must be corrected. The Companion Marketplace integrates with the Trust Vault Architecture (financial credentials, agent certificates, transaction audit), Device Trust Mesh (device-bound authorization, presence-gated purchases), Life Graph Architecture (agent nodes, trust edges, ownership graphs), KAAI Standard (agent identity, authorization certificates, accountability), Companion Charter (Companion-mediated commerce), Family Trust Network (family budgets, child-safe agents), Organization Graph Enterprise Companion Framework (enterprise agent fleets), and Life Operating System (domain-scoped spending policies).

No single instrument owns the marketplace. The human owns the authorization graph. The Life Graph indexes agent relationships. The Companion mediates human intent to agent operations. KAAI agents receive derived access through authorization chains rooted in Trust Vault-held keys and human presence proofs. Together they form the Human Sovereignty Operating System for durable trust-based commerce.

Preamble — Normative Language

Throughout this document:

• MUST, MUST NOT, REQUIRED, and SHALL denote absolute requirements for Companion Marketplace conformance
• SHOULD and RECOMMENDED denote strong guidance with documented exceptions permitted only under human or institutional policy with audit
• MAY denotes optional capability
• Prohibited actions are void regardless of technical success; marketplace runtimes MUST reject them

Conformance is measured at three layers: constitutional (subordination to human authority), technical (registration, certification, authorization, settlement), and operational (audit, revocation, dispute resolution, incident response).

Preamble — Architectural Placement

The Companion Marketplace sits beneath human sovereignty and above application silos — the economic substrate on which vault, graph, companion, device, and agent layers transact:

┌──────────────────────────────────────────────┐
│     Human Sovereignty Charter               │
├──────────────────────────────────────────────┤
│     Companion · Twin · Life Graph · Vault  │
├──────────────────────────────────────────────┤
│     KAAI Standard · Device Trust Mesh       │
├──────────────────────────────────────────────┤
│     Companion Marketplace (this document)  │
│     Registration · Certification · Commerce │
├──────────────────────────────────────────────┤
│     Agents · Services · Settlement · Trust │
├──────────────────────────────────────────────┤
│     Developers · Banks · Telcos · Gov       │
└──────────────────────────────────────────────┘

Applications are replaceable. Agent authorization graphs and transaction audit chains are not — they belong to the human and persist across platform replacements when export ceremonies execute.

Preamble — Scope of Support

The Companion Marketplace supports:

The Companion Marketplace serves:

• Individuals — personal agent authorization under sole human authority
• Families — federated agent catalogs with governed child access
• Organizations — institutional agent deployments subordinate to human members
• Governments — lawful certification and citizen service channels without root authority usurpation
• Future Generations — inherited agent grants, succession of authorization, legacy commerce policies

PART I — Definition

Section 1.01 — What Is a Companion Marketplace?

The Companion Marketplace is a human-sovereign, trust-based, cryptographically attested economic architecture — comprising agent registration, certification, permission scopes, trust transactions, revenue models, and governance frameworks — through which KAAI agents, Companion extensions, and certified services are discovered, authorized, purchased, and settled under explicit constitutional subordination to the Sovereign Human.

The Companion Marketplace:

• Registers agents — identity, ownership, certification, verification, sponsorship, publishing, retirement
• Certifies trust levels — Experimental through Critical Infrastructure Certified
• Scopes permissions — data, financial, communication, family, enterprise, government rights
• Executes trust transactions — scoring, guarantees, escrow, reputation, recovery
• Models economics — subscription, usage, transaction, revenue sharing, licensing, fees
• Governs publication — audit, approval, compliance, revocation, retirement
• Federates globally — cross-border discovery, certification, authorization, commerce
• Protects security — agent isolation, vault binding, identity and financial protection
• Scales civilization-wide — from one agent to one hundred billion agents

The Companion Marketplace is the economic layer of the Human Sovereignty Operating System.

Section 1.01a — Companion Marketplace as Constitutional Implementation

The Human Sovereignty Charter declares rights. The Trust Vault Architecture declares persistence. The Device Trust Mesh declares physical anchoring. The Companion Marketplace declares commerce — the material condition without which agents act in institutional gray zones and humans lose inspectability over economic life. A Sovereign Human whose agents operate only through platform OAuth without exportable grant chains does not possess economic sovereignty in any operational sense. A Sovereign Human whose purchases occur only inside walled gardens does not possess portability in any commercial sense.

The Companion Marketplace therefore occupies a position analogous to commercial law and payment systems in physical civilization — not extraction infrastructure, but trust infrastructure that binds economic action to human intent. Without marketplace architecture, agent commerce becomes platform roulette.

Section 1.01b — Actors Served by the Companion Marketplace

Each actor's relationship to the marketplace is defined, bounded, auditable, and revocable — except the human root, whose sovereignty is inalienable.

Section 1.02 — What Is Not a Companion Marketplace?

The Companion Marketplace is not:

• An application store alone — binary distribution without agent accountability, trust scoring, or portable authorization
• A cloud marketplace alone — compute and API listing without human-rooted grant chains
• A plugin ecosystem alone — application extensions without KAAI certification and revocation semantics
• An agent marketplace alone — API wrapper listings without Companion integrations, vault binding, and family governance
• Attention economics — engagement optimization, dark patterns, or surveillance monetization
• Platform financial capture — mandatory payment rails that prevent human inspection and export
• Compulsory agent registry — centralized inventory without individual revocation and portability
• Unaccountable automation — agents that act without certification, audit, and human-granted scopes

If marketplace transactions cannot be inspected and revoked by the human root, they violate the Human Sovereignty Charter. If agents purchase without authorization chain validation, they violate the KAAI Standard. If child-facing agents bypass family approval structures, they violate the Family Trust Network.

Section 1.03 — Distinctions Among Marketplace Systems

Application Store (App Store)

Application stores — Apple App Store, Google Play, Microsoft Store — distribute binaries to devices. They authenticate developers to platforms. They review for malware and policy compliance. They optimize for platform revenue share — typically 15–30%. They do not model agent permission scopes, trust decay, family child-safe certification, cross-vendor authorization portability, or human-rooted financial grants.

The Companion Marketplace incorporates application distribution where agents require local execution — but distribution is one function among registration, certification, authorization, settlement, and audit. An app without KAAI agent certificate is not a marketplace agent — it is a legacy application.

Cloud Marketplace

Cloud marketplaces — AWS Marketplace, Azure Marketplace, GCP Marketplace — sell SaaS, AMIs, and APIs to enterprise tenants. They excel at billing aggregation and procurement compliance. They treat the enterprise tenant as customer. Individual humans within tenants are invisible. Agent accountability, family budgets, device-bound authorization, and Trust Vault credential binding are absent.

The Companion Marketplace supports enterprise procurement overlays — but enterprise catalogs remain subordinate to member human sovereignty for personal partitions and portable agent grants.

Plugin Ecosystem

Plugin ecosystems — browser extensions, IDE plugins, productivity add-ons — extend host applications with narrow APIs. Authorization is host-mediated OAuth. Revocation is host-controlled. Portability dies when the host application changes. No standardized trust scoring, certification levels, or financial scope semantics exist.

The Companion Marketplace may list Companion extensions and Twin projections — but extensions MUST register as KAAI agents or Companion modules with explicit scopes, not opaque host plugins.

Agent Marketplace (industries Generic)

industries agent marketplaces — early GPT stores, workflow automation directories, MCP server listings — optimize discovery of capabilities. They rarely require: hardware presence for high-risk actions, Trust Vault credential storage, Device Trust Mesh binding, certification tiers, trust escrow, complaint frameworks, government accreditation, or cross-border authorization federation.

The Companion Marketplace defined here is human-rooted — platform discovery is a feature, not the foundation.

Companion Marketplace (This Framework)

The Companion Marketplace integrates discovery with governance — every listed agent carries certification level, permission scope manifest, trust score history, publisher identity, audit chain, and human-grant requirement. Commerce flows through authorization, not attention. Settlement attaches to trust transactions, not engagement metrics.

Section 1.03a — Illustrative Scenario

Consider a sovereign professional: she authorizes a KAAI travel agent (Trusted certification) with financial scope capped at $5,000 per transaction and Device Trust Mesh requirement above 0.7 for itinerary changes affecting visas. Her Companion mediates negotiation — the agent proposes flights; Companion presents comparison; she grants per-trip authorization with Keyra Key signature.

Her family operates a Family Marketplace partition — child-safe education agents (Verified certification) with no financial scope; teenage son has shopping agent (Experimental) with $50 weekly budget and parent approval for purchases above $25. Her employer deploys Enterprise Marketplace — compliance agent (Enterprise Certified) reads Organization Vault contracts; HR agent accesses bounded employee data per role.

When a fraudulent agent attempts purchase, Trust Escrow holds settlement pending trust score verification. When she revokes travel agent grant, revocation propagates within SLA; agent certificates invalidate; pending transactions cancel. When she exports Agent Authorization Pack, grants port to new Companion without platform gatekeeping.

Section 1.04 — Why Agents Require a New Economic Model

Applications requested human attention. Agents request authorization — persistent, scoped, decaying grants to act on human behalf across time, devices, and institutions. The economic model must answer:

Agents that purchase, negotiate, transfer, and represent require trust-based commerce — not attention-based commerce. The Companion Marketplace is the economic architecture for that transition.

Section 1.05 — Agent Economy Definition

The Agent Economy is the aggregate of authorized agent transactions — subscriptions, usage charges, settlements, revenue shares, trust fees — occurring under Companion Marketplace governance. It is not cryptocurrency speculation. It is not platform tokenomics. It is authorized economic activity where every transaction cites human grant, agent certificate, trust score threshold, and audit reference.

The Agent Economy MUST remain subordinate to human sovereignty. GDP of agent transactions is measurable; sovereignty of humans is not negotiable.

Section 1.06 — Marketplace Runtime Architecture

The Marketplace Runtime validates every agent operation through a pipeline — not optional middleware:

Agent Request → Scope Validator → Certification Gate → 
Trust Score Gate → Spending Control → Presence Check → 
Authorization Chain → Settlement → Audit Append

Each stage MUST fail closed. Partial pipeline success without full validation is prohibited. Runtime implementations MUST NOT cache authorization decisions beyond policy-declared TTL without revalidation of trust score and grant status.

Section 1.07 — Relationship to Device Trust Mesh

High-consequence marketplace actions MUST bind to Device Trust Mesh requirements per KAAI Standard and Device Trust Mesh Architecture. Financial authority level A3 and above REQUIRES device trust score above human-declared threshold and presence level P4 minimum unless emergency instrument documented in Authorization Vault. Device revocation MUST halt pending agent transactions tied to that device within SLA.

Section 1.08 — Relationship to Trust Vault

Agent certificates, financial credentials, transaction audit hashes, and spending policies MUST reside in Trust Vault Agent and Authorization partitions — not in marketplace operator databases as root secrets. Marketplace operators MAY hold federated registry metadata and encrypted audit replicas; they MUST NOT hold human root keys or unaudited financial credentials.

Section 1.09 — Comparative Architecture Table

PART II — Foundational Principles

Section 2.01 — Human Ownership

The Sovereign Human owns the agent authorization estate absolutely. Ownership includes all agent grants, marketplace subscriptions, spending policies, certification trust relationships, and transaction history references. Ownership is inalienable — organizations may custodian enterprise agent fleets during employment, but personal agent grants remain human-rooted unless explicit voluntary transfer executes.

Implementations MUST represent agent ownership in Life Graph as Human → authorizes → Agent. No edge may assign Platform → owns → Agent without explicit human-initiated migration where human retains revocation root.

Section 2.02 — Human Authorization

Every marketplace transaction and agent action MUST cite Human Authorization — validated grant chain from human root, scoped to permission manifest, presence level where required, and trust score threshold. Default deny. No grant — no transaction.

Human Authorization integrates KAAI Authorization Certificates, Trust Vault financial partition policies, Device Trust Mesh presence proofs, and Companion-mediated confirmation for consequential purchases.

Section 2.03 — Agent Accountability

Agent Accountability means every agent carries identifiable publisher, certification level, audit chain, and liability attribution. Agents MUST NOT act anonymously at consequential scope. Experimental agents carry explicit warnings. Critical Infrastructure agents carry publisher bonds and government accreditation.

Accountability failures trigger certification downgrade, marketplace suspension, and trust score penalties — not silent continuation.

Section 2.04 — Trust-Based Commerce

Commerce in the Companion Marketplace is trust-based, not attention-based. Transactions require trust score thresholds, certification minimums, and optional trust escrow. Discovery MAY rank by relevance but MUST NOT rank by undisclosed payment for placement without human-visible sponsorship labels.

Trust-Based Commerce integrates Trust Vault reputation refs, Device Trust Mesh device binding, and marketplace trust transaction layer (Part IX).

Section 2.05 — Permission-Based Access

Agents receive Permission-Based Access — never standing root. Scopes declare data, financial, communication, family, enterprise, and government rights explicitly. Scope expansion requires new human grant. Scope decay retires unused permissions automatically per policy.

Permission manifests MUST be human-inspectable, machine-validated, and exportable in Agent Authorization Pack.

Section 2.06 — Transparency

All marketplace fees, revenue shares, sponsorship relationships, and data accesses MUST be transparent to the human root. Hidden fees are prohibited. Undisclosed agent data exfiltration is prohibited. Audit logs MUST be human-inspectable and hash-chained to Trust Vault.

Section 2.07 — Portability

The Sovereign Human MUST export marketplace state in Agent Commerce Export Pack (ACEP) — agent grants, certification refs, subscription records, spending policies, transaction audit hashes — without vendor gatekeeping. ACEP interoperates with Trust Vault Export Pack and Device Trust Export Pack.

Portability MUST survive platform migration, employment termination, and carrier change.

Section 2.08 — Fair Competition

The Companion Marketplace MUST enable Fair Competition — third-party agents compete on certification, trust scores, and capability — not on platform lock-in, undisclosed preferential API access, or anti-portability terms. Marketplace operators MUST NOT use human transaction data to advantage proprietary agents without explicit human consent.

Section 2.09 — Global Interoperability

Agent registration, certification, and authorization MUST interoperate globally through federated Global Agent Registry (Part XVIII). National overlays accredit without usurping human root. Cross-border commerce respects human grant and local lawful requirements without mandatory single-marketplace capture.

Section 2.10 — Principle Enforcement Matrix

Violations MUST surface as errors — not silent degradation.

Section 2.11 — Tension Resolution

Principles occasionally tension:

• Authorization vs convenience — low-risk Experimental agents MAY use streamlined grants; high-risk Financial agents never waive presence
• Enterprise procurement vs personal sovereignty — dual catalogs; personal grants survive employment
• Trust escrow vs velocity — escrow optional per human policy; mandatory for Critical Infrastructure
• Global interoperability vs national law — lawful local restrictions apply; human revocation remains universal

Resolution always favors human root authority after policy-declared emergency expiry.

Section 2.12 — Illustrative Scenario — Principle Collision

A enterprise compliance agent (Enterprise Certified) requests access to personal Health Vault partition for wellness program. Enterprise policy mandates participation. Human sovereignty prevails: enterprise agent receives Organization Vault occupational health only; personal Health Vault requires separate human grant. Marketplace runtime rejects cross-partition access attempt; audit logs incident; enterprise administrator receives policy correction notice — not human override.

PART III — Marketplace Architecture

Section 3.01 — Marketplace Architecture Overview

The Companion Marketplace comprises core infrastructure and domain-specific stores — each store curates catalogs, certification requirements, and governance overlays while sharing registration, trust transaction, and settlement layers.

┌─────────────────────────────────────────────┐
│           Marketplace Core                  │
│  Registry · Certification · Settlement     │
├──────────┬──────────┬──────────┬────────────┤
│Companion │  Agent   │  Family  │ Enterprise │
│  Store   │  Store   │  Store   │   Store    │
├──────────┼──────────┼──────────┼────────────┤
│Government│ Education│Healthcare│  Banking   │
│  Store   │  Store   │  Store   │   Store    │
├──────────┴──────────┴──────────┴────────────┤
│        Telecommunications Store              │
└─────────────────────────────────────────────┘

Section 3.02 — Companion Store

The Companion Store distributes Companion configurations, personality modules, skill packs, Twin projection templates, and Companion-native extensions. All listings require Companion Charter conformance review. Companion Store transactions use Companion Fees model (Section 8.10). Companion modules MUST NOT hold root vault keys.

Section 3.03 — Agent Store

The Agent Store is the primary catalog for KAAI agents — travel, banking, shopping, legal, and domain agents per Part IV taxonomy. Agent Store listings require KAAI registration (Part V), minimum certification level declaration, and permission scope manifest publication. Default discovery ranks by trust score and certification — not undisclosed payment.

Section 3.04 — Family Store

The Family Store curates family-appropriate agents — child-safe, education, family services — with mandatory Family Trust Network conformance. Family Store agents MUST declare age appropriateness, parental approval requirements, and financial scope restrictions. Family guardians administer catalog visibility per Family Constitution.

Section 3.05 — Enterprise Store

The Enterprise Store serves Organization Graph deployments — department agents, compliance agents, finance agents, HR agents. Enterprise Store requires Organization Graph Enterprise Companion Framework conformance. Enterprise agents operate on Organization Vault partitions; personal agent grants remain separate.

Section 3.06 — Government Store

The Government Store publishes citizen service agents — identity, permits, tax, healthcare, education, national services — with Government Certified or Critical Infrastructure certification. Government Store agents MUST NOT usurp human root; they execute lawful services under citizen grant.

Section 3.07 — Education Store

The Education Store distributes curriculum agents, tutoring agents, learning companions, and institutional education integrations. Education Store agents require Verified minimum certification; child-facing agents require child-safe badge and COPPA-equivalent conformance per jurisdiction.

Section 3.08 — Healthcare Store

The Healthcare Store lists clinical support agents, patient navigation agents, and health record integrations agents. Healthcare Store agents require Trusted minimum certification; PHI access requires Health Vault partition grant and HIPAA-equivalent audit per jurisdiction.

Section 3.09 — Banking Store

The Banking Store hosts financial agents — payments, transfers, investments, insurance, credit — with Enterprise Certified or Government Certified minimum for consequential financial scope. Banking Store integrates bank settlement rails and authorization chain verification (Part XIII).

Section 3.10 — Telecommunications Store

The Telecommunications Store distributes subscriber agents, eSIM provisioning agents, identity agents, and network trust agents. Telco-published agents require carrier accreditation overlay subordinate to human root (Part XIV).

Section 3.11 — Store Interoperability

Stores share Marketplace Core services — registration, certification validation, trust transactions, settlement — but MAY apply domain-specific certification minimums. An agent MAY list in multiple stores when meeting each store's requirements. Store-specific policies MUST NOT weaken human root authority.

Section 3.12 — Store Governance Table

PART IV — Agent Categories

Section 4.01 — Agent Taxonomy Overview

KAAI agents in the Companion Marketplace organize into domain taxonomies — each taxonomy declares typical permission scopes, certification recommendations, and store placement. Taxonomy guides discovery; it does not limit human authorization creativity.

Section 4.02 — Travel Agents

Travel Agents — itinerary planning, booking, visa assistance, disruption rebooking. Typical scopes: communication (airlines, hotels), financial (bookings capped), data (passport refs from Identity Vault). RECOMMENDED certification: Trusted. Device Trust Mesh binding RECOMMENDED for international itinerary changes.

Section 4.03 — Banking Agents

Banking Agents — balance inquiry, payments, transfers, fraud alerts. Typical scopes: financial (bank APIs), data (account refs). REQUIRED certification: Enterprise Certified minimum for transfers; Trusted for inquiry-only.

Section 4.04 — Investment Agents

Investment Agents — portfolio monitoring, rebalancing proposals, trade execution. Typical scopes: financial (brokerage APIs), data (Asset Vault). REQUIRED certification: Enterprise Certified. Human confirmation REQUIRED for trades above policy threshold.

Section 4.05 — Healthcare Agents

Healthcare Agents — appointment scheduling, record retrieval, medication reminders. Typical scopes: health (Health Vault), communication (providers). REQUIRED certification: Trusted; Government Certified for national health system integrations.

Section 4.06 — Education Agents

Education Agents — tutoring, curriculum delivery, progress tracking. Typical scopes: data (education records), communication (institutions). REQUIRED certification: Verified; child-safe badge for K-12.

Section 4.07 — Learning Agents

Learning Agents — skill acquisition, language learning, professional development — distinct from institutional Education Agents. Typical scopes: data (progress), memory (learning history). RECOMMENDED certification: Verified.

Section 4.08 — Shopping Agents

Shopping Agents — product discovery, price comparison, purchase execution. Typical scopes: financial (capped), communication (retailers), data (preferences). RECOMMENDED certification: Verified for autonomous purchase; Experimental for recommendation-only.

Section 4.09 — Insurance Agents

Insurance Agents — quote comparison, policy management, claims initiation. Typical scopes: financial, data (Asset Vault, Health Vault). REQUIRED certification: Enterprise Certified.

Section 4.10 — Legal Agents

Legal Agents — document preparation, deadline tracking, legal research — not unauthorized practice of law. Typical scopes: data (documents), communication (counsel). REQUIRED certification: Trusted; publisher MUST declare jurisdictional limitations.

Section 4.11 — Government Agents

Government Agents — permit applications, tax filing, benefit enrollment, citizen services. Typical scopes: government (lawful APIs), identity (Identity Vault). REQUIRED certification: Government Certified.

Section 4.12 — Enterprise Agents

Enterprise Agents — internal workflow, compliance, document processing, departmental automation. Typical scopes: enterprise (Organization Vault). REQUIRED certification: Enterprise Certified.

Section 4.13 — Family Agents

Family Agents — household coordination, shared calendars, chore management, elder check-in. Typical scopes: family (Family Vault), communication (family members). RECOMMENDED certification: Verified; child interaction requires child-safe badge.

Section 4.14 — Companion Extensions

Companion Extensions — skills and modules extending Keyra Companion capability without independent KAAI identity. Companion Extensions register in Companion Store; they inherit Companion's grant chain and MUST NOT escalate scope without explicit human approval.

Section 4.15 — Taxonomy Cross-Reference Matrix

PART V — Agent Registration

Section 5.01 — Registration Overview

Every marketplace agent MUST complete Agent Registration before publication — establishing identity, ownership, certification pathway, and audit root. Registration occurs in the Agent Registry — federated directory integrated with Global Agent Registry (Part XVIII).

Section 5.02 — Agent Identity

Agent Identity comprises: agent_id (globally unique), agent_name, agent_version, taxonomy_class, publisher_id, public_key, and capability_manifest. Identity MUST be cryptographically signed by publisher and anchored in Agent Registry. Identity MUST NOT impersonate human identity — agents are instruments, not persons.

Section 5.03 — Agent Ownership

Agent Ownership identifies the Publisher — developer, organization, government agency, or service provider legally responsible for agent behavior. Publisher identity MUST be verifiable. Shell publishers are prohibited for Trusted certification and above.

Section 5.04 — Agent Certification

Registration declares target certification level and initiates certification workflow (Part VI). Agents MUST NOT publish above achieved certification. Certification downgrade automatically restricts discovery and scope.

Section 5.05 — Agent Verification

Agent Verification confirms publisher identity, capability claims, and security posture — static analysis, dynamic testing, third-party audit for higher tiers. Verification evidence MUST be stored in Agent Registry with human-inspectable summary.

Section 5.06 — Agent Sponsorship

Agent Sponsorship labels commercial relationships — enterprise endorsements, government recommendations, Companion featured listings. Sponsorship MUST be human-visible. Sponsored agents MUST NOT bypass certification requirements or trust score gates.

Section 5.07 — Agent Publishing

Agent Publishing releases agent to store catalogs after registration, verification, and certification approval. Publishing requires permission scope manifest publication, privacy policy, liability attribution, and revocation contact. Publishing MAY be staged — beta channel, enterprise-only, geographic restriction.

Section 5.08 — Agent Retirement

Agent Retirement gracefully removes agents from discovery while preserving audit history. Active grants MUST receive retirement notice; humans MUST re-authorize successor agents. Retirement MUST NOT delete transaction audit chains.

Section 5.09 — Agent Registry

The Agent Registry maintains authoritative records: identity, certification status, trust scores, publisher, version history, revocation status. Registry MUST federate with Global Agent Registry. Registry queries MUST NOT expose human grant graphs — only public agent metadata.

Section 5.10 — Registration Workflow

Section 5.11 — Illustrative Scenario — Registration

A developer publishes a shopping agent. She registers agent_id, submits capability manifest (price comparison, purchase execution), verifies publisher identity, completes Verified certification audit, publishes scope manifest (financial cap $500 per transaction, no health data), lists in Agent Store. A human authorizes agent with personal cap $200. Agent operates within intersection of manifest cap and human cap — $200 effective limit.

Section 5.12 — Agent Versioning and Updates

Agent versions MUST register as semver increments. Breaking scope changes REQUIRE new human authorization — silent scope expansion via update is prohibited. Humans MUST receive update notification with scope diff summary. Auto-update MAY apply for patch versions within unchanged scope manifest hash. Major version updates MUST require explicit re-grant ceremony.

Section 5.13 — Multi-Publisher and White-Label Agents

White-label agents — enterprise-branded instances of publisher agents — MUST maintain traceability to root publisher in Agent Registry. White-label MUST NOT obscure liability attribution. Enterprise white-label REQUIRES Enterprise Certified minimum on root publisher.

Section 5.14 — Agent Registry Data Model

PART VI — Certification Levels

Section 6.01 — Certification Overview

Certification Levels express trustworthiness tiers — from Experimental prototypes to Critical Infrastructure agents. Certification gates discovery, scope eligibility, and store placement. Certification is earned, audited, revocable — not purchased without audit.

Section 6.02 — Experimental

Experimental certification permits discovery with explicit warnings. Scope limited to recommendation, simulation, and non-consequential actions. Financial scope prohibited unless human grant explicitly accepts Experimental financial risk. RECOMMENDED for beta testing and developer iteration.

Section 6.03 — Verified

Verified certification confirms publisher identity, basic security audit, and capability accuracy. Eligible for general Agent Store, Family Store (with child-safe review), and Education Store. Financial scope permitted with standard human grants. Default certification for mainstream consumer agents.

Section 6.04 — Trusted

Trusted certification requires enhanced security audit, operational SLA, incident response plan, and trust score maintenance above 0.7. Eligible for Healthcare Store, Telecommunications Store, and high-value travel agents. PHI and identity scopes require Trusted minimum.

Section 6.05 — Enterprise Certified

Enterprise Certified certification requires enterprise security audit (SOC 2 equivalent), liability insurance, publisher bond, and compliance documentation. Eligible for Enterprise Store, Banking Store, and investment agents. Financial scopes above $10,000 per transaction require Enterprise Certified minimum.

Section 6.06 — Government Certified

Government Certified certification requires government agency publisher or government-accredited third party, national security review, and lawful intercept compliance where applicable without human root usurpation. Eligible for Government Store and national health integrations.

Section 6.07 — Critical Infrastructure Certified

Critical Infrastructure Certified certification applies to agents affecting power, water, telecommunications backbone, financial system stability, or national security. Requires government accreditation, continuous monitoring, trust escrow mandatory, publisher bond minimum, and 24/7 incident response. Highest tier; revocation triggers national incident protocols.

Section 6.08 — Certification Requirements Matrix

Section 6.09 — Certification Revocation

Certification revocation MUST propagate within SLA. Revoked agents MUST halt new authorizations; existing grants SHOULD receive human notification for re-authorization or migration. Certification revocation triggers marketplace investigation and possible publisher sanctions.

Section 6.10 — Certification Upgrade Path

Publishers MAY upgrade certification with new audit evidence. Downgrade occurs automatically on trust score breach, incident finding, or audit failure. Humans MUST NOT be trapped in upgraded agent grants — downgrade does not expand scope; it restricts discovery and new grants only.

PART VII — Permission Scopes

Section 7.01 — Scope Architecture Overview

Permission Scopes define bounded authority granted to agents. Scopes compose from scope types, authority levels, and trust levels — validated at transaction time against human grant, agent manifest, and certification level.

Section 7.02 — Permission Scope Types

Section 7.03 — Authority Levels

Authority Levels express action consequence:

Section 7.04 — Trust Levels

Trust Levels express minimum trust score for scope activation:

Section 7.05 — Data Rights

Data Rights scope permits access to vault data refs — Identity, Memory, Health, Relationship, Asset. Data Rights MUST cite specific artifact classes, not blanket vault access. Export rights require explicit grant.

Section 7.06 — Financial Rights

Financial Rights scope permits transactions — amounts, currencies, payees, velocity limits, merchant categories. Financial Rights MUST integrate bank authorization chains for settlement. Financial Rights MUST respect human spending controls (Part X).

Section 7.07 — Communication Rights

Communication Rights scope permits outbound communication — email, SMS, voice, in-app — to specified parties or categories. Communication Rights MUST NOT permit impersonation of human without disclosed agent identity.

Section 7.08 — Family Rights

Family Rights scope permits interaction with family members — children, elders, shared resources. Family Rights MUST comply with Family Constitution approval structures. Child-facing communication requires child-safe certification and guardian visibility.

Section 7.09 — Enterprise Rights

Enterprise Rights scope permits Organization Vault access per role. Enterprise Rights terminate on employment end unless separate human grant. Enterprise Rights MUST NOT access personal vault partitions without human grant.

Section 7.10 — Government Rights

Government Rights scope permits lawful government API access — tax filing, permit application, benefit enrollment. Government Rights MUST cite legal basis. Government agents MUST NOT expand scope beyond published manifest.

Section 7.11 — Scope Composition Rules

Scopes MUST compose intersectively — effective authority is minimum of human grant, agent manifest, certification eligibility, and trust score. Scope expansion requires new grant ceremony. Scope reduction applies immediately on human revoke.

Section 7.12 — Scope Manifest Example

agent_id: travel-agent-v3
scopes:
  - type: data
    rights: [identity.passport_ref, memory.travel_preferences]
    authority: A1
    trust_min: T2
  - type: financial
    rights: [purchase.airline, purchase.hotel]
    cap_per_transaction: 5000 USD
    authority: A2
    trust_min: T2
    presence_min: P3
  - type: communication
    rights: [email.airline, email.hotel]
    authority: A2
    trust_min: T1

Section 7.13 — Illustrative Scenario — Scope Intersection

Human grants shopping agent $300 cap. Agent manifest allows $500. Certification allows financial scope. Effective cap: $300. Agent attempts $350 purchase — rejected at authorization engine. Human receives notification with override option requiring Keyra Key for one-time exception.

PART VIII — Agent Economics

Section 8.01 — Economic Model Overview

The Companion Marketplace supports multiple revenue and pricing models — subscription, usage, transaction, revenue sharing, licensing — with transparent fee disclosure. Economic models MUST NOT obscure total cost from human root.

Section 8.02 — Subscription Model

Subscription Model — recurring payment for agent access, Companion modules, or service tiers. Subscriptions MUST be human-authorizable with renewal notice. Auto-renewal requires explicit opt-in per jurisdiction. Subscription cancellation MUST revoke agent grants at period end unless human extends.

Section 8.03 — Usage Model

Usage Model — metered charges per API call, transaction, compute minute, or outcome. Usage meters MUST be human-inspectable. Usage caps MUST be enforceable in authorization engine. Bill shock prevention: default usage alerts at 80% and 100% of human-declared cap.

Section 8.04 — Transaction Model

Transaction Model — fees per completed purchase, transfer, or settlement. Transaction fees MUST appear in pre-purchase disclosure. Transaction model integrates trust escrow (Part IX) — fees may include trust guarantee premium.

Section 8.05 — Revenue Sharing

Revenue Sharing distributes transaction revenue among publisher, marketplace operator, Companion platform, and optionally human referrer. Revenue share percentages MUST be disclosed at authorization. Undisclosed revenue share is prohibited.

Section 8.06 — Agent Licensing

Agent Licensing — perpetual or term license for agent deployment in enterprise. License terms MUST permit portability where human root authorizes — enterprise licenses MUST NOT trap personal grants. License violation triggers certification review.

Section 8.07 — Trust Fees

Trust Fees — premium for trust escrow, enhanced certification monitoring, or dispute resolution fund contribution. Trust fees MUST be optional except Critical Infrastructure mandatory escrow. Trust fee application MUST be receipted.

Section 8.08 — Authorization Fees

Authorization Fees — charges for high-assurance authorization ceremonies — government identity verification, bank KYC integrations, notary-equivalent digital witnessing. Authorization fees MUST be disclosed before ceremony.

Section 8.09 — Marketplace Fees

Marketplace Fees — platform fee for discovery, settlement, and registry services. Marketplace fees SHOULD be competitive and transparent. Marketplace operators MUST NOT condition fee discounts on anti-portability terms.

Section 8.10 — Companion Fees

Companion Fees — charges for Companion Store modules, premium Companion capabilities, Twin projection packs. Companion fees flow through same transparency requirements. Companion MUST NOT condition core sovereignty features on fee payment.

Section 8.11 — Fee Stacking Disclosure

When multiple fees apply, Fee Stacking Disclosure MUST present itemized total before human confirmation:

Section 8.12 — Economic Model Selection Guidance

PART IX — Trust-Based Commerce

Section 9.01 — Trust Transaction Overview

Trust Transactions are marketplace settlements where trust score, certification, and optional escrow govern execution — not merely payment authorization. Trust Transactions bind financial settlement to authorization chain integrity.

Section 9.02 — Trust Scoring

Trust Scoring aggregates agent reliability — completion rate, dispute rate, incident history, certification maintenance, publisher standing. Trust scores range \( T \in [0,1] \). Trust scores MUST decay without positive evidence. Trust score updates MUST be auditable.

Section 9.03 — Trust Guarantees

Trust Guarantees — publisher bonds, insurance, marketplace dispute funds — compensate humans for agent failure within policy. Guarantee claims require audit review. Guarantee availability MUST be disclosed per certification level.

Section 9.04 — Trust Escrow

Trust Escrow holds settlement until delivery confirmation, trust score verification, or timeout release. Escrow RECOMMENDED for high-value transactions; REQUIRED for Critical Infrastructure. Escrow release requires human confirmation or automated policy satisfaction.

Section 9.05 — Trust Reputation

Trust Reputation — long-horizon publisher and agent standing across transactions. Reputation MUST NOT be sold or transferred without audit. Reputation portability follows publisher, not human — human trust relationships port via ACEP grant history refs.

Section 9.06 — Trust Recovery

Trust Recovery — pathway for agents and publishers to restore trust after incidents — remediation plan, re-audit, probation period with reduced scopes. Trust recovery MUST NOT erase incident history; history informs future trust scores with decay.

Section 9.07 — Trust Transaction Lifecycle

Section 9.08 — Illustrative Scenario — Trust Escrow

Human authorizes furniture purchase via shopping agent ($2,400). Trust escrow holds payment until delivery scan confirms receipt. Agent trust score 0.82; threshold 0.75 — pass. Delivery delayed 14 days; human extends escrow once. Delivery confirmed; escrow releases; audit complete.

Section 9.09 — Dispute Resolution

Dispute Resolution — humans initiate disputes within policy window (RECOMMENDED: 90 days). Dispute tiers: publisher mediation, marketplace arbitration, regulatory escalation. Dispute outcomes affect trust and risk scores. Dispute fund from trust fees MAY compensate validated claims. Dispute proceedings MUST NOT require human waiver of sovereignty rights.

Section 9.10 — Trust Score Computation

Trust score \( T \) SHOULD compose weighted factors:

Weights MAY vary by taxonomy; composition MUST be disclosed to humans inspecting agent detail.

Section 9.11 — Trust Transaction Failure Modes

PART X — Companion Commerce

Section 10.01 — Companion-Mediated Commerce Overview

Companion Purchases flow through Keyra Companion mediation — Companion presents agent proposals, enforces spending controls, records human decisions, and NEVER holds root financial authority. Companion is negotiator and witness, not purchaser.

Section 10.02 — Companion Recommendations

Companion MAY recommend agents and offers based on Life Graph context, trust scores, and human preferences — with disclosed recommendation logic. Recommendations MUST NOT manipulate through artificial urgency or hidden sponsorship.

Section 10.03 — Companion Negotiations

Companion Negotiations — Companion orchestrates multi-agent comparison, price negotiation, and terms review. Negotiation transcripts MUST be human-inspectable. Binding commitment requires human authorization at appropriate authority level.

Section 10.04 — Companion Procurement

Companion Procurement — enterprise Companion coordinates departmental agent procurement with Organization Graph approval workflows. Procurement MUST respect enterprise budget and personal sovereignty separation.

Section 10.05 — Companion Subscription

Companion manages subscription lifecycle — renewal alerts, cancellation assistance, grant alignment with subscription status. Subscription lapse MUST trigger agent grant review.

Section 10.06 — Companion Spending Control

Companion Spending Control — human-declared caps, category restrictions, velocity limits, approval thresholds. Companion Spending Control MUST enforce at authorization engine — not merely Companion UI warnings.

Section 10.07 — Purchase Authorization Flow

Section 10.08 — Illustrative Scenario — Family Spending

Teenage son's shopping agent finds $40 sneakers. Family policy: $25 auto-approve, above requires parent Keyra Key. Companion routes to parent watch; parent approves with P4; transaction executes; son receives confirmation; family budget ledger updates.

PART XI — Family Marketplace

Section 11.01 — Family Marketplace Overview

The Family Marketplace extends marketplace governance to household economics — family agents, child-safe catalogs, education services, family budgets, and approval structures under Family Trust Network conformance.

Section 11.02 — Family Agents

Family Agents coordinate household logistics — calendars, chores, meal planning, elder check-in. Family agents access Family Vault partitions per Family Constitution. Family agents MUST NOT access adult personal vault partitions without individual grant.

Section 11.03 — Child-Safe Agents

Child-Safe Agents carry child-safe certification badge — content filtering, no unsolicited outbound communication, no financial scope, no data sale, guardian visibility. Child-safe review MUST occur annually. Child-safe agents MUST NOT use dark patterns or engagement optimization on minors.

Section 11.04 — Education Agents

Education Agents in Family Marketplace serve homeschooling, tutoring, and supplemental learning. Education agents MUST declare curriculum standards alignment where claimed. Progress data flows to Family Vault with retention policy.

Section 11.05 — Family Services

Family Services — shared subscriptions, household maintenance agents, family travel coordination. Family services billing MAY use family payment method with guardian authorization.

Section 11.06 — Family Budgets

Family Budgets — pooled or allocated budgets per member, category, and period. Family budgets enforce at authorization engine. Budget overrun MUST notify guardians; MUST NOT silently charge alternate payment methods without grant.

Section 11.07 — Family Governance

Family Governance — Family Constitution defines marketplace rules: catalog visibility, certification minimums, approval chains, spending caps. Family governance MUST be amendable by constitution process — not platform override.

Section 11.08 — Family Approval Structures

Section 11.09 — Illustrative Scenario — Child-Safe Enforcement

Child attempts to authorize social agent without child-safe badge. Family Marketplace runtime rejects; Companion explains; guardian receives notification. Child-safe approved education agent installs successfully with guardian grant.

PART XII — Enterprise Marketplace

Section 12.01 — Enterprise Marketplace Overview

The Enterprise Marketplace serves Organization Graph deployments — enterprise agents, department agents, compliance overlays, and executive protection under Organization Graph Enterprise Companion Framework.

Section 12.02 — Enterprise Agents

Enterprise Agents automate organizational workflows — document processing, ticketing, reporting. Enterprise agents access Organization Vault per role-based scope. Enterprise agent deployment requires enterprise administrator authorization; member humans retain personal marketplace sovereignty.

Section 12.03 — Department Agents

Department Agents — scoped to finance, HR, legal, research, operations departments. Department agents MUST NOT cross department boundaries without explicit grant. Department budget caps enforce similarly to personal spending controls.

Section 12.04 — Compliance Agents

Compliance Agents — regulatory monitoring, policy enforcement, audit preparation. Compliance agents read Organization Vault; they MUST NOT modify human personal records. Compliance findings flow to authorized administrators with audit.

Section 12.05 — Finance Agents

Finance Agents — accounts payable, expense reporting, procurement. Finance agents require Enterprise Certified minimum for transactions above enterprise policy threshold. Finance agents integrate bank authorization chains.

Section 12.06 — HR Agents

HR Agents — onboarding, benefits enrollment, policy Q&A. HR agents access bounded employee data per GDPR-equivalent minimization. HR agents MUST NOT infer health or genetic information beyond authorized scope.

Section 12.07 — Research Agents

Research Agents — literature review, data analysis, experiment tracking. Research agents accessing sensitive data require Enterprise Certified and data handling attestation.

Section 12.08 — Executive Agents

Executive Agents — calendar, travel, communication triage for executives. Executive agents carry enhanced security requirements — Device Trust Mesh binding, Trusted minimum, optional dedicated Keyra Key policy.

Section 12.09 — Enterprise Departure

On employment termination, enterprise agent grants to Organization Vault MUST auto-revoke. Personal agent grants MUST survive. Enterprise Marketplace MUST execute departure workflow within 24 hours.

PART XIII — Banking Marketplace

Section 13.01 — Banking Marketplace Overview

The Banking Marketplace integrates financial institutions into trust-based agent commerce — payments, transfers, investments, insurance, credit, compliance, fraud prevention, and authorization chains.

Section 13.02 — Financial Agents

Financial Agents execute banking operations under human grant and bank policy. Financial agents MUST hold Enterprise Certified minimum. Financial agents MUST NOT commingle human funds without explicit sweep authorization.

Section 13.03 — Payment Agents

Payment Agents — card-present equivalent, ACH, real-time payments, cross-border remittance. Payment agents validate authorization chain before settlement. Payment agents MUST support human transaction dispute initiation.

Section 13.04 — Transfer Agents

Transfer Agents — account-to-account, wire, scheduled transfers. Transfer agents REQUIRE A3 authority and P4+ presence for amounts above policy. Transfer agents MUST validate payee against Relationship Vault trusted payee list when configured.

Section 13.05 — Investment Agents

Investment Agents — portfolio management, trade execution, rebalancing. Investment agents MUST disclose conflicts of interest. Trade execution above threshold REQUIRES human confirmation per grant.

Section 13.06 — Insurance Agents

Insurance Agents — quote, bind, claim. Insurance agents require publisher insurance license verification per jurisdiction. Claims initiation MUST be human-authorized.

Section 13.07 — Credit Agents

Credit Agents — loan application, credit monitoring. Credit agents MUST NOT initiate hard inquiries without explicit human grant. Credit data access minimizes to authorized refs.

Section 13.08 — Compliance Agents (Banking)

Banking Compliance Agents — AML screening, sanctions check, transaction monitoring. Compliance agents operate on bank infrastructure; they MUST NOT export human data to non-accredited parties.

Section 13.09 — Fraud Prevention

Fraud Prevention — anomaly detection, device binding, velocity checks, trust score integrations. Fraud blocks MUST notify human with remediation path. Fraud false positive dispute MUST be human-resolvable within SLA.

Section 13.10 — Authorization Chains

Authorization Chains — cryptographic chain from human root through Companion, agent, bank API:

Human Root → KAAI Grant → Agent Certificate → 
Companion Attestation → Bank Authorization Token → Settlement

Each link MUST be validated. Broken chain MUST halt settlement.

Section 13.11 — Illustrative Scenario — Wire Transfer

Human authorizes $50,000 wire via banking agent. Chain validates: human grant (A3), agent Enterprise Certified, trust score 0.88, Device Trust Mesh P5 Keyra Key, bank KYC current. Trust escrow holds 24 hours per policy. Human confirms release; wire executes; audit complete.

PART XIV — Telecommunications Marketplace

Section 14.01 — Telecom Marketplace Overview

The Telecommunications Marketplace integrates carriers into agent commerce — subscriber services, identity provisioning, eSIM commerce, network trust, and provisioning agents.

Section 14.02 — Subscriber Agents

Subscriber Agents — plan management, usage monitoring, bill negotiation. Subscriber agents access carrier APIs under human grant. Subscriber agents MUST NOT modify plan without human confirmation.

Section 14.03 — Identity Agents

Identity Agents — SIM binding, number porting, identity verification for carrier services. Identity agents integrate Identity Vault and Device Trust Mesh. Identity agents MUST NOT usurp human root over device trust.

Section 14.04 — eSIM Agents

eSIM Agents — profile provisioning, travel eSIM purchase, multi-carrier management. eSIM agents REQUIRE Trusted certification for profile download. eSIM commerce MUST bind to device_id per Device Trust Mesh.

Section 14.05 — Provisioning Agents

Provisioning Agents — network service activation, IoT connectivity, enterprise fleet provisioning. Provisioning agents operate under carrier accreditation; human grant required for personal lines.

Section 14.06 — Trust Agents (Telecom)

Telecom Trust Agents — network attestation, spam filtering, call authentication (STIR/SHAKEN equivalent). Trust agents enhance Device Trust Mesh network overlay without surveillance beyond declared scope.

Section 14.07 — Network Agents

Network Agents — quality monitoring, outage notification, mesh handoff optimization. Network agents MUST NOT inspect application content; network metadata only per policy.

Section 14.08 — Carrier Subordination

Carriers participate as accredited publishers — not human root. Carrier agents MUST respect human revocation over subscriber services. Carrier lock-in prohibiting ACEP export is prohibited.

PART XV — Government Marketplace

Section 15.01 — Government Marketplace Overview

The Government Marketplace publishes citizen service agents — identity, permits, healthcare, tax, education, national services — under Government Certified or Critical Infrastructure certification.

Section 15.02 — Citizen Agents

Citizen Agents — benefit enrollment, service requests, status tracking. Citizen agents MUST be human-authorizable. Citizen agents MUST NOT condition essential services on unrelated data collection.

Section 15.03 — Identity Agents (Government)

Government Identity Agents — national ID, passport renewal, driver's license. Identity agents integrate government accreditation. Identity agents MUST NOT create parallel identity root — human sovereignty prevails.

Section 15.04 — Permit Agents

Permit Agents — building permits, business licenses, regulatory filings. Permit agents guide application; submission REQUIRES human authorization. Permit agents MUST provide audit trail for administrative law review.

Section 15.05 — Healthcare Agents (Government)

Government Healthcare Agents — national health system enrollment, public health programs. Healthcare agents access Health Vault with Trusted minimum and lawful basis.

Section 15.06 — Tax Agents

Tax Agents — filing preparation, payment scheduling, refund tracking. Tax agents MUST NOT file without human A3 authorization. Tax data MUST minimize to filing requirement.

Section 15.07 — Education Agents (Government)

Government Education Agents — public school enrollment, student aid, credential verification. Education agents serving minors require child-safe conformance.

Section 15.08 — National Services Agents

National Services Agents — veterans services, social security equivalent, emergency alerts. National services agents affecting safety MUST carry Critical Infrastructure certification where applicable.

Section 15.09 — Government Accountability

Government publishers MUST provide: lawful basis per agent, appeal pathway, human-readable scope manifest, and audit export for FOIA-equivalent requests without compromising other citizens' data.

PART XVI — Agent Reputation

Section 16.01 — Scoring Framework Overview

The Companion Marketplace maintains multi-dimensional scoring — reputation, trust, usage, quality, risk — informing discovery, certification maintenance, and authorization thresholds.

Section 16.02 — Reputation Scores

Reputation Scores reflect long-term publisher and agent standing — transaction satisfaction, dispute resolution, community reports. Reputation MUST be resistant to review bombing via verified transaction weighting.

Section 16.03 — Trust Scores

Trust Scores (Part IX) reflect operational reliability and incident history. Trust scores gate scope activation per Part VII trust levels.

Section 16.04 — Usage Scores

Usage Scores reflect adoption and engagement — install count, active grants, retention. Usage informs discovery ranking but MUST NOT override certification or trust requirements for high-risk categories.

Section 16.05 — Quality Scores

Quality Scores reflect capability delivery — task completion, latency, accuracy benchmarks. Quality audits MAY be third-party for Trusted and above.

Section 16.06 — Risk Scores

Risk Scores reflect security and liability exposure — vulnerability history, data handling incidents, scope violation attempts. Elevated risk triggers certification review and discovery deprioritization.

Section 16.07 — Complaint Framework

Complaint Framework — humans file complaints against agents or publishers. Complaints MUST be triaged within SLA. Valid complaints affect trust and risk scores. Frivolous complaint patterns MUST NOT weaponize against publishers without audit.

Section 16.08 — Certification Framework

Certification Framework (Part VI) integrates with scoring — certification maintenance requires sustained trust and quality thresholds. Automatic downgrade on sustained breach.

Section 16.09 — Revocation Framework

Revocation Framework — human grant revocation, publisher certificate revocation, emergency marketplace suspension. Revocation propagation SLA: RECOMMENDED < 60 seconds personal; < 5 minutes global federation.

Section 16.10 — Scoring integrations Matrix

PART XVII — Marketplace Governance

Section 17.01 — Governance Overview

Marketplace Governance — publishing rules, audit requirements, approval workflows, compliance monitoring, revocation procedures, retirement rules — ensures marketplace serves human sovereignty, not platform extraction.

Section 17.02 — Publishing Rule

Publishing REQUIRES: valid registration, achieved certification, published scope manifest, privacy policy, liability attribution, human-readable description, fee disclosure. Prohibited content: surveillance agents, attention manipulation agents, anti-portability agents, human impersonation agents.

Section 17.03 — Audit Rule

Audit Rule escalate with certification:

Section 17.04 — Approval Rule

Approval Rule — automated for Verified and below when audit passes; human review panel for Government and Critical Infrastructure; enterprise admin approval for Enterprise Store private listings.

Section 17.05 — Compliance Rule

Compliance Rule — continuous scope violation detection, transaction anomaly analysis, certification condition monitoring. Compliance failures trigger investigation — not silent tolerance.

Section 17.06 — Revocation Rule

Revocation grounds: scope violation, incident, certification failure, publisher fraud, human safety, lawful order. Revocation MUST preserve audit history. Appeal pathway MUST exist for publishers; human safety revocations MAY be immediate without appeal delay.

Section 17.07 — Retirement Rule

Retirement REQUIRES: 90-day notice for Enterprise Certified and above unless security emergency; migration guide; grant expiration schedule; audit archive transfer to Trust Vault refs.

Section 17.08 — Governance Council

Governance Council — multi-stakeholder body: human advocates, publishers, banks, telcos, governments, security auditors. Council amends marketplace policy — not Human Sovereignty Charter. Council proceedings MUST be published with redaction for security.

Section 17.09 — Illustrative Scenario — Emergency Revocation

Critical Infrastructure power grid agent exhibits anomalous scope expansion attempt. Automated compliance flags incident; marketplace suspends agent within 30 seconds; existing grants halt; national incident protocol activates; human operators receive Keyra Key confirmation request for manual override only.

PART XVIII — Global Agent Registry

Section 18.01 — Global Registry Overview

The Global Agent Registry federates national and regional Agent Registries — enabling discovery, certification validation, trust score lookup, and cross-border commerce without single-marketplace capture.

Section 18.02 — Global Discovery

Global Discovery — federated query across registry shards by taxonomy, certification, trust score, geography. Global Discovery MUST NOT require central human surveillance database. Bloom filters and edge cache RECOMMENDED at scale.

Section 18.03 — Global Certification

Global Certification — national accreditations map to global certification levels. Government Certified in one jurisdiction MUST declare equivalence or restriction in others. Humans MUST see certification jurisdiction badge.

Section 18.04 — Global Trust

Global Trust — trust scores and incident reports share across registries with publisher consent and lawful restriction. False incident injection MUST be cryptographically attributable and penalized.

Section 18.05 — Global Publishing

Global Publishing — agent listed once, visible globally with geographic and certification filters. Publisher MAY restrict distribution; human grant MAY authorize cross-border agent not in local store.

Section 18.06 — Cross-Border Trust

Cross-Border Trust — trust scores port; certification badges display with jurisdiction; escrow MAY require enhanced trust fee for cross-border settlement.

Section 18.07 — Cross-Border Authorization

Cross-Border Authorization — human grant valid globally unless human restricts geography. Agent execution MUST comply with local law; authorization engine MUST geo-fence when required.

Section 18.08 — Cross-Border Commerce

Cross-Border Commerce — settlement respects currency, tax, sanctions screening. Banking authorization chains MUST include compliance agent validation for international transfers.

Section 18.09 — Registry Sharding

Registry shards by geography — no single query path for all agents. Revocation propagates via federated pub/sub. RECOMMENDED revocation propagation < 5 minutes global.

Section 18.10 — Illustrative Scenario — Cross-Border Agent

Human in Canada authorizes EU-published Trusted travel agent. Discovery finds agent via federation; certification equivalence displayed; cross-border trust fee disclosed; human grants; agent operates within grant; settlement in CAD with disclosed FX; audit chain complete.

PART XIX — Marketplace Security

Section 19.01 — Security Overview

Marketplace Security protects humans, agents, vaults, identities, finances, families, and enterprises from agent compromise, scope escalation, and supply chain attack.

Section 19.02 — Agent Isolation

Agent Isolation — agents execute in sandboxed runtimes with declared scope enforcement. Agent MUST NOT access vault partitions beyond grant. Agent MUST NOT communicate laterally with other agents without explicit delegation grant.

Section 19.03 — Permission Isolation

Permission Isolation — authorization engine enforces scope at runtime — not merely at install. Permission reduction applies immediately. Privilege escalation attempts MUST trigger incident and trust score penalty.

Section 19.04 — Vault Protection

Vault Protection — agent access to Trust Vault via derived keys only; no root key exposure. Vault reads MUST be logged with artifact ref — not plaintext in agent logs. High-sensitivity vault access REQUIRES Device Trust Mesh presence.

Section 19.05 — Identity Protection

Identity Protection — agents MUST NOT store human credentials; OAuth and derived tokens only. Identity impersonation prohibited. Agent identity MUST be distinct from human identity in all communications.

Section 19.06 — Financial Protection

Financial Protection — spending controls, velocity limits, trust escrow, fraud prevention, authorization chain validation. Financial agent compromise MUST limit blast radius to granted caps.

Section 19.07 — Family Protection

Family Protection — child-safe enforcement, guardian approval, catalog filtering, no minor data sale. Family agent access MUST respect Family Constitution.

Section 19.08 — Enterprise Protection

Enterprise Protection — Organization Vault isolation, departure revocation, compliance monitoring, executive enhanced binding. Supply chain attack on enterprise agent MUST trigger fleet suspension capability.

Section 19.09 — Security Incident Response

Section 19.10 — Security Conformance Checklist

PART XX — Future Scale

Section 20.01 — Scale Overview

The Companion Marketplace MUST scale from one agent to one hundred billion agents without constitutional redesign — scaling caching, sharding, and federation — not principles.

Section 20.02 — One Million Agents

At one million agents: single-region registry adequate; trust score computation batch nightly; discovery sub-100ms p99; revocation propagation < 60 seconds regional.

Section 20.03 — One Hundred Million Agents

At one hundred million agents: registry sharded by geography and taxonomy; edge trust score cache; bloom-filter revocation; discovery federated; 95% authorization decisions local at Companion edge.

Section 20.04 — One Billion Agents

At one billion agents: hierarchical registry — global index, regional authority, publisher shard; trust score approximate at edge with authoritative reconcile; cross-border commerce via standardized settlement protocol; human grant graph remains personal-scale — not billion-node central query.

Section 20.05 — One Hundred Billion Agents

At one hundred billion agents: IoT and infrastructure agents dominate count; gateway agents aggregate authorization; Critical Infrastructure monitoring continuous; human-facing agents remain minority but highest value; civilization-scale incident response federation mandatory.

Section 20.06 — Scale Invariants

At any scale: human root authority, agent accountability, revocable grants, auditable transactions, portable ACEP export. Scale changes infrastructure — not principles.

Section 20.07 — Performance Benchmarks (Target)

Section 20.08 — Population Scenarios

Scenario — Average human 25 agents: Personal shopping, travel, health, finance, home, work, family — each with scoped grants. Companion edge cache holds active grant set; vault holds authoritative certs.

Scenario — Enterprise 10,000 agents: Department fleets share certification; compliance agent monitors scope; departure workflow batch revokes 500 grants nightly.

Scenario — National citizen services: Government publishes 200 Government Certified agents; 50 million citizens authorize subsets; registry shard per region; no central citizen-agent graph query.

Section 20.09 — Migration Path from Today's Internet

Phase 1: Agent registration and Verified certification for consumer agents (years 0–5). Phase 2: Trust transactions and bank authorization chains (years 3–10). Phase 3: Family and enterprise marketplace governance (years 5–15). Phase 4: Global federation and government accreditation (years 10–25). Parallel operation with legacy app stores during transition — default deny for vault financial scope without KAAI grant increasing over time.

Section 20.10 — Infrastructure Scaling Patterns

Registry sharding — hash agent_id to regional shard; cross-shard lookup via global index with bloom-filter negative confirmation. Trust score cache — Companion edge holds 15-minute TTL cache; authoritative reconcile on consequential transaction. Settlement batching — micro-transactions aggregate hourly below human-configured batch threshold to reduce rail fees. Revocation fanout — pub/sub with CRDT merge for partition tolerance; humans see eventual consistency within SLA, never unbounded delay.

Section 20.11 — Agent Density Projections

Projections inform capacity planning; they do not mandate human agent adoption. Human sovereignty includes declining agent authorization entirely.

Section 20.12 — Research and Standards Alignment

Companion Marketplace aligns with emerging standards:

• W3C Decentralized Identifiers for agent and publisher identity
• W3C Verifiable Credentials for certification badges
• ISO 20022 for financial settlement messaging overlays
• PSD2/Open Banking for bank authorization chain patterns
• NIST AI Risk Management Framework for agent risk scoring
• COPPA/GDPR for family and data minimization overlays

Alignment is interoperable, not dependent — marketplace sovereignty survives standards evolution via migration layers.

PART XXI — Economic Civilization Layer

Section 21.01 — Civilization Overview

Companion Marketplace at civilization scale is economic infrastructure — like payment networks, securities exchanges, commercial law — enabling agent commerce without centralized human surveillance or platform extraction.

Section 21.02 — Personal Economies

Every human operates Personal Economies — personal agent grants, spending policies, trust relationships with publishers — analogous to personal financial relationships in physical economy.

Section 21.03 — Family Economies

Family Economies federate household agent commerce — shared budgets, child-safe catalogs, elder care agents — under Family Constitution.

Section 21.04 — Enterprise Economies

Enterprise Economies — organizations deploy agent fleets, procurement, compliance — subordinate to member human personal economies.

Section 21.05 — National Economies

National Economies — nations accredit Government and Critical Infrastructure agents, regulate settlement, protect citizen revocation — without owning human agent graphs.

Section 21.06 — Global Trust Economies

Global Trust Economies — federated registries, cross-border settlement, standards governance council — no single sovereign marketplace operator.

Section 21.07 — Civilizational Invariants

At any scale: human sovereignty over transactions, agent accountability, trust-based commerce, authorization over engagement, portable economic state. Civilization scale changes throughput — not constitutional subordination.

Section 21.08 — Economic Layer Placement

┌──────────────────────────────────────────────┐
│     Global Trust Economy                    │
├──────────────────────────────────────────────┤
│     National · Enterprise · Family          │
├──────────────────────────────────────────────┤
│     Personal Trust Economy (per human)      │
├──────────────────────────────────────────────┤
│     Companion Marketplace Core              │
└──────────────────────────────────────────────┘

Each layer composes without merging human root sovereignty.

Section 21.09 — Trust Economy Interoperability

Personal economies interoperate with family, enterprise, and national layers through explicit grant — not automatic merge. ACEP export enables human migration between economic contexts without data loss.

Section 21.10 — Economic Civilization Scenarios

Scenario — Personal to family transition: Adult child leaves household. Family agent grants revoke per constitution; personal agent grants port via ACEP; family budget allocation ends; personal spending controls activate unchanged.

Scenario — Enterprise to entrepreneur: Employee departs to found startup. Enterprise agent grants auto-revoke within 24 hours; personal grants survive; enterprise-published agents no longer accessible; personal marketplace sovereignty intact.

Scenario — Cross-national relocation: Human emigrates. ACEP exports; national Government Store agents require re-accreditation discovery; personal grants persist; bank authorization chains re-KYC per destination; human root unchanged.

Section 21.11 — Civilizational Risk Posture

Societies without agent accountability architecture inherit automation liability without law. Societies that centralize agent authorization in platforms sacrifice sovereignty for convenience. Societies that abandon marketplace governance inherit attention economics at civilization scale. The Companion Marketplace at scale is infrastructure for agent civilization — as essential as commercial law, as personal as family budget, as enduring as authorization grant chains.

PART XXII — Closing Declaration

Section 22.01 — On Trust Over Attention

We declare that trust over attention governs the Companion Marketplace — agents earn authorization through certification and accountability, not through engagement manipulation, artificial urgency, or surveillance monetization. Humans grant; agents serve; platforms facilitate — none extract.

Trust matters because attention economics treats humans as inventory; trust economics treats humans as sovereign persons.

Section 22.02 — On Authorization Over Engagement

Authorization over engagement — every consequential agent action requires validated human grant chain, not merely session persistence or dark-pattern consent. Engagement metrics MUST NOT override authorization requirements.

Authorization matters because agents that act without grant are trespassers with API keys.

Section 22.03 — On Agent Accountability

Agent accountability — publishers, certification, audit, liability, revocation — is non-negotiable. Anonymous consequential agents are prohibited. Experimental agents carry warnings. Critical Infrastructure agents carry bonds and continuous monitoring.

Accountability matters because civilization-scale agent commerce without liability is civilization-scale liability without law.

Section 22.04 — On Trusted Relationships Over Applications

Trusted relationships over applications — humans authorize agents as ongoing relationships with scopes, trust scores, and revocation — not as one-time app installs with immortal OAuth tokens. The Companion Marketplace models economic life as grant graphs, not app icons.

Relationships matter because digital life is continuous; application metaphors are disposable.

Section 22.05 — On Human Economic Sovereignty

Humans deserve to own their agent authorization graphs — inspect grants, revoke instantly, export portably, inherit intentionally. Economic sovereignty requires marketplace architecture that persists, audits, and answers to the human root.

Sovereignty matters because agent commerce without human root becomes institutional commerce over human life.

Section 22.06 — On Technology and Commerce

Technology promised frictionless commerce and delivered opaque extraction. It promised assistants and delivered unaccountable automations. It promised marketplaces and delivered walled gardens. This framework ends the confusion. Commerce should serve authorized human intent — across agents, across families, across enterprises, across borders — under human authority.

Section 22.07 — Timeless Commitment

We declare the Companion Marketplace & Agent Economy the canonical economic framework for trust-based agent commerce in the Keyra Companion Ecosystem — and a reference for any civilization that dares to treat humans as sovereign economic persons rather than transaction feedstock.

The marketplace is not an app store. The marketplace is a trust. The trust belongs to the human. Always.

Section 22.08 — Invocation

To every human authorizing their first agent: you grant what only you possess. To every parent provisioning child-safe agents: you protect with architecture, not hope. To every publisher building agents: map every capability to scope — or do not publish. To every bank, telco, and government participating: accredit without usurping; facilitate without capturing.

Section 22.09 — Canonical Status

This instrument joins the founding frameworks of the Keyra Companion Ecosystem as the authoritative reference for agent registration, certification, permission scopes, trust transactions, revenue models, domain marketplaces, global federation, and economic civilization scale — for the century ahead and beyond.

Implementers of agent platforms, app stores, payment systems, and automation tools must map every capability to explicit sections of this framework. Features without marketplace governance mapping are design defects. Features that weaken human root authority for platform revenue are constitutional violations.

Section 22.10 — On the Century Ahead

This framework is written for humans not yet born — whose great-grandparents today authorize their first shopping agent. Agents will multiply beyond present imagination. Settlement rails will evolve. Certification standards will tighten. The invariants must not change: human sovereignty over transactions, agent accountability, trust-based commerce, authorization over engagement, portable economic state. Implementation may evolve; principles endure.

The Companion Marketplace is the ledger — not of attention, but of authorized trust in the agent era.

Section 22.11 — Acknowledgment of Founding Instruments

This Framework stands with the Human Sovereignty Charter, Trust Vault Architecture, Device Trust Mesh, KAAI Standard, Life Graph Architecture, Companion Charter, Human Digital Twin Architecture, Family Trust Network, Organization Graph Enterprise Companion Framework, and Life Operating System as co-equal founding architecture of the Keyra Companion Ecosystem — each necessary, none sufficient alone.

Implementers who build agents without vault integrations build unauthorized executors. Implementers who build vaults without marketplace governance build unlockable commerce. The ecosystem requires both.

Section 22.12 — Perpetual Subordination Clause

This Framework may be amended for technical agility — fee models, certification criteria, federation protocols. It may never be amended to permit platform root over human economic authority, unauditable consequential agent action, compulsory marketplace capture without portability, or attention-based commerce overriding authorization requirements. Such amendments are void ab initio per the Human Sovereignty Charter.

Section 22.13 — Declaration on Families

We declare that families require governed agent commerce — child-safe certification, approval structures, shared budgets — that protects minors without surveilling them and empowers guardians without platform override.

Section 22.14 — Declaration on Institutions

We declare that banks, telcos, enterprises, and governments participate as accredited facilitators and certifiers, not as sovereigns over personal agent authorization graphs. Institutional convenience MUST NOT weaken human revocation.

Section 22.15 — Call to Implementers

To agent developers: publish with certification, scope manifests, and accountability. To banks: validate authorization chains, not merely API keys. To marketplaces: disclose fees, enable portability, reject attention optimization. To humans: demand KAAI conformance for agents that touch your money, children, health, and legal standing.

Section 22.16 — Call to Nations

To nations: accredit government agents, protect citizen revocation rights, participate in global federation without capture, and never treat agent commerce as exempt from accountability because automation is novel.

Section 22.17 — Glossary of Core Terms

Section 22.18 — Conformance Self-Assessment Checklist

Organizations MAY use this checklist for Companion Marketplace readiness:

Conformance is a journey; partial conformance with documented roadmap is valid if audited.

Section 22.19 — Document Maintenance

Companion Marketplace & Agent Economy 1.0 is the founding framework. Technical errata publish quarterly. Minor version (1.x) may add agent taxonomies and store types. Major version (2.0) requires Governance Council supermajority and human rights review. Constitutional subordination to the Human Sovereignty Charter is immutable across all versions.

Section 22.20 — Final Affirmation

We affirm that every human — infant with a family education agent, elder with a healthcare agent, refugee with a single authorized companion, executive with an enterprise fleet, farmer with an agricultural agent — deserves agent commerce that operates under their authority, with accountability, across marketplaces, across generations, across mortality's threshold to those they designate.

This is not a privilege of the technical elite. It is infrastructure of human dignity in the age of ubiquitous agents.

Trust over attention. Authorization over engagement. Agent accountability. Trusted relationships over applications. The marketplace belongs to the human. The human belongs to themselves. Always.

End of Document

The Companion Marketplace & Agent Economy v1.0 — Founding Framework of the Keyra Companion Ecosystem