<?xml version="1.0" encoding="UTF-8"?>
<rfc xmlns:xi="http://www.w3.org/2001/XInclude" category="info" docName="draft-das-evaluator-governance-00" ipr="trust200902" submissionType="independent" xml:lang="en" version="3" tocInclude="true" tocDepth="3" sortRefs="true" symRefs="true">
<front>
<title abbrev="Evaluator Authority and Lifecycle">Location-Independent Evaluator Governance for Autonomous Multi-Agent Systems: Protected Appointment, Evidence-Bounded Decisions, Cross-Domain Coordination, and Separation of Execution Authority</title>
<seriesInfo name="Internet-Draft" value="draft-das-evaluator-governance-00"/>
<author fullname="Sangam Kumar Das" initials="S." surname="Das">
<organization>Independent Inventor</organization>
<address>
<postal><city>Balasore</city><region>Odisha</region><code>756001</code><country>India</country></postal>
<phone>+91 98613 63532</phone>
<email>info@sangamdas.com</email>
</address>
</author>
<date year="2026" month="November" day="16"/>
<area>Security</area>
<workgroup>Independent Submission</workgroup>
<keyword>evaluator</keyword><keyword>multi-agent AI</keyword><keyword>appointment</keyword><keyword>review round</keyword><keyword>execution finality</keyword><keyword>AI safety</keyword>
<abstract>
<t>Autonomous agents increasingly rely on an evaluator to review plans, evidence and outcomes before work continues, yet an evaluation result is often mistaken for permission to act. This document describes a vendor-neutral architecture in which the evaluator is a functional role, not a product: it may be a rule engine, model, program, enclave, service or human, placed locally, remotely or on a shared platform. It separates six independent lifecycles, requires a protected task-specific appointment derived from resource-owner authority, admits evidence by source, role, freshness and independence, preserves mandatory dissent, and emits typed determinations such as HOLD, DISPUTED and UNKNOWN. Its central rule is that a valid determination is not an executable authorization: separate owner-controlled authorization and a native enforcement boundary remain necessary. Twenty-seven embodiments, a threat model, data model and examiner index are provided.</t>
</abstract><note title="Source and Patent Information">
<t>Technical source: Indian Provisional Patent Application 202631131650, dated 3 November 2026 (as reported by the applicant; corroborated only by an eFiling payment record showing CBR date 03/11/2026, not by the official acknowledgment, which the editor has not seen), titled "Systems and Methods for Location-Independent Evaluator Architectures, Dynamic and Persistent Evaluation Lifecycles, Protected Appointment, Evidence Verification, and Authority-Separated Decision Control in Autonomous and Distributed Systems". The source states 27 embodiments and 540 numbered claims (60 independent, 480 dependent). Foundational reference: PCT/IB2026/055615, filed 4 June 2026, published 16 July 2026 as WO 2026/150382 A2. See <xref target="ipr"/>.</t>
<t>Where to read the full PCT filing: search "Sangam Das" on WIPO PATENTSCOPE; search PCT/IB2026/055615 (pctib202655615); or read the PCT file directly in Technical Disclosure Commons at https://www.tdcommons.org/dpubs_series/12542/.</t>
<t>The IPR disclosure for this document will be made at the appropriate time through the official IETF process and in accordance with IETF guidelines (see <xref target="ipr"/>).</t>
<t>AI-assisted drafting was used. Readers must verify patent numbers, dates, claim text, product capabilities and standards references against primary sources. This is an individual proposal, not an IETF standard and not an endorsed position.</t>
</note></front>
<middle>
<section anchor="intro">
<name>Introduction</name>
<t>Autonomous and multi-agent systems increasingly place an evaluator between a proposed plan and the point where an effect occurs. The evaluator may inspect evidence, estimate consequences, check policy, review an outcome, and recommend whether work should continue. Three questions are frequently merged: who may evaluate, what evidence the evaluator may rely on, and who may actually cause the effect. This document keeps them apart.</t>
<t>The document describes one architecture, drawn from a single source filing with 27 embodiments, in an individual, informational form. It does not define a new wire protocol; it specifies functional requirements, candidate record schemas and test conditions that existing mechanisms (see <xref target="interop"/>) can carry.</t>
<t>Central rule: a valid evaluation determination is not automatically an executable authorization.</t>
<section anchor="problem">
<name>Problem Statement</name>
<t>Existing practice commonly shows one or more of the following gaps: an evaluator is chosen by the agent it reviews; independence is counted as accounts rather than control groups; a signed report is treated as truth about the physical or ledger outcome; an approval is reused after its appointment, round or policy generation has changed; a majority hides a mandatory dissent; a lost acknowledgment is retried and duplicates an effect; and a coordinator's DONE message is treated as closure while effect-capable work remains.</t>
</section>
<section anchor="scope">
<name>Scope and Non-Goals</name>
<ul>
<li>In scope: functional evaluator role, protected appointment, lifecycles, evidence admission, typed non-executable determinations, review rounds, multi-agent controls, cross-domain governance, recovery, revocation, membership, minimal-disclosure evidence, and partial-outcome handling.</li>
<li>Not in scope: a new transport or wire format; a claim that evaluators are accurate; a claim of global atomicity; instantaneous revocation during disconnection; certification; patentability or infringement conclusions; any company endorsement.</li>
<li>Distinct from related drafts: causal semantic diagnosis (see <xref target="REL-CAUSAL"/>) diagnoses where meaning changed; execution finality (see <xref target="REL-FOUND"/>, <xref target="REL-CBEA"/>, <xref target="REL-IEV"/>) governs whether an effect is admitted. This document governs who may evaluate, on what evidence, for how long, with what typed output, and how that output relates to the separate authority.</li>
</ul>
</section>
</section>
<section anchor="terms">
<name>Terminology and Conventions</name>
<t>The key words in this document follow BCP 14 <xref target="RFC2119"/> <xref target="RFC8174"/> only where requirement language is used, and sparingly. Most statements are architectural descriptions rather than protocol requirements.</t>
<dl newline="true">
<dt>Evaluator</dt>
<dd><t>A functional role that assesses evidence, a plan, a consequence or an outcome under a bounded mandate and produces a determination. It holds no effectuation authority by virtue of the role.</t></dd>
<dt>Supervisor / coordinator</dt>
<dd><t>A role that organizes tasks or agents and may carry out limited, authorized coordination. It is not an evaluator and not an owner.</t></dd>
<dt>Authorization issuer / owner-controlled authority</dt>
<dd><t>The party or process that, under resource-owner policy, issues a bounded permission for a specific act.</t></dd>
<dt>Native resource gate / Finality Sink</dt>
<dd><t>The enforcement boundary at which an effect is actually admitted or withheld by the resource owner.</t></dd>
<dt>Appointment</dt>
<dd><t>A protected, task-specific designation of an evaluator derived from the authority of the mandatory resource owners.</t></dd>
<dt>Determination</dt>
<dd><t>A typed, non-executable output of an evaluation, bound to task, round, policy generation and validity interval.</t></dd>
<dt>Control group</dt>
<dd><t>The set of credentials, hosts or principals that are in fact under one administrative or economic control.</t></dd>
<dt>Review round</dt>
<dd><t>A bounded set of reviews over a fixed evidence set that closes with completeness, independence and dissent checks.</t></dd>
<dt>Native outcome</dt>
<dd><t>Evidence originating at the resource owner or by protected direct observation about what actually occurred.</t></dd>
<dt>Protected-local state</dt>
<dd><t>State held inside a protected enforcement environment that is not exported as a portable signed artifact.</t></dd>
</dl>
</section>
<section anchor="arch">
<name>Principal Architecture and Trust Boundaries</name>
<t>The principal architecture is shown in <xref target="fig-main"/>. Evaluation authority ends at the boundary where a non-executable determination is produced; continuation is separately authorized and any effect is admitted only by the owner-controlled native boundary.</t>
<figure anchor="fig-main">
<name>Principal architecture (evaluation authority ends at the dotted line)</name>
<artwork type="ascii-art"><![CDATA[
 Agents / human actors
        |
        v
 Task, plan and evaluation request
        |
        v
 Authority and resource-owner resolution
        |
        v
 Evaluator eligibility + appointment verification
        |
   mandate valid? --NO--> DENY / HOLD
        | YES
        v
 Authorized evidence intake
        |
        v
 Source, role, task, freshness, truth checks
        |
        v
 Policy-bounded evaluation
        |
        v
 Review completeness, independence, dissent
        |  incomplete --> HOLD   contested --> DISPUTED
        |  unresolved --> UNKNOWN / REQUEST_EVIDENCE
        v
 Typed NON-EXECUTABLE determination
 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
 evaluator authority ends above; owner authority begins below
        v
 Separate protected continuation authorization
        |   (owner policy; may refuse)
        v
 Owner-controlled native Finality Sink
        |   admits exactly the authorized act or withholds
        v
 Native outcome evidence (success / partial / UNKNOWN)]]></artwork>
</figure>
<dl newline="true">
<dt>Branches</dt>
<dd><t>A failed mandate yields DENY or HOLD. Incomplete review yields HOLD. Contested review yields DISPUTED. Unresolved or missing evidence yields UNKNOWN or REQUEST_EVIDENCE. None of these is an authorization.</t></dd>
<dt>Trust boundaries</dt>
<dd><t>Between agents and the evaluator; between the evaluator and the authorization issuer; between the issuer and the native gate. No authority is inherited across a boundary <xref target="fig-trust"/>.</t></dd>
<dt>Where evaluator authority ends</dt>
<dd><t>At the typed determination. It is a premise that owner policy may consider, not a token that a gate honors.</t></dd>
</dl>
<figure anchor="fig-trust">
<name>Four roles and their trust boundaries</name>
<artwork type="ascii-art"><![CDATA[
 +-----------+   +-------------+   +-------------+   +-----------+
 | Evaluator |   | Supervisor/ |   | Authorization|  | Native    |
 | assesses  |   | coordinator |   | issuer/owner |  | gate/Sink |
 | evidence  |   | organizes   |   | grants bound |  | admits    |
 |           |   | tasks       |   | permission   |  | the effect|
 +-----+-----+   +------+------+   +------+-------+  +-----+-----+
       |  boundary 1    |   boundary 2    |  boundary 3    |
       +----- no authority is inherited across a boundary ------+
 One host may run several roles; co-location inherits nothing.]]></artwork>
</figure>
</section>
<section anchor="evalfn">
<name>Evaluator Functional Architecture and Location Independence</name>
<t>An evaluator is defined by what it does, not by what it is. The same function may be an in-process component, a local protected service, a hardware or firmware element, a trusted execution environment, a container or virtual machine, a remote service, a persistent platform, a human, a committee, or a hybrid. Rule engines, policy engines, solvers, statistical models, generative AI and verification programs are all admissible placements. A single-component evaluator is not excluded merely because it is a single component.</t>
<figure anchor="fig-loc">
<name>Evaluator function and location-neutral deployment</name>
<artwork type="ascii-art"><![CDATA[
 Same function, different placements:
 [in-process component]  [local protected service]  [TEE/hardware]
 [container/VM]          [remote service]           [human/committee]
        \                      |                       /
         +-- protected intake path (versioned evidence) --+
                           |
                    bounded determination
                           |
         protected output path (admissibility checked)
 Co-location is not non-independence; remote is not independent.
 Independence = real control relations, conflicts, access.]]></artwork>
</figure>
<t>Location independence is not absence of control. Co-location with an agent does not necessarily defeat independence if control relationships, evidence access and conflicts are separated; third-party hosting does not prove independence if the host is under common control or economic dependence. Independence is assessed on actual control relationships, evidence access, conflicts of interest and the ability to influence the outcome, under policy that remains owner-specific.</t>
</section>
<section anchor="deploy">
<name>Evaluator Deployment Models</name>
<dl newline="true">
<dt>Type I: Ephemeral</dt>
<dd><t>Created for one task or round, retired when appointment, round or validity ends. Strong isolation; setup cost per task <xref target="fig-eph"/>.</t></dd>
<dt>Type II: Persistent</dt>
<dd><t>A long-lived protected service that serves many independent projects, each with its own mandate, rounds, validity and retention <xref target="fig-per"/>.</t></dd>
<dt>Type III: Hybrid</dt>
<dd><t>A persistent platform hosting task-specific contexts with separate expiry, namespaces and revocation <xref target="fig-hyb"/>.</t></dd>
</dl>
<figure anchor="fig-eph">
<name>Ephemeral evaluator lifecycle</name>
<artwork type="ascii-art"><![CDATA[
 eligible? -> appoint -> activate instance -> intake evidence
    |NO                                          |
    v                                            v
  refuse                                  evaluate -> determination
                                                 |
                      handoff <------------------+
                         |
 appointment expires or is revoked --> instance retired
                         |
 evidence retained (protected); old authority NOT reusable]]></artwork>
</figure>
<figure anchor="fig-per">
<name>Persistent evaluator lifecycle</name>
<artwork type="ascii-art"><![CDATA[
 Host (long-lived)
   |-- Project A mandate --> rounds A1, A2   (expiry A)
   |-- Project B mandate --> rounds B1       (expiry B)
 Separate data and evidence per project.
 A expires: host and B continue; no inheritance from A.
 Platform compromise: revoke all mandates, revalidate each.]]></artwork>
</figure>
<figure anchor="fig-hyb">
<name>Hybrid evaluator platform</name>
<artwork type="ascii-art"><![CDATA[
 Persistent platform (shared maintenance)
   +- context X (org 1 task) -- own expiry, own namespace
   +- context Y (org 2 task) -- own expiry, own namespace
 per-context revocation | context migration -> revalidate
 shared-platform compromise -> all contexts untrusted until
 re-established]]></artwork>
</figure>
<t>A persistent evaluator does not inherit authority across projects. If the platform is compromised, every context is treated as untrusted until re-established. Moving a context between hosts requires revalidation.</t>
</section>
<section anchor="life">
<name>Six Independent Lifecycles</name>
<t>Six lifecycles are independent. None is assumed to nest inside another.</t>
<dl newline="true">
<dt>L1 Infrastructure</dt>
<dd><t>Host or platform exists. State: provisioned, active, suspended, decommissioned.</t></dd>
<dt>L2 Runtime instance</dt>
<dd><t>A running evaluator process. State: created, running, retired.</t></dd>
<dt>L3 Appointment</dt>
<dd><t>Designation for a task. State: proposed, owner-approved, active, expired, revoked.</t></dd>
<dt>L4 Evaluation round</dt>
<dd><t>Bounded review over a fixed evidence set. State: open, admitting, closed, disputed.</t></dd>
<dt>L5 Determination validity</dt>
<dd><t>Interval during which a determination is usable as a premise. State: valid, stale, withdrawn. It can end before the appointment ends.</t></dd>
<dt>L6 Evidence retention</dt>
<dd><t>How long evidence and audit material are kept. It may outlive all other lifecycles and grants no permission.</t></dd>
</dl>
<figure anchor="fig-life">
<name>Six independent lifetimes (not nested)</name>
<artwork type="ascii-art"><![CDATA[
 time ------------------------------------------------------>
 infrastructure  |=================================...........
 runtime instance|    |=========|     (retired early)
 appointment     |      |================|
 eval round      |        |====|
 determination   |           |========|  (may go stale first)
 evidence retain |           |==========================...
 Round may close while task is active; evidence outlives all
 permissions; a valid signature does not extend validity.]]></artwork>
</figure>
<t>Formal predicates (informal notation):</t>
<ul>
<li><tt>MandateValid(e,t,now) := Appointed(e,t) AND NOT Revoked(e,t) AND now &lt; Expiry(e,t) AND Uses(e,t) &lt; Limit(e,t)</tt></li>
<li><tt>RoundOpen(r,now) := State(r)=open AND now in Window(r)</tt></li>
<li><tt>DetValid(d,now) := Bound(d,task,round,policyGen) AND now in Validity(d) AND NOT Withdrawn(d) AND SnapshotValid(d)</tt></li>
<li><tt>EvidenceAdmissible(x,r) := L(x)=ok for all ten checks in <xref target="evidence"/></tt></li>
</ul>
<t>Negative tests: (a) a retired instance presents a valid-looking signature (rejected: L2); (b) evidence arrives after round closure (rejected: L4); (c) a determination is reused after policy generation changes (stale: L5); (d) evidence retention continues while authority is gone (permitted, grants nothing: L6); (e) appointment expires while the host continues (no authority for that task: L3).</t>
</section>
<section anchor="roles">
<name>Control Roles and Separation</name>
<t>Four control roles are distinguished. A single host may run several roles; co-location inherits no authority.</t>
<table anchor="tbl-roles">
<name>Role-separation matrix</name>
<thead><tr><th>Role</th><th>May do</th><th>May not do</th></tr></thead>
<tbody>
<tr><td>Evaluator</td><td>assess evidence; produce typed determination</td><td>admit an effect; issue a passport; self-appoint</td></tr>
<tr><td>Supervisor / coordinator</td><td>organize agents and tasks within a bounded lease</td><td>grant native authority; act as evaluator for its own plan</td></tr>
<tr><td>Authorization issuer / owner authority</td><td>issue bounded permission per owner policy</td><td>bypass the native gate; reuse consumed authority</td></tr>
<tr><td>Native gate / Finality Sink</td><td>admit or withhold the actual act</td><td>accept a determination or approval as sufficient</td></tr>
</tbody>
</table>
<figure anchor="fig-own">
<name>Owner and appointment resolution</name>
<artwork type="ascii-art"><![CDATA[
 request -> resolve actual mandatory owners
              |                 |
        ambiguous?          resolved
            YES: HOLD            |
                       owner consents (all mandatory)
                          |            |
                    missing/forged   complete
                    -> REJECT        -> appointment issued
                                     -> scope, expiry, use count
 Agent self-declaration of neutrality is never an input.]]></artwork>
</figure>
</section>
<section anchor="appoint">
<name>Protected Appointment and Eligibility</name>
<t>A protected appointment is task-specific and derived from owner authority. A candidate agent's self-declaration is never an input. The appointment record is expected to include the following elements:</t>
<ol>
<li>identity of the evaluator and its software or human identity</li>
<li>task or operation scope</li>
<li>resource owners and their consent</li>
<li>mandatory owners and vetoes</li>
<li>qualification and independence assertions with the control-group basis</li>
<li>policy generation</li>
<li>evidence classes the evaluator may access</li>
<li>permitted determination types</li>
<li>review-round limits and reviewer requirements</li>
<li>validity interval for determinations</li>
<li>appointment expiry</li>
<li>use-count limit</li>
<li>revocation reference and serial</li>
<li>host or runtime binding</li>
<li>dissent and quorum rules (owner-specific)</li>
<li>retention and disclosure constraints</li>
<li>jurisdiction and purpose constraints</li>
<li>successor or substitution rules</li>
</ol>
<t>Failure to verify any mandatory element results in refusal or HOLD, never in permissive default.</t>
</section>
<section anchor="evidence">
<name>Evidence Admission, Provenance and Outcome Truth</name>
<t>Ten distinctions apply to every evidence item:</t>
<ol>
<li>authenticated source origin</li>
<li>authorized source role</li>
<li>content integrity</li>
<li>task and operation correspondence</li>
<li>freshness</li>
<li>applicable policy generation</li>
<li>source independence</li>
<li>correlation among observations</li>
<li>actual native-resource outcome</li>
<li>unresolved or contradictory evidence</li>
</ol>
<ul>
<li>A signature proves origin and integrity, not outcome truth.</li>
<li>Independence is counted by underlying control group, not by account or key count.</li>
<li>Agent claims, tool or transport acknowledgments, coordinator flags and evaluator reports are weaker than resource-origin receipts or protected direct observation.</li>
<li>Evidence content is data, never instruction to the evaluator.</li>
<li>Origin proof or inclusion proof is not proof of physical truth; a dishonest resource owner is a stated residual limit.</li>
</ul>
<figure anchor="fig-evid">
<name>Evidence provenance and source-truth separation</name>
<artwork type="ascii-art"><![CDATA[
 source -> origin authenticated? -> role eligible? -> content
 integrity? -> bound to task/operation? -> fresh? -> policy
 generation? -> independent (control group)? -> correlated?
      each NO -> excluded or flagged
 admitted evidence  !=  outcome truth
 outcome truth needs resource-origin evidence or protected
 direct observation; agent success claims, tool acks, transport
 acks, coordinator flags and evaluator reports are weaker.]]></artwork>
</figure>
</section>
<section anchor="dec">
<name>Typed Non-Executable Determinations and Review Rounds</name>
<t>Determinations are typed and non-executable.</t>
<table anchor="tbl-dec">
<name>Typed determinations</name>
<thead><tr><th>Type</th><th>Meaning</th><th>Executable?</th></tr></thead>
<tbody>
<tr><td>APPROVED_FOR_CONSIDERATION</td><td>premise for owner authorization process</td><td>No</td></tr>
<tr><td>APPROVED</td><td>evaluator finds conditions met under mandate</td><td>No; premise only</td></tr>
<tr><td>DENIED</td><td>conditions not met</td><td>n/a</td></tr>
<tr><td>HOLD</td><td>incomplete mandate, evidence or review</td><td>No</td></tr>
<tr><td>DISPUTED</td><td>reviewers or sources contest</td><td>No</td></tr>
<tr><td>UNKNOWN</td><td>outcome or evidence unresolved</td><td>No; not DENIED, not SUCCESS</td></tr>
<tr><td>REQUEST_EVIDENCE</td><td>named additional evidence needed</td><td>No</td></tr>
<tr><td>REVISE</td><td>plan change suggested</td><td>No; needs recommitment</td></tr>
<tr><td>RENEGOTIATE</td><td>parties must renegotiate</td><td>No</td></tr>
<tr><td>HUMAN_REVIEW</td><td>escalate to human</td><td>No</td></tr>
<tr><td>SUBSTITUTE</td><td>replace an agent</td><td>No; needs fresh authorization</td></tr>
<tr><td>ABORT</td><td>stop the task</td><td>Not a native effect by itself</td></tr>
</tbody>
</table>
<t>UNKNOWN is not DENIED and not SUCCESS. Dissent is preserved with the determination; it is never silently dropped.</t>
<figure anchor="fig-indep">
<name>Reviewer independence, control groups, dissent</name>
<artwork type="ascii-art"><![CDATA[
 reviewers r1..rn -> map credentials -> control groups g1..gk
 diversity rule met on k (not n)?  NO -> HOLD
 mandatory reviewer present?       NO -> HOLD
 dissent recorded and shown?       NO -> reject round
 contested outcome -> DISPUTED (not a majority vote)
 round closed -> determination eligible as a premise only]]></artwork>
</figure>
</section>
<section anchor="xwalk">
<name>Embodiment Overview and Claim Crosswalk</name>
<t>The following table is a crosswalk of the 27 embodiments to the claims of the source filing. Claim numbers are those of the filing as supplied and should be verified against the official record.</t>
<table anchor="tbl-xwalk">
<name>E01-E27 crosswalk</name>
<thead><tr><th>Emb.</th><th>Code</th><th>Indep. claims</th><th>Dep. claims</th><th>Title (source)</th></tr></thead>
<tbody>
<tr><td>E01</td><td>CE-NJC-01</td><td>1-2</td><td>61-76</td><td>Protected Multi-Agent Negotiation, Joint-Plan Commitment, and Independent Evaluator-Governed Continuation</td></tr>
<tr><td>E02</td><td>CE-BFQ-02</td><td>3-5</td><td>77-100</td><td>Bilateral Communication Freeze, Distributed Quiescence, and Protected Evaluation-State Convergence</td></tr>
<tr><td>E03</td><td>CE-NSI-01</td><td>6-7</td><td>101-116</td><td>Negotiation-State Integrity, Causal Message Containment, and Continuous Evaluation-Snapshot Validity</td></tr>
<tr><td>E04</td><td>CE-INE-01</td><td>8-9</td><td>117-132</td><td>Independent Multi-Agent Consequence Evaluator, Protected Evidence Admission, and Next-Actor Selection</td></tr>
<tr><td>E05</td><td>CE-VCR-01</td><td>10</td><td>133-140</td><td>Isolated Virtual Consequence Reconstruction and Non-Effective Scenario Execution</td></tr>
<tr><td>E06</td><td>CE-RBL-01</td><td>11-12</td><td>141-156</td><td>Risk, Benefit-Loss and Tail-Exposure Evaluation With Independent Domain Thresholds</td></tr>
<tr><td>E07</td><td>CE-NAS-01</td><td>13-14</td><td>157-172</td><td>Evaluator-Directed Next-Actor Selection and Controlled Next-Turn Instructions</td></tr>
<tr><td>E08</td><td>CE-ACR-01</td><td>15-16</td><td>173-188</td><td>Corrective Instructions, Protected Plan Revision, and Controlled Multiparty Renegotiation</td></tr>
<tr><td>E09</td><td>CE-ARR-01</td><td>17-18</td><td>189-204</td><td>Agent Role Reconfiguration, Bounded Delegation, and Protected Capability Transition</td></tr>
<tr><td>E10</td><td>CE-SCR-01</td><td>19-20</td><td>205-220</td><td>Selective Communication Resumption, Single-Turn Authority, Automatic Re-Freezing, and Inter-Phase Hold</td></tr>
<tr><td>E11</td><td>CE-NMP-01</td><td>21-22</td><td>221-236</td><td>Protected Next-Move Passport, Actor-Bound Issuance, Single-Use Finality Verification, and Outcome-Uncertainty Reconciliation</td></tr>
<tr><td>E12</td><td>CE-PFS-01</td><td>23-24</td><td>237-252</td><td>Protected Finality Sink Enforcement Before Communication, Privileged Execution, External Dispatch and Physical Actuation</td></tr>
<tr><td>E13</td><td>CE-NSA-01</td><td>25-26</td><td>253-268</td><td>Risk-Triggered Neutral Substitute-Agent Selection, Appointment, and Protected Reauthorization</td></tr>
<tr><td>E14</td><td>CE-PWM-01</td><td>27-28</td><td>269-284</td><td>Protected Pending-Work Escrow, Task-State Migration, and Substitute Reauthorization</td></tr>
<tr><td>E15</td><td>CE-PCR-01</td><td>29-31</td><td>285-308</td><td>Phased Execution, Protected Completion Receipts, and Verified Phase Advancement</td></tr>
<tr><td>E16</td><td>CE-VFO-01</td><td>32-33</td><td>309-324</td><td>Verified Feedback, Actual-Outcome Evaluation, and Protected Continuation</td></tr>
<tr><td>E17</td><td>CE-EAC-01</td><td>34-35</td><td>325-340</td><td>Evaluator Integrity, Source Provenance, Collusion-Resistant Review, and Protected Continuation</td></tr>
<tr><td>E18</td><td>CE-PAR-01</td><td>36-38</td><td>341-364</td><td>Cross-Domain Network Partitions, Unknown Outcomes, Durable Fencing, Migration and Safe Recovery</td></tr>
<tr><td>E19</td><td>CE-SCL-01</td><td>39-40</td><td>365-380</td><td>Protected Session Closure, Obligation Reconciliation, Native Domain Sealing, and Irreversible Authority Retirement</td></tr>
<tr><td>E20</td><td>CE-JSA-01</td><td>41-42</td><td>381-396</td><td>Verified Mandatory Approvers, Authenticated Hosting and Installation, and Revocable Supervisor Identity</td></tr>
<tr><td>E21</td><td>CE-IEO-01</td><td>43-44</td><td>397-412</td><td>Independent Cross-Domain Evaluator Appointment, Actual-Outcome Verification, Reviewer Accountability and Dispute Containment</td></tr>
<tr><td>E22</td><td>CE-CTP-01</td><td>45-46</td><td>413-428</td><td>Cross-Domain Trust and Policy Compatibility</td></tr>
<tr><td>E23</td><td>CE-LDC-01</td><td>47-48</td><td>429-444</td><td>Leaderless Cross-Domain Agent Coordination</td></tr>
<tr><td>E24</td><td>CE-SER-01</td><td>49-50</td><td>445-460</td><td>Self-Expiring Supervisor and Evaluator Authority, Authenticated Revocation, and Generation-Fenced Replacement</td></tr>
<tr><td>E25</td><td>CE-DMG-01</td><td>51-52</td><td>461-476</td><td>Dynamic Multi-Organization Membership and Principal-Independent Threshold Governance</td></tr>
<tr><td>E26</td><td>CE-PCE-01</td><td>53-55</td><td>477-500</td><td>Purpose-Bound Cross-Organizational Outcome Evidence</td></tr>
<tr><td>E27</td><td>CE-MEF-01</td><td>56-60</td><td>501-540</td><td>Multi-Domain Protected Execution Commitment, Partial-Outcome Reconciliation and Cross-Organizational Recovery</td></tr>
</tbody>
</table>
</section>
<section anchor="e1">
<name>Embodiments E01-E16</name>
<t>Embodiments E01-E16 establish the multi-agent evaluation handoff. Each is summarized with the required profile fields. E12 is the embodiment that restates the foundational Finality Sink in the evaluator setting; this document does not claim a new execution-finality protocol.</t>
<figure anchor="fig-joint">
<name>Joint plan commitment and non-effective handoff</name>
<artwork type="ascii-art"><![CDATA[
 Agent A <--proposals/counters--> Agent B
        \                         /
         candidate plan v1 (bound to accepted state)
                  |
     required principals commit? --NO--> keep negotiating
                  | YES
        plan = committed but NON-EFFECTIVE
                  |
        freeze + pending-work reconciliation OK?
           NO -> HOLD       YES -> evaluator handoff
 Resource effects still need separate owner authorization.]]></artwork>
</figure>
<figure anchor="fig-freeze">
<name>Freeze, pending-message reconciliation, quiescence</name>
<artwork type="ascii-art"><![CDATA[
 freeze request -> each edge acknowledges?
   partial ack -> NOT frozen (unresolved)
   all ack -> inventory in-flight, callbacks, queues, tools
        -> late message arrives?  YES -> classify, quarantine
        -> all accounted?  YES -> quiescent snapshot
                           NO  -> UNRESOLVED (no evaluation)
 Applies only to interfaces actually under enforcement.]]></artwork>
</figure>
<figure anchor="fig-lineage">
<name>Causal lineage and snapshot invalidation</name>
<artwork type="ascii-art"><![CDATA[
 msg m1 -> m2 -> m3 (per-sender sequence, causal links)
 snapshot S bound to plan v1 and evidence set E1
 material change (new message / evidence / policy gen)
     -> S invalidated -> determination on S becomes stale
     -> re-snapshot -> re-evaluate
 superseded or unaccounted message cannot alter plan silently]]></artwork>
</figure>
<figure anchor="fig-sim">
<name>Virtual consequence simulation and risk classification</name>
<artwork type="ascii-art"><![CDATA[
 verified plan + authentic observations
        -> isolated non-effective simulator (no egress)
        -> scenarios + uncertainty
        -> per-domain risk, tail exposure, thresholds
 owner veto? YES -> blocked    unknown fidelity -> qualified
 result = evidence only; never a commit or proof of behavior]]></artwork>
</figure>
<figure anchor="fig-next">
<name>Next actor, passport and automatic refreeze</name>
<artwork type="ascii-art"><![CDATA[
 evaluator -> select actor (eligible?) -> bounded instruction
 owner approves -> passport (actor, task, op, plan, phase,
                  epoch, policy, single use)
 gate checks passport:  wrong actor / replay / expired -> REJECT
 one turn admitted -> turn ends -> automatic refreeze
 unknown delivery -> reconcile (no blind retry)]]></artwork>
</figure>
<figure anchor="fig-subst">
<name>Substitute selection and pending-work escrow</name>
<artwork type="ascii-art"><![CDATA[
 risk trigger -> screen candidates (neutral, qualified, no
 conflict) -> de-appoint original -> appoint substitute
 escrow unfinished work (canonical, minimal) -> fence source
 -> destination accepts -> fresh authorization
 old credentials and consumed authority are NOT inherited]]></artwork>
</figure>
<figure anchor="fig-phase">
<name>Phase receipt verification and feedback continuation</name>
<artwork type="ascii-art"><![CDATA[
 phase N done? -> resource-origin receipt verified?
    NO -> block advancement (no skipping)
    YES -> feedback evaluated (untrusted instructions contained)
        -> outcome classified: success / partial / UNKNOWN
        -> proposed next step -> separate authorization
 coordinator "done" and agent self-report are not receipts]]></artwork>
</figure>
<section anchor="e01">
<name>E01: Protected Multi-Agent Negotiation, Joint-Plan Commitment, and Independent Evaluator-Governed Continuation</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E01 (CE-NJC-01), source page 39. Protected Multi-Agent Negotiation, Joint-Plan Commitment, and Independent Evaluator-Governed Continuation</t></dd>
<dt>Problem</dt>
<dd><t>Independently governed agents agree a joint plan, but an agreement is not an authorization to act; continuation must pass an independent evaluator after a verified freeze.</t></dd>
<dt>Actors</dt>
<dd><t>Two or more agents under different administrative domains; mandatory authority principals; an independent neutral evaluator.</t></dd>
<dt>Protected state</dt>
<dd><t>Versioned candidate plan bound to an accepted communication state and required principals; commitment records; freeze and pending-work records.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: proposals, counteroffers, constraints, approvals. Out: a committed non-effective candidate plan and an evaluator-governed handoff.</t></dd>
<dt>State transitions</dt>
<dd><t>negotiating, committed plan, frozen, evaluating, continuation proposed.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>No commitment without required principals; no effect from a jointly signed plan; handoff withheld until freeze and reconciliation succeed.</t></dd>
<dt>Threats</dt>
<dd><t>Forged consent; missing mandatory owner; plan altered after commitment; one agent's self-declared authority.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>A jointly signed plan stays non-effective with respect to real resource operations; effects are admitted only at actual enforcement points.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 16 and 33 test runs, and 42 tests passed on modeled code paths (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>First embodiment; basis for E02 to E16.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>joint plan commitment; multi-agent negotiation; evaluator handoff; non-effective plan</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 1, 2. Claim 1: Protected Multi-Agent Agreement and Independently Controlled Evaluation Handoff; Claim 2: Protected Multi-Agent Negotiation and Evaluator-Governed Continuation Method Dependent claims 61-76 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e02">
<name>E02: Bilateral Communication Freeze, Distributed Quiescence, and Protected Evaluation-State Convergence</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E02 (CE-BFQ-02), source page 173. Bilateral Communication Freeze, Distributed Quiescence, and Protected Evaluation-State Convergence</t></dd>
<dt>Problem</dt>
<dd><t>A declared stop is not a quiescent snapshot: in-flight messages, delayed callbacks, deferred tool calls and queue deliveries can still change state.</t></dd>
<dt>Actors</dt>
<dd><t>Two or more agents; freeze controller; independently administered enforcement interfaces.</t></dd>
<dt>Protected state</dt>
<dd><t>Freeze epoch, per-edge acknowledgments, pending-work inventory, late-operation disposition records.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: commitment to one plan version. Out: verified freeze with accounted pending work, or unresolved status.</t></dd>
<dt>State transitions</dt>
<dd><t>freeze requested, partial acknowledgment, quiescent, unresolved.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Partial freeze acknowledgment is not a freeze; late or duplicated messages are not silently admitted; unresolved delivery blocks evaluation.</t></dd>
<dt>Threats</dt>
<dd><t>In-flight message after freeze; unfenced background work; different state generations; replayed deliveries.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Actual covered communication interfaces and queues; the restriction is asserted only for interfaces actually subject to enforcement.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 28 and 17 test runs (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E01 commitment.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>communication freeze; distributed quiescence; barrier; message fencing</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 3, 4, 5. Claim 3: Independently Enforced Cross-Domain Communication Freeze and Distributed Quiescence System; Claim 4: Atomic Fencing, Pending-Work Reconciliation, and Verified Freeze Method; Claim 5: Protected Admission Fencing and Late-Operation Disposition Dependent claims 77-100 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e03">
<name>E03: Negotiation-State Integrity, Causal Message Containment, and Continuous Evaluation-Snapshot Validity</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E03 (CE-NSI-01), source page 309. Negotiation-State Integrity, Causal Message Containment, and Continuous Evaluation-Snapshot Validity</t></dd>
<dt>Problem</dt>
<dd><t>A later message can silently change the plan an evaluator is reviewing unless message causality and snapshot validity are tracked.</t></dd>
<dt>Actors</dt>
<dd><t>Agents; evaluator; state-integrity controller.</t></dd>
<dt>Protected state</dt>
<dd><t>Causal event records (per-sender sequencing), snapshot identifiers, invalidation records, version-bound plan and evidence references.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: messages and accepted task states. Out: a stable verifiable reference to the exact plan and interaction state, or invalidation.</t></dd>
<dt>State transitions</dt>
<dd><t>snapshot valid, material change observed, snapshot invalidated.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Superseded or unaccounted communications cannot silently alter the evaluated plan; invalid snapshot cannot support a determination.</t></dd>
<dt>Threats</dt>
<dd><t>Out-of-order or replayed messages; snapshot reuse after material change.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Protected state controller for plan and evidence references; does not itself admit external effects.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: no test-run output located by the editor in the pages examined. These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E01, E02.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>causal containment; snapshot validity; negotiation-state integrity</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 6, 7. Claim 6: Protected Multi-Agent Causal-State and Evaluation-Snapshot Integrity System; Claim 7: Causal Message Containment and Continuous Snapshot-Validity Method Dependent claims 101-116 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e04">
<name>E04: Independent Multi-Agent Consequence Evaluator, Protected Evidence Admission, and Next-Actor Selection</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E04 (CE-INE-01), source page 375. Independent Multi-Agent Consequence Evaluator, Protected Evidence Admission, and Next-Actor Selection</t></dd>
<dt>Problem</dt>
<dd><t>The evaluator must receive exactly the plan and evidence state without gaining the resource credentials needed to effectuate anything.</t></dd>
<dt>Actors</dt>
<dd><t>Independent evaluator; evidence sources; agents; organizational policy owners.</t></dd>
<dt>Protected state</dt>
<dd><t>Admitted evidence set with source roles; evaluator appointment reference; typed recommendation record.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: frozen plan, authenticated evidence. Out: a non-effective decision and a risk-constrained next-actor or action-class recommendation.</t></dd>
<dt>State transitions</dt>
<dd><t>evidence admitted, evaluated, recommendation issued.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Evaluator recommendation is not permission; inadmissible evidence is excluded; missing evidence yields a hold or request, not approval.</t></dd>
<dt>Threats</dt>
<dd><t>Evaluator evidence-access escalation; malicious instructions in evidence; evaluator self-appointment.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>The evaluator is separated from the effectuation credentials; permission is handled elsewhere.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 50 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E01 to E03.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>independent consequence evaluator; evidence admission; non-effective decision</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 8, 9. Claim 8: Independent Multi-Agent Consequence Evaluation and Non-Effective Decision System; Claim 9: Source-Aware Evaluation, Risk-Constrained Actor Selection, and Protected Decision-Handoff Method Dependent claims 117-132 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e05">
<name>E05: Isolated Virtual Consequence Reconstruction and Non-Effective Scenario Execution</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E05 (CE-VCR-01), source page 422. Isolated Virtual Consequence Reconstruction and Non-Effective Scenario Execution</t></dd>
<dt>Problem</dt>
<dd><t>A simulation can mislead: it is neither a real commit nor proof that destination infrastructure behaves as predicted.</t></dd>
<dt>Actors</dt>
<dd><t>Evaluator; isolated simulation environment; resource-state observers.</t></dd>
<dt>Protected state</dt>
<dd><t>Virtual state reconstruction, scenario records, uncertainty annotations, phase-bound decision evidence.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: verified plan, authentic observations. Out: predicted consequences with stated uncertainty.</t></dd>
<dt>State transitions</dt>
<dd><t>reconstructed, scenario run, evidence recorded.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Simulation result is never treated as proof of actual execution; model uncertainty and fidelity limits are preserved.</t></dd>
<dt>Threats</dt>
<dd><t>False simulation-as-completion assertion; data egress from the simulator.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Isolated non-effective environment with guarded egress; no resource effect.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 62 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E04.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>virtual consequence reconstruction; non-effective simulation; digital twin</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 10. Claim 10: Non-Effective Consequence Evaluation of Multi-Agent Candidate Operations Dependent claims 133-140 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e06">
<name>E06: Risk, Benefit-Loss and Tail-Exposure Evaluation With Independent Domain Thresholds</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E06 (CE-RBL-01), source page 486. Risk, Benefit-Loss and Tail-Exposure Evaluation With Independent Domain Thresholds</t></dd>
<dt>Problem</dt>
<dd><t>One organization's acceptable risk score must not override another owner's mandatory restriction, and no universal scalar risk exists.</t></dd>
<dt>Actors</dt>
<dd><t>Evaluator; each domain owner with its own thresholds; agents.</t></dd>
<dt>Protected state</dt>
<dd><t>Per-domain thresholds, loss and tail-exposure measures, veto records, calibration records.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: consequence evidence and policies. Out: risk classification per domain and conditioned continuation recommendation.</t></dd>
<dt>State transitions</dt>
<dd><t>measured, thresholds applied, veto or pass.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>A domain veto is not outvoted; missing calibration limits conclusions; no invented universal policy.</t></dd>
<dt>Threats</dt>
<dd><t>Threshold manipulation; miscalibrated risk; cumulative exposure ignored.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Each owner's own enforcement boundary applies its own threshold.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 69 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E04, E05.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>tail exposure; risk benefit loss; domain threshold; owner veto</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 11, 12. Claim 11: Owner-Specific Consequence Bound Enforcement for Cooperative Operations; Claim 12: Calibration-Bounded Risk Reassessment With Adverse-Outcome Feedback Dependent claims 141-156 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e07">
<name>E07: Evaluator-Directed Next-Actor Selection and Controlled Next-Turn Instructions</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E07 (CE-NAS-01), source page 538. Evaluator-Directed Next-Actor Selection and Controlled Next-Turn Instructions</t></dd>
<dt>Problem</dt>
<dd><t>Selecting an actor is not granting its credentials; routing needs eligibility, restrictions and a separately enforced continuation.</t></dd>
<dt>Actors</dt>
<dd><t>Evaluator; candidate next actors; owners.</t></dd>
<dt>Protected state</dt>
<dd><t>Actor eligibility records, bounded next-turn instruction, task-type class.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: evaluation and eligibility. Out: selected next actor and bounded instruction.</t></dd>
<dt>State transitions</dt>
<dd><t>candidates screened, actor selected, instruction bounded.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Selection of an agent does not confer resource credentials; conflicting concurrent moves are arbitrated, not both admitted.</t></dd>
<dt>Threats</dt>
<dd><t>Next-actor impersonation; conflicting concurrent next moves.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Continuation enforced separately at the actor's resource or communication gate.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 62 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E04 to E06.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>next actor selection; bounded instruction; arbitration of concurrent moves</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 13, 14. Claim 13: Protected Eligibility-Based Next-Actor Allocation; Claim 14: Conflict-Aware Arbitration of Concurrent Agent Next Moves Dependent claims 157-172 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e08">
<name>E08: Corrective Instructions, Protected Plan Revision, and Controlled Multiparty Renegotiation</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E08 (CE-ACR-01), source page 582. Corrective Instructions, Protected Plan Revision, and Controlled Multiparty Renegotiation</t></dd>
<dt>Problem</dt>
<dd><t>A corrective instruction must not rewrite an already protected joint plan without an authorized transition.</t></dd>
<dt>Actors</dt>
<dd><t>Evaluator; agents; owners who must re-commit.</t></dd>
<dt>Protected state</dt>
<dd><t>Patch records scoped to fields, version invalidation, fresh recommitment records.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: evaluation finding. Out: authorized plan revision after fresh multi-owner recommitment, or hold.</t></dd>
<dt>State transitions</dt>
<dd><t>correction proposed, versions invalidated, renegotiated, recommitted.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>A model-generated proposal does not overwrite the committed plan; revision requires new consents.</t></dd>
<dt>Threats</dt>
<dd><t>Unauthorized renegotiation; stale commitments surviving revision.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Re-entry into a valid evaluation cycle with the same owner-side enforcement.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 40 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E01, E04, E07.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>corrective patch; plan revision; renegotiation; recommitment</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 15, 16. Claim 15: Field-Scoped Corrective Patch Admission for Committed Plans; Claim 16: Causal Renegotiation and Fresh Multi-Owner Recommitment Dependent claims 173-188 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e09">
<name>E09: Agent Role Reconfiguration, Bounded Delegation, and Protected Capability Transition</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E09 (CE-ARR-01), source page 624. Agent Role Reconfiguration, Bounded Delegation, and Protected Capability Transition</t></dd>
<dt>Problem</dt>
<dd><t>Changing an agent's role or tools is a capability change that must be separate from authority for an external effect.</t></dd>
<dt>Actors</dt>
<dd><t>Evaluator; owners; agents; capability enforcement controller.</t></dd>
<dt>Protected state</dt>
<dd><t>Role and capability generations, delegation records, outstanding-work and credential fencing records.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: evaluator recommendation, owner approval. Out: versioned capability transition.</t></dd>
<dt>State transitions</dt>
<dd><t>role proposed, owner-approved, transition enforced.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Old credentials, retained workers and cached role assignments are fenced; role change does not authorize effects.</t></dd>
<dt>Threats</dt>
<dd><t>Credential retention; stale role cache; unauthorized delegation.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Protected versioned capability-enforcement point.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 59 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E07, E08.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>role reconfiguration; bounded delegation; capability generation</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 17, 18. Claim 17: Owner-Governed Dynamic Agent Role Reconfiguration; Claim 18: Safe Strategy Replacement With Outstanding-Work and Credential Fencing Dependent claims 189-204 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e10">
<name>E10: Selective Communication Resumption, Single-Turn Authority, Automatic Re-Freezing, and Inter-Phase Hold</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E10 (CE-SCR-01), source page 670. Selective Communication Resumption, Single-Turn Authority, Automatic Re-Freezing, and Inter-Phase Hold</t></dd>
<dt>Problem</dt>
<dd><t>Allowing one needed message should not reopen unrestricted communication.</t></dd>
<dt>Actors</dt>
<dd><t>Evaluator; agents; communication gates.</t></dd>
<dt>Protected state</dt>
<dd><t>Single-turn authority record, communication-edge permission, refreeze state.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: evaluator need for a specific communication. Out: one permitted message then automatic return to hold.</t></dd>
<dt>State transitions</dt>
<dd><t>held, one turn admitted, refrozen.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Turn termination returns to a protected evaluation state; unknown delivery is handled, not assumed.</t></dd>
<dt>Threats</dt>
<dd><t>Race between reopening and refreezing; reuse of the turn authority.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Ingress and egress admission at the actual communication boundary.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 53 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E02, E07.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>single-turn authority; selective resumption; automatic refreezing; inter-phase hold</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 19, 20. Claim 19: One-Message Cross-Domain Communication Reopening; Claim 20: Race-Safe Re-Freezing and Unknown-Delivery Handling Dependent claims 205-220 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e11">
<name>E11: Protected Next-Move Passport, Actor-Bound Issuance, Single-Use Finality Verification, and Outcome-Uncertainty Reconciliation</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E11 (CE-NMP-01), source page 719. Protected Next-Move Passport, Actor-Bound Issuance, Single-Use Finality Verification, and Outcome-Uncertainty Reconciliation</t></dd>
<dt>Problem</dt>
<dd><t>A next-move authorization must not become a bearer token that any holder can reuse.</t></dd>
<dt>Actors</dt>
<dd><t>Issuer, selected actor, owners, consuming gate.</t></dd>
<dt>Protected state</dt>
<dd><t>Passport (or protected-local slot) bound to actor, task, operation, plan version, phase, freeze epoch, policy, resource state and single use; consumption state.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: evaluator selection and owner approval. Out: a single-use bound authorization and consumption record.</t></dd>
<dt>State transitions</dt>
<dd><t>issued, presented, consumed, expired or revoked, unknown.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Replay, wrong actor, wrong operation or consumed passport are rejected; unknown outcome goes to reconciliation, not retry.</t></dd>
<dt>Threats</dt>
<dd><t>Passport replay; next-actor impersonation; attenuation errors.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Verified at the consuming gate before the act proceeds.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 62 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E07, E10.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>next-move passport; single-use authorization; non-bearer; consumption state</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 21, 22. Claim 21: Protected Actor- and Operation-Bound Continuation Authority; Claim 22: Protected Next-Move Grant Attenuation, Revocation and Consumption State Dependent claims 221-236 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e12">
<name>E12: Protected Finality Sink Enforcement Before Communication, Privileged Execution, External Dispatch and Physical Actuation</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E12 (CE-PFS-01), source page 768. Protected Finality Sink Enforcement Before Communication, Privileged Execution, External Dispatch and Physical Actuation</t></dd>
<dt>Problem</dt>
<dd><t>A valid evaluation and authorization are not the same as admission of the actual pending act at the resource.</t></dd>
<dt>Actors</dt>
<dd><t>Resource owner; Finality Sink; agents.</t></dd>
<dt>Protected state</dt>
<dd><t>Reconstructed pending act, authorization reference, alternate-path closure records, external-dispatch outbox.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: authorization and pending act. Out: admit exactly the authorized act or withhold.</t></dd>
<dt>State transitions</dt>
<dd><t>act reconstructed, compared, admitted or withheld.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Non-matching act is withheld; model evaluation does not bypass owner admission.</t></dd>
<dt>Threats</dt>
<dd><t>Alternate unmonitored paths; act altered after authorization.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>The participating effectuation boundary before communication release, data export, privileged execution, remote dispatch, payment, network configuration or actuation.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 53 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Applies the foundational Finality Sink to the evaluator setting.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>finality sink; pending act reconstruction; point of effect; outbox</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 23, 24. Claim 23: Native Pending-Act Reconstruction and Exact-State Authorization; Claim 24: Alternate-Path Closure and Protected External-Dispatch Outbox Dependent claims 237-252 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e13">
<name>E13: Risk-Triggered Neutral Substitute-Agent Selection, Appointment, and Protected Reauthorization</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E13 (CE-NSA-01), source page 815. Risk-Triggered Neutral Substitute-Agent Selection, Appointment, and Protected Reauthorization</t></dd>
<dt>Problem</dt>
<dd><t>When risk triggers replacement of an agent, the substitute must be neutral, qualified and separately authorized, not an inheritor of credentials.</t></dd>
<dt>Actors</dt>
<dd><t>Evaluator; original agents; substitute candidates; owners.</t></dd>
<dt>Protected state</dt>
<dd><t>Eligibility and conflict screening records, de-appointment record, substitute designation.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: risk trigger. Out: appointed substitute with fresh authorization.</t></dd>
<dt>State transitions</dt>
<dd><t>risk triggered, candidates screened, substitute appointed.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>No automatic inheritance of the replaced agent's credentials; conflicted candidates rejected.</t></dd>
<dt>Threats</dt>
<dd><t>Substitute credential inheritance; hidden conflict of interest.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Substitute acts only after reauthorization at resource gates.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 56 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E04, E09.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>substitute agent; neutrality; de-appointment; reauthorization</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 25, 26. Claim 25: Neutral Substitute-Agent Eligibility and Conflict Screening; Claim 26: Original-Agent De-Appointment and Protected Substitute Designation Dependent claims 253-268 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e14">
<name>E14: Protected Pending-Work Escrow, Task-State Migration, and Substitute Reauthorization</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E14 (CE-PWM-01), source page 863. Protected Pending-Work Escrow, Task-State Migration, and Substitute Reauthorization</t></dd>
<dt>Problem</dt>
<dd><t>Transferring task data must not transfer native effect permission or consumed authority.</t></dd>
<dt>Actors</dt>
<dd><t>Source agent or controller; destination; owners; evaluator.</t></dd>
<dt>Protected state</dt>
<dd><t>Canonical minimal-disclosure escrow of unfinished work, source fencing, destination acceptance records.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: unfinished work and chosen substitute. Out: committed escrow, destination acceptance, fresh authorization.</t></dd>
<dt>State transitions</dt>
<dd><t>escrowed, source fenced, destination accepts, reauthorized.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Unproven outcomes and consumed authorizations are not inherited; unaccepted escrow is not live.</t></dd>
<dt>Threats</dt>
<dd><t>Escrow substitution; source continuing to act after fence.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Source fencing and destination-side admission.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 61 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E13.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>pending-work escrow; task-state migration; source fencing</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 27, 28. Claim 27: Canonical Minimal-Disclosure Escrow of Unfinished Multi-Agent Work; Claim 28: Recoverable Cross-Domain Task Escrow Commit and Destination Acceptance Dependent claims 269-284 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e15">
<name>E15: Phased Execution, Protected Completion Receipts, and Verified Phase Advancement</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E15 (CE-PCR-01), source page 928. Phased Execution, Protected Completion Receipts, and Verified Phase Advancement</t></dd>
<dt>Problem</dt>
<dd><t>A coordinator's claim that a phase is done is not evidence; phases need resource-origin receipts and prerequisites.</t></dd>
<dt>Actors</dt>
<dd><t>Phase scheduler; resource owners issuing receipts; evaluator.</t></dd>
<dt>Protected state</dt>
<dd><t>Phase identities, source receipts, prerequisite records, advancement state.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: receipts. Out: verified phase closure and separately authorized advancement.</t></dd>
<dt>State transitions</dt>
<dd><t>phase open, receipt verified, closed, next phase authorized.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Missing receipt blocks advancement; phases cannot be skipped; unknown outcome is reconciled.</t></dd>
<dt>Threats</dt>
<dd><t>Missing phase receipt; forged completion evidence.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Resource-originated receipt verification before advancement.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 66 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E12, E14.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>phased execution; completion receipt; phase advancement</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 29, 30, 31. Claim 29: Resource-Originated Atomic Task Completion Evidence; Claim 30: External Effect Intent Correlation and Unknown-Outcome Reconciliation; Claim 31: Protected Evidence-Based Phase Closure and Separately Governed Advancement Dependent claims 285-308 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e16">
<name>E16: Verified Feedback, Actual-Outcome Evaluation, and Protected Continuation</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E16 (CE-VFO-01), source page 1005. Verified Feedback, Actual-Outcome Evaluation, and Protected Continuation</t></dd>
<dt>Problem</dt>
<dd><t>Agent self-reported success must not stand in for actual outcome evidence; feedback can also carry hostile instructions.</t></dd>
<dt>Actors</dt>
<dd><t>Evaluator; resource owners; feedback sources.</t></dd>
<dt>Protected state</dt>
<dd><t>Source-attributed feedback records, evidence-origin correlation, outcome classification.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: feedback and outcome evidence. Out: outcome classification supporting a proposed continuation.</t></dd>
<dt>State transitions</dt>
<dd><t>feedback received, origin correlated, outcome classified.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Untrusted feedback instructions are contained; unknown is not success; continuation needs fresh authorization.</t></dd>
<dt>Threats</dt>
<dd><t>Authenticated but false outcome reports; instruction injection in feedback.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Continuation is separately authorized at owner gates.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 61 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E15.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>verified feedback; actual outcome; outcome truth; feedback containment</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 32, 33. Claim 32: Independent Evidence-Origin Correlation and Actual-Outcome Classification; Claim 33: Untrusted Feedback Instruction Containment and Outcome-Truth Continuity Dependent claims 309-324 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
</section>
<section anchor="e2">
<name>Embodiments E17-E27</name>
<t>Embodiments E17-E27 are treated individually. E15 relates to phase receipts, E18 to partition recovery, and E27 to native partial outcomes; all three are read in relation to the existing execution-finality drafts and add no new protocol.</t>
<figure anchor="fig-part">
<name>Partition, UNKNOWN outcome and fencing</name>
<artwork type="ascii-art"><![CDATA[
 act sent -> response lost -> status unknown
 read-only source-authoritative status inquiry
   committed -> record;  not committed -> new authorization
   unreachable -> UNKNOWN (no automatic retry)
 old coordinator generation g1 < new g2 -> resource rejects g1
 recovery and retry are separately authorized]]></artwork>
</figure>
<figure anchor="fig-close">
<name>Session closure, domain seals and tombstones</name>
<artwork type="ascii-art"><![CDATA[
 close request -> inventory obligations in each mandatory
 domain, queues, delegated work, grants
 every domain sealed with source-authenticated evidence?
   NO -> closure withheld
   YES -> revoke/exhaust grants -> tombstone (no reopen)
 late completion evidence recorded; terminated authority is
 never revived. Coordinator DONE is not closure.]]></artwork>
</figure>
<figure anchor="fig-sup">
<name>Supervisor installation and revocation</name>
<artwork type="ascii-art"><![CDATA[
 owners consent -> joint appointment -> host bound to instance
 -> installed at each domain -> per-call authorization
 lease expires / owner withdraws -> revocation serial n+1
 -> local fence acknowledged
 disconnected domain: authority ends by lease bound, not by
 an assumed instant signal; successor needs new generation]]></artwork>
</figure>
<figure anchor="fig-trust2">
<name>Trust and policy compatibility across domains</name>
<artwork type="ascii-art"><![CDATA[
 domain A policy  domain B policy
       \             /
   authenticated trust formation
        |
   verified policy assertions -> schema + semantic check
   compatible? NO -> refuse (no optimistic merge)
   YES -> constraint intersection (conservative)
        -> bounded joint authority; non-delegable stay local
 drift (owner or policy generation) -> re-verify]]></artwork>
</figure>
<figure anchor="fig-leader">
<name>Leaderless coordination with owner-side predecessors</name>
<artwork type="ascii-art"><![CDATA[
 org X op1 --native receipt--> org Y checks predecessor
 org Y admits op2 only if receipt verifies at Y
 no common supervisor required; no global authority
 convergence evidence is not permission
 forged or missing predecessor -> refuse at owner side]]></artwork>
</figure>
<figure anchor="fig-member">
<name>Membership, owner voting and reconstitution</name>
<artwork type="ascii-art"><![CDATA[
 change proposed -> snapshot eligible electorate (epoch e)
 ballots (source-authenticated, one per organization)
 mandatory vetoes honored? joiner consents?
   NO -> blocked   YES -> stage epoch e+1
 old epoch fenced; unresolved effects settled
 local installation complete -> activate e+1
 resource-local authorization still required]]></artwork>
</figure>
<figure anchor="fig-disc">
<name>Purpose-bound selective disclosure and audit</name>
<artwork type="ascii-art"><![CDATA[
 evaluator request (purpose, audience) -> custodian checks
 purpose + audience match? NO -> refuse
 YES -> reveal attribute subset + commitment proof
 evidence lineage + disclosure audit (minimal retention)
 proof of origin or inclusion != proof of physical truth]]></artwork>
</figure>
<figure anchor="fig-pc">
<name>Multi-domain PREPARE, COMMIT and partial-effect reconciliation</name>
<artwork type="ascii-art"><![CDATA[
 domain 1: PREPARE ok   domain 2: PREPARE ok   domain 3: denied
 owner-specific authorization per domain -> COMMIT at gates
 results: committed / denied / UNKNOWN
 any domain not committed -> PARTIAL_EFFECT manifest
 recovery choices: inquire, retry (authorized), compensate
 no global atomic transaction is assumed]]></artwork>
</figure>
<figure anchor="fig-comp">
<name>Independently authorized compensation</name>
<artwork type="ascii-art"><![CDATA[
 partial effect recorded -> compensation proposed
 owner of affected resource consents?  NO -> refused
 YES -> compensation is a NEW native effect (own receipt)
 reversible? -> may reverse;  irreversible -> corrective act
 original committed record stays in history]]></artwork>
</figure>
<figure anchor="fig-comp2">
<name>Complete native-domain evidence verification</name>
<artwork type="ascii-art"><![CDATA[
 roster of mandatory domains
 for each: resource-origin receipt present and verified?
   any missing -> composite status INCOMPLETE
 all present -> composite completion record
 record is evidence only (not executable authority)]]></artwork>
</figure>
<section anchor="e17">
<name>E17: Evaluator Integrity, Source Provenance, Collusion-Resistant Review, and Protected Continuation</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E17 (CE-EAC-01), source page 1078. Evaluator Integrity, Source Provenance, Collusion-Resistant Review, and Protected Continuation</t></dd>
<dt>Problem</dt>
<dd><t>A signature or majority vote does not establish a reliable evaluation; independence and dissent need protected treatment.</t></dd>
<dt>Actors</dt>
<dd><t>Evaluator; reviewers; evidence sources; owners.</t></dd>
<dt>Protected state</dt>
<dd><t>Appointment, reviewer roster, control-group mapping, review-round record, dissent record.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: evidence and reviews. Out: closed round with preserved dissent, or fail-closed hold.</t></dd>
<dt>State transitions</dt>
<dd><t>round open, reviews admitted, completeness checked, closed.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Mandatory dissent is not suppressed; correlated sources are not counted as independent; failure of mandatory conditions fails closed.</t></dd>
<dt>Threats</dt>
<dd><t>Reviewer collusion; compromised evaluator; correlated observations.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Continuation depends on a complete, independent review and separate authorization.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 120 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E04, E16.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>evaluator integrity; control group; mandatory dissent; review round closure</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 34, 35. Claim 34: Evaluator Eligibility With Underlying Control-Group Independence; Claim 35: Closed Review-Round Completeness and Mandatory Dissent Preservation Dependent claims 325-340 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e18">
<name>E18: Cross-Domain Network Partitions, Unknown Outcomes, Durable Fencing, Migration and Safe Recovery</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E18 (CE-PAR-01), source page 1133. Cross-Domain Network Partitions, Unknown Outcomes, Durable Fencing, Migration and Safe Recovery</t></dd>
<dt>Problem</dt>
<dd><t>A lost acknowledgment is not a failed effect, and retries or recovery need separate authorization and fencing.</t></dd>
<dt>Actors</dt>
<dd><t>Coordinators (old and new); resource owners; workers.</t></dd>
<dt>Protected state</dt>
<dd><t>Original-operation identity, native-effect fingerprints, idempotency keys, durable fencing generations, checkpoint lineage.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: partition or lost response. Out: authoritative status or unknown; fenced stale leaders.</t></dd>
<dt>State transitions</dt>
<dd><t>executing, response lost, status queried, resolved or unknown.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Unknown or partial effects are not assumed rolled back or retried; stale workers rejected.</t></dd>
<dt>Threats</dt>
<dd><t>Split-brain coordination; unknown effect retried; stale leader.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Read-only source-authoritative status inquiry and resource-side monotonic fencing.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 123 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E11, E15.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>unknown outcome; durable fencing; idempotency key; partition recovery</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 36, 37, 38. Claim 36: Resource-Authoritative Reconciliation After Lost Execution Response; Claim 37: Monotonic Resource-Side Stale-Leader Fencing; Claim 38: Checkpoint-Lineage-Controlled Agent Migration and Recovered Execution Dependent claims 341-364 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e19">
<name>E19: Protected Session Closure, Obligation Reconciliation, Native Domain Sealing, and Irreversible Authority Retirement</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E19 (CE-SCL-01), source page 1187. Protected Session Closure, Obligation Reconciliation, Native Domain Sealing, and Irreversible Authority Retirement</t></dd>
<dt>Problem</dt>
<dd><t>A coordinator's DONE message does not prove every effect-capable path is closed.</t></dd>
<dt>Actors</dt>
<dd><t>Coordinators; each mandatory domain owner.</t></dd>
<dt>Protected state</dt>
<dd><t>Obligation inventory, per-domain seals, closure tombstone, grant revocation records.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: closure request. Out: sealed closure or withheld closure.</t></dd>
<dt>State transitions</dt>
<dd><t>closing, domains sealed, tombstone set.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Closure withheld while obligations or effect-capable grants remain; terminated session authority is not revived.</t></dd>
<dt>Threats</dt>
<dd><t>False complete-session closure; residual grants after closure.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Native-domain sealing at each domain.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 84 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E15, E18.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>session closure; domain seal; tombstone; authority retirement</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 39, 40. Claim 39: Complete Native-Obligation Inventory for Agent Session Termination; Claim 40: Independently Sealed Domains and Non-Reopenable Session Tombstone Dependent claims 365-380 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e20">
<name>E20: Verified Mandatory Approvers, Authenticated Hosting and Installation, and Revocable Supervisor Identity</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E20 (CE-JSA-01), source page 1226. Verified Mandatory Approvers, Authenticated Hosting and Installation, and Revocable Supervisor Identity</t></dd>
<dt>Problem</dt>
<dd><t>A supervisor cannot appoint itself; mandatory owners must consent and the supervisor must be installed at each domain.</t></dd>
<dt>Actors</dt>
<dd><t>Mandatory owners; supervisor; hosting runtime.</t></dd>
<dt>Protected state</dt>
<dd><t>Appointment, host-instance binding, installation records, per-call authorization, lease.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: owner consents. Out: authenticated supervisor installed with limited coordination functions.</t></dd>
<dt>State transitions</dt>
<dd><t>proposed, approved, hosted, installed, active, withdrawn.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>No self-appointed supervisor; service creation is not authorization to operate; coordination is not native effect authority.</t></dd>
<dt>Threats</dt>
<dd><t>Unauthorized supervisor creation or installation.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Local installation and per-call authorization at each domain.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 91 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E12, E19.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>joint supervisor; mandatory approver; host binding</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 41, 42. Claim 41: Owner-Authorized Bounded Cross-Organization Supervisor; Claim 42: Supervisor Host Instance Binding and Independent Local Installation Dependent claims 381-396 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e21">
<name>E21: Independent Cross-Domain Evaluator Appointment, Actual-Outcome Verification, Reviewer Accountability and Dispute Containment</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E21 (CE-IEO-01), source page 1286. Independent Cross-Domain Evaluator Appointment, Actual-Outcome Verification, Reviewer Accountability and Dispute Containment</t></dd>
<dt>Problem</dt>
<dd><t>Evaluators across organizations need owner-derived appointment, outcome sources independent by control group, and dispute containment.</t></dd>
<dt>Actors</dt>
<dd><t>Owners; roster of evaluators; reviewers; outcome sources.</t></dd>
<dt>Protected state</dt>
<dd><t>Evaluator roster, qualification and consent records, review round, dispute and appeal records.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: owner roster. Out: protected review result, including UNKNOWN and HOLD, with separate continuation.</t></dd>
<dt>State transitions</dt>
<dd><t>appointed, round open, closed, disputed, appealed.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Disputes are contained and new rounds isolated; unavailable evaluator never grants permission.</t></dd>
<dt>Threats</dt>
<dd><t>Collusion; false independence; dispute spillover.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Continuation separately authorized at owner gates.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 82 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E17, E20.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>cross-domain evaluator; reviewer accountability; dispute containment; appeal</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 43, 44. Claim 43: Cross-Organization Authorization of Independent Outcome Evaluation; Claim 44: Authenticated Evaluation Dispute, Appeal and New-Round Isolation Dependent claims 397-412 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e22">
<name>E22: Cross-Domain Trust and Policy Compatibility</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E22 (CE-CTP-01), source page 1356. Cross-Domain Trust and Policy Compatibility</t></dd>
<dt>Problem</dt>
<dd><t>Organizations do not share one authority root or one universal policy; compatibility must be verified, not assumed.</t></dd>
<dt>Actors</dt>
<dd><t>Independent organizations; resource owners.</t></dd>
<dt>Protected state</dt>
<dd><t>Trust records, verified policy assertions, schema and semantic compatibility results, non-delegable conditions, policy generations.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: policy assertions. Out: constraint intersection and bounded joint authority, or refusal for incompatibility.</t></dd>
<dt>State transitions</dt>
<dd><t>trust formed, policies compared, joint authority bounded, drift detected.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Incompatible policies are not merged optimistically; non-delegable conditions stay with owners.</t></dd>
<dt>Threats</dt>
<dd><t>Incompatible cross-domain policies; stale policy generation.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Each owner keeps local resource admission.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 121 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E20, E21.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>trust formation; policy semantic compatibility; constraint intersection</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 45, 46. Claim 45: Semantic Compatibility and Owner-Bounded Cross-Domain Cooperation; Claim 46: Continuous Cross-Domain Owner and Policy Drift Containment Dependent claims 413-428 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e23">
<name>E23: Leaderless Cross-Domain Agent Coordination</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E23 (CE-LDC-01), source page 1425. Leaderless Cross-Domain Agent Coordination</t></dd>
<dt>Problem</dt>
<dd><t>Cooperation without a common supervisor needs dependencies enforced at the resource owner, not just a task graph in an agent.</t></dd>
<dt>Actors</dt>
<dd><t>Independent resource owners; local supervisors.</t></dd>
<dt>Protected state</dt>
<dd><t>Causal dependency evidence, predecessor receipts, remote-status leases.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: predecessor receipts. Out: owner-side admission or refusal; convergence evidence only.</t></dd>
<dt>State transitions</dt>
<dd><t>predecessor required, receipt verified, admitted.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Convergence evidence is not executable permission; spoofed dependency is refused.</t></dd>
<dt>Threats</dt>
<dd><t>Dependency spoofing; split coordination views.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Resource-owner-side predecessor check.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 103 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E15, E22.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>leaderless coordination; causal dependency; predecessor receipt</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 47, 48. Claim 47: Leaderless Causal Task Execution Across Independent Resource Owners; Claim 48: Owner-Side Causal Receipt and Remote-Status Lease Validation Dependent claims 429-444 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e24">
<name>E24: Self-Expiring Supervisor and Evaluator Authority, Authenticated Revocation, and Generation-Fenced Replacement</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E24 (CE-SER-01), source page 1515. Self-Expiring Supervisor and Evaluator Authority, Authenticated Revocation, and Generation-Fenced Replacement</t></dd>
<dt>Problem</dt>
<dd><t>Delegated authority must expire or be withdrawn safely without claiming instant remote revocation.</t></dd>
<dt>Actors</dt>
<dd><t>Owners; supervisor or evaluator; successor.</t></dd>
<dt>Protected state</dt>
<dd><t>Lease, monotonic revocation serial, tombstones, local fencing acknowledgments, generation-fenced successor appointment.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: lease or withdrawal. Out: expired or revoked authority; safe successor.</t></dd>
<dt>State transitions</dt>
<dd><t>active, expiring, revoked, fenced, successor active.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Disconnected domain not yet informed is not claimed revoked; stale host rejected.</t></dd>
<dt>Threats</dt>
<dd><t>Revocation delay; stale supervisor during replacement; key compromise.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Local fencing at each domain.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 115 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E20, E21.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>authority lease; revocation convergence; generation fencing; successor activation</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 49, 50. Claim 49: Bounded Supervisor Lease and Authenticated Revocation Convergence; Claim 50: Safe Successor Activation After Partitioned Supervisory Withdrawal Dependent claims 445-460 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e25">
<name>E25: Dynamic Multi-Organization Membership and Principal-Independent Threshold Governance</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E25 (CE-DMG-01), source page 1579. Dynamic Multi-Organization Membership and Principal-Independent Threshold Governance</t></dd>
<dt>Problem</dt>
<dd><t>Voting authority belongs to organizations, not agent counts, and membership changes need epochs.</t></dd>
<dt>Actors</dt>
<dd><t>Member organizations; joiners; mandatory-veto holders.</t></dd>
<dt>Protected state</dt>
<dd><t>Eligible-electorate snapshot, membership epoch, ballots, joiner consent.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: membership change and ballots. Out: reconstituted membership with new epoch activation.</t></dd>
<dt>State transitions</dt>
<dd><t>proposed, voted, staged, old epoch fenced, new epoch active.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>One organization with many agents does not gain many votes; non-delegable vetoes hold; unresolved effects block activation.</t></dd>
<dt>Threats</dt>
<dd><t>Vote inflation; mandatory veto bypass.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Resource-local authorization after reconstitution.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 178 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E20, E22, E24.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>membership epoch; principal-independent threshold; owner veto; reconstitution</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 51, 52. Claim 51: Principal-Independent Threshold Governance With Non-Delegable Owner Veto; Claim 52: Protected Membership Reconstitution and Epoch-Safe Join/Leave Dependent claims 461-476 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e26">
<name>E26: Purpose-Bound Cross-Organizational Outcome Evidence</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E26 (CE-PCE-01), source page 1650. Purpose-Bound Cross-Organizational Outcome Evidence</t></dd>
<dt>Problem</dt>
<dd><t>Evaluators should receive no more confidential evidence than needed, and origin proof is not proof of physical truth.</t></dd>
<dt>Actors</dt>
<dd><t>Evidence custodians; verifier audience; owners.</t></dd>
<dt>Protected state</dt>
<dd><t>Purpose-bound requests, source commitments, attribute proofs, disclosure lineage, audit anchors.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: purpose-bound request. Out: selective, audience-bound evidence and audit record.</t></dd>
<dt>State transitions</dt>
<dd><t>requested, committed, disclosed, audited.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>Out-of-purpose or wrong-audience release refused; inclusion proof not treated as truth.</t></dd>
<dt>Threats</dt>
<dd><t>Selective-disclosure abuse; metadata leakage; forged inclusion proofs.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Evidence release gate at the custodian.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 98 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E16, E21.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>minimal disclosure; purpose-bound evidence; selective disclosure; evidence lineage</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 53, 54, 55. Claim 53: Source-Authenticated Selective Evidence Commitments; Claim 54: Audience- and Purpose-Bound Confidential Outcome Evidence Release; Claim 55: Minimal-Retention Evidence Audit With Source Sequence and Anchor Continuity Dependent claims 477-500 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
<section anchor="e27">
<name>E27: Multi-Domain Protected Execution Commitment, Partial-Outcome Reconciliation and Cross-Organizational Recovery</name>
<dl newline="true">
<dt>Title / identifier</dt>
<dd><t>E27 (CE-MEF-01), source page 1715. Multi-Domain Protected Execution Commitment, Partial-Outcome Reconciliation and Cross-Organizational Recovery</t></dd>
<dt>Problem</dt>
<dd><t>There is no global atomic transaction across independently administered domains; partial results and compensation need honest handling.</t></dd>
<dt>Actors</dt>
<dd><t>Resource owners in several domains; coordinator; evaluator.</t></dd>
<dt>Protected state</dt>
<dd><t>PREPARE and COMMIT records, predecessor manifests, partial-outcome manifest, compensation authorization, completeness roster.</t></dd>
<dt>Inputs / outputs</dt>
<dd><t>In: composite plan. Out: per-domain native outcomes and a non-executable completion record.</t></dd>
<dt>State transitions</dt>
<dd><t>prepared, committed, partial, unknown, compensated, complete.</t></dd>
<dt>Negative outcomes</dt>
<dd><t>PARTIAL_EFFECT and UNKNOWN are shown honestly; compensation is a new authorized act; completion record is not executable.</t></dd>
<dt>Threats</dt>
<dd><t>Partial commitment; compensation without consent; irreversible effect treated as rolled back.</t></dd>
<dt>Alternatives</dt>
<dd><t>Alternatives are other placements of the same function and existing mechanisms described in <xref target="interop"/>; selection is policy-specific.</t></dd>
<dt>Effects and enforcement boundary</dt>
<dd><t>Owner-specific protected prepare and commit at each native domain gate.</t></dd>
<dt>Test requirements</dt>
<dd><t>Negative tests per <xref target="threats"/>; source reports: 106 tests (reported). These are local tests of the source's modeled code paths and do not establish production security.</t></dd>
<dt>Relation to earlier embodiments</dt>
<dd><t>Builds on E12, E15, E18, E19.</t></dd>
<dt>Relation to execution finality</dt>
<dd><t>The evaluator output stays non-executable; effects are admitted only at a participating native boundary as described in <xref target="REL-FOUND"/>.</t></dd>
<dt>Industry terminology</dt>
<dd><t>prepare commit; partial effect; compensation; completeness evidence</t></dd>
<dt>Claim references</dt>
<dd><t>Independent claims 56, 57, 58, 59, 60. Claim 56: Independent Prepare-and-Commit Authorization at Native Domain Gates; Claim 57: Owner-Installed Mandatory Predecessor Manifest Enforcement; Claim 58: Partial Composite-Effect Reconciliation and Controlled Resume; Claim 59: Separately Authorized Compensation as a New Native Effect; Claim 60: Composite Native-Evidence Completeness Without Composite Execution Authority Dependent claims 501-540 (indexed, not individually analyzed here).</t></dd>
</dl>
</section>
</section>
<section anchor="model">
<name>Protocol-Independent Data Model</name>
<t>Illustrative, protocol-independent records follow. Field names are candidates only; JSON or CBOR encodings are examples. Signed portable artifacts are distinguished from protected local state, which is not exported.</t>
<section anchor="rec-evaluationrequest">
<name>EvaluationRequest</name>
<t>Illustrative example (portable):</t>
<sourcecode type="json"><![CDATA[{
 "id": "req-1",
 "task": "t-9",
 "plan_ref": "sha256:...",
 "owners": [
  "o1",
  "o2"
 ],
 "purpose": "pre-effect review"
}]]></sourcecode>
</section>
<section anchor="rec-evaluatorappointment">
<name>EvaluatorAppointment</name>
<t>Illustrative example (portable, signed by owners):</t>
<sourcecode type="json"><![CDATA[{
 "evaluator": "ev-3",
 "task": "t-9",
 "owners_consent": [
  "o1",
  "o2"
 ],
 "policy_gen": 7,
 "expiry": "2026-12-01T00:00:00Z",
 "uses": 1,
 "revocation_serial": 0
}]]></sourcecode>
</section>
<section anchor="rec-ownerauthoritymanifest">
<name>OwnerAuthorityManifest</name>
<t>Illustrative example (portable, signed):</t>
<sourcecode type="json"><![CDATA[{
 "task": "t-9",
 "mandatory_owners": [
  "o1",
  "o2"
 ],
 "vetoes": [
  "o2"
 ],
 "non_delegable": [
  "payment"
 ]
}]]></sourcecode>
</section>
<section anchor="rec-evaluationround">
<name>EvaluationRound</name>
<t>Illustrative example (protected local plus signed summary):</t>
<sourcecode type="json"><![CDATA[{
 "round": "r-1",
 "evidence_set": "sha256:...",
 "state": "closed",
 "reviewers": 3,
 "control_groups": 3,
 "dissent": []
}]]></sourcecode>
</section>
<section anchor="rec-evidencerecord">
<name>EvidenceRecord</name>
<t>Illustrative example (portable, source-signed):</t>
<sourcecode type="json"><![CDATA[{
 "id": "ev-12",
 "source": "o1",
 "role": "resource_origin",
 "task": "t-9",
 "fresh_until": "...",
 "policy_gen": 7,
 "class": "native_receipt"
}]]></sourcecode>
</section>
<section anchor="rec-evaluatordetermination">
<name>EvaluatorDetermination</name>
<t>Illustrative example (portable, signed):</t>
<sourcecode type="json"><![CDATA[{
 "type": "HOLD",
 "round": "r-1",
 "task": "t-9",
 "valid": [
  "..",
  "..."
 ],
 "executable": false
}]]></sourcecode>
</section>
<section anchor="rec-supervisorappointment">
<name>SupervisorAppointment</name>
<t>Illustrative example (portable, signed):</t>
<sourcecode type="json"><![CDATA[{
 "supervisor": "sv-1",
 "owners_consent": [
  "o1",
  "o2"
 ],
 "host_binding": "inst-5",
 "lease": "...",
 "serial": 3
}]]></sourcecode>
</section>
<section anchor="rec-nativeoutcomereceipt">
<name>NativeOutcomeReceipt</name>
<t>Illustrative example (resource-origin, signed or protected local):</t>
<sourcecode type="json"><![CDATA[{
 "op": "op-77",
 "domain": "d1",
 "result": "COMMITTED",
 "fingerprint": "...",
 "gen": 12
}]]></sourcecode>
</section>
<section anchor="rec-partialcompletionmanifest">
<name>PartialCompletionManifest</name>
<t>Illustrative example (portable, signed):</t>
<sourcecode type="json"><![CDATA[{
 "composite": "c-2",
 "domains": {
  "d1": "COMMITTED",
  "d2": "UNKNOWN",
  "d3": "DENIED"
 },
 "status": "PARTIAL_EFFECT"
}]]></sourcecode>
</section>
<section anchor="rec-closureseal">
<name>ClosureSeal</name>
<t>Illustrative example (portable, per-domain signed):</t>
<sourcecode type="json"><![CDATA[{
 "session": "s-4",
 "domains_sealed": [
  "d1",
  "d2"
 ],
 "tombstone": true
}]]></sourcecode>
</section>
<section anchor="rec-selectivedisclosureproof">
<name>SelectiveDisclosureProof</name>
<t>Illustrative example (portable, source-authenticated):</t>
<sourcecode type="json"><![CDATA[{
 "purpose": "outcome audit",
 "audience": "rev-1",
 "attributes": [
  "amount_range"
 ],
 "commitment": "..."
}]]></sourcecode>
</section>
</section>
<section anchor="pred">
<name>Formal Predicates and Pseudocode</name>
<t>Pseudocode is illustrative and non-normative; it is not a reference implementation and the source's attached tests are local tests only.</t>
<section>
<name>Evaluator admission</name>
<sourcecode type="python"><![CDATA[def admit_evaluator(e, task, now):
    a = appointments.get(e, task)
    if a is None or a.revoked or now >= a.expiry: return HOLD
    if not owners_consented(a, task.mandatory_owners): return HOLD
    if not independent(e, task.parties): return DENY
    return OK]]></sourcecode>
</section>
<section>
<name>Evidence admission</name>
<sourcecode type="python"><![CDATA[def admit_evidence(x, r):
    checks = [origin(x), role(x), integrity(x),
              binds(x, r.task), fresh(x), policy_gen(x),
              independent(x), uncorrelated(x)]
    if not all(checks): return EXCLUDE
    return ADMIT   # admitted != outcome truth]]></sourcecode>
</section>
<section>
<name>Determination verification</name>
<sourcecode type="python"><![CDATA[def verify_det(d, now):
    return (sig_ok(d) and d.executable is False and
            now in d.validity and snapshot_valid(d.snapshot)
            and not withdrawn(d))]]></sourcecode>
</section>
<section>
<name>Policy-bound continuation</name>
<sourcecode type="python"><![CDATA[def continue_(d, owner_policy):
    if not verify_det(d, now()): return HOLD
    grant = owner_policy.authorize(d)   # separate decision
    return grant   # gate still admits or withholds]]></sourcecode>
</section>
<section>
<name>Revocation</name>
<sourcecode type="python"><![CDATA[def revoke(a):
    a.serial += 1; publish(a)
    for dom in domains(a): fence(dom, a.generation)
    # disconnected domains end by lease expiry, not by signal]]></sourcecode>
</section>
<section>
<name>Review-round closure</name>
<sourcecode type="python"><![CDATA[def close_round(r):
    if not mandatory_present(r): return HOLD
    if control_groups(r) < r.policy.min_groups: return HOLD
    if not dissent_included(r): return REJECT
    r.state = 'closed'; return seal(r)]]></sourcecode>
</section>
<section>
<name>Outcome reconciliation</name>
<sourcecode type="python"><![CDATA[def reconcile(op):
    s = source_status(op)       # read-only, source-authoritative
    if s is None: return UNKNOWN  # never auto-retry
    return COMMITTED if s.committed else NEEDS_NEW_AUTH]]></sourcecode>
</section>
<section>
<name>Relationships That Are Not Implied</name>
<t>Seven central relationships are NOT implied:</t>
<ul>
<li>A determination does not imply authorization.</li>
<li>A valid signature does not imply outcome truth.</li>
<li>Admitted evidence does not imply independent evidence.</li>
<li>Approval by many credentials does not imply many independent reviewers.</li>
<li>Agreement of agents does not imply permission to act.</li>
<li>A simulation does not imply that the effect occurred or will occur as predicted.</li>
<li>A convergence record or completion record does not imply executable authority.</li>
</ul>
</section>
</section>
<section anchor="threats">
<name>Threat Model and Negative Tests</name>
<t>The following threats each carry eight fields. They are generic categories; mitigation depends on deployment and residual limitations are stated.</t>
<section anchor="t01">
<name>T01 Agent self-appointment as evaluator</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>act as evaluator</t></dd>
<dt>Capability</dt>
<dd><t>controls an agent</t></dd>
<dt>Trust boundary</dt>
<dd><t>agent to appointment authority</t></dd>
<dt>Invariant at risk</dt>
<dd><t>appointment comes from the required authority source</t></dd>
<dt>Required verification</dt>
<dd><t>check appointment issuer and scope</t></dd>
<dt>Failure response</dt>
<dd><t>reject; HOLD</t></dd>
<dt>Residual limitation</dt>
<dd><t>issuer compromise</t></dd>
<dt>Negative test</dt>
<dd><t>agent declares itself neutral; determination refused</t></dd>
</dl>
</section>
<section anchor="t02">
<name>T02 Forged organizational consent</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>fake an owner's approval</t></dd>
<dt>Capability</dt>
<dd><t>forges or replays approval</t></dd>
<dt>Trust boundary</dt>
<dd><t>owner to appointment</t></dd>
<dt>Invariant at risk</dt>
<dd><t>consent is current and from the actual owner</t></dd>
<dt>Required verification</dt>
<dd><t>verify consent source and freshness</t></dd>
<dt>Failure response</dt>
<dd><t>reject appointment</t></dd>
<dt>Residual limitation</dt>
<dd><t>stolen credentials</t></dd>
<dt>Negative test</dt>
<dd><t>forged consent record; appointment refused</t></dd>
</dl>
</section>
<section anchor="t03">
<name>T03 Missing mandatory resource owner</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>omit an owner</t></dd>
<dt>Capability</dt>
<dd><t>controls owner list</t></dd>
<dt>Trust boundary</dt>
<dd><t>owner resolution</t></dd>
<dt>Invariant at risk</dt>
<dd><t>all mandatory owners resolved</t></dd>
<dt>Required verification</dt>
<dd><t>compare against owner manifest</t></dd>
<dt>Failure response</dt>
<dd><t>HOLD</t></dd>
<dt>Residual limitation</dt>
<dd><t>manifest error</t></dd>
<dt>Negative test</dt>
<dd><t>one owner removed; no approval</t></dd>
</dl>
</section>
<section anchor="t04">
<name>T04 Unauthorized supervisor creation</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>create a supervisor</t></dd>
<dt>Capability</dt>
<dd><t>controls one domain</t></dd>
<dt>Trust boundary</dt>
<dd><t>owner to supervisor</t></dd>
<dt>Invariant at risk</dt>
<dd><t>supervisor needs mandatory owner consent</t></dd>
<dt>Required verification</dt>
<dd><t>check joint appointment</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>weak owner process</t></dd>
<dt>Negative test</dt>
<dd><t>single-owner creation; refused</t></dd>
</dl>
</section>
<section anchor="t05">
<name>T05 Unauthorized supervisor installation</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>install a rogue supervisor</t></dd>
<dt>Capability</dt>
<dd><t>access to a host</t></dd>
<dt>Trust boundary</dt>
<dd><t>host to domain</t></dd>
<dt>Invariant at risk</dt>
<dd><t>installation bound to approved instance</t></dd>
<dt>Required verification</dt>
<dd><t>verify host-instance binding</t></dd>
<dt>Failure response</dt>
<dd><t>reject installation</t></dd>
<dt>Residual limitation</dt>
<dd><t>host compromise</t></dd>
<dt>Negative test</dt>
<dd><t>unapproved instance install; refused</t></dd>
</dl>
</section>
<section anchor="t06">
<name>T06 Improper role co-location</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>hold evaluator and gate roles unrestrained</t></dd>
<dt>Capability</dt>
<dd><t>controls a shared host</t></dd>
<dt>Trust boundary</dt>
<dd><t>role boundary</t></dd>
<dt>Invariant at risk</dt>
<dd><t>co-location grants no inherited authority</t></dd>
<dt>Required verification</dt>
<dd><t>check separation of role authority</t></dd>
<dt>Failure response</dt>
<dd><t>deny inherited authority</t></dd>
<dt>Residual limitation</dt>
<dd><t>weak isolation</t></dd>
<dt>Negative test</dt>
<dd><t>same host evaluator tries gate use; denied</t></dd>
</dl>
</section>
<section anchor="t07">
<name>T07 False claims of independence</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>pass off a related evaluator as neutral</t></dd>
<dt>Capability</dt>
<dd><t>claims neutrality</t></dd>
<dt>Trust boundary</dt>
<dd><t>independence check</t></dd>
<dt>Invariant at risk</dt>
<dd><t>independence from actual control relationships</t></dd>
<dt>Required verification</dt>
<dd><t>check control groups</t></dd>
<dt>Failure response</dt>
<dd><t>reject evaluator</t></dd>
<dt>Residual limitation</dt>
<dd><t>hidden control</t></dd>
<dt>Negative test</dt>
<dd><t>self-declared neutral; rejected</t></dd>
</dl>
</section>
<section anchor="t08">
<name>T08 Multiple credentials in one control group</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>inflate reviewers</t></dd>
<dt>Capability</dt>
<dd><t>holds many credentials</t></dd>
<dt>Trust boundary</dt>
<dd><t>reviewer diversity</t></dd>
<dt>Invariant at risk</dt>
<dd><t>count control groups not accounts</t></dd>
<dt>Required verification</dt>
<dd><t>map credentials to groups</t></dd>
<dt>Failure response</dt>
<dd><t>count once</t></dd>
<dt>Residual limitation</dt>
<dd><t>unknown affiliation</t></dd>
<dt>Negative test</dt>
<dd><t>ten keys one group; counted once</t></dd>
</dl>
</section>
<section anchor="t09">
<name>T09 Reviewer collusion</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>force a result</t></dd>
<dt>Capability</dt>
<dd><t>colluding reviewers</t></dd>
<dt>Trust boundary</dt>
<dd><t>review round</t></dd>
<dt>Invariant at risk</dt>
<dd><t>dissent and independence preserved</t></dd>
<dt>Required verification</dt>
<dd><t>check source independence</t></dd>
<dt>Failure response</dt>
<dd><t>HOLD or DISPUTED</t></dd>
<dt>Residual limitation</dt>
<dd><t>covert collusion</t></dd>
<dt>Negative test</dt>
<dd><t>colluding pair; round not approved</t></dd>
</dl>
</section>
<section anchor="t10">
<name>T10 Suppressed dissent</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>hide a dissent</t></dd>
<dt>Capability</dt>
<dd><t>controls aggregator</t></dd>
<dt>Trust boundary</dt>
<dd><t>review aggregation</t></dd>
<dt>Invariant at risk</dt>
<dd><t>mandatory dissent cannot be dropped</t></dd>
<dt>Required verification</dt>
<dd><t>verify dissent included</t></dd>
<dt>Failure response</dt>
<dd><t>reject round</t></dd>
<dt>Residual limitation</dt>
<dd><t>unrecorded dissent</t></dd>
<dt>Negative test</dt>
<dd><t>dissent removed; closure refused</t></dd>
</dl>
</section>
<section anchor="t11">
<name>T11 Stale evaluator mandate</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>use expired mandate</t></dd>
<dt>Capability</dt>
<dd><t>retains old credential</t></dd>
<dt>Trust boundary</dt>
<dd><t>appointment lifetime</t></dd>
<dt>Invariant at risk</dt>
<dd><t>mandate currently valid</t></dd>
<dt>Required verification</dt>
<dd><t>check validity at use</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>clock error</t></dd>
<dt>Negative test</dt>
<dd><t>expired mandate presented; refused</t></dd>
</dl>
</section>
<section anchor="t12">
<name>T12 Stale review-round authorization</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>admit evidence after closure</t></dd>
<dt>Capability</dt>
<dd><t>retains round token</t></dd>
<dt>Trust boundary</dt>
<dd><t>round lifetime</t></dd>
<dt>Invariant at risk</dt>
<dd><t>closed round admits nothing</t></dd>
<dt>Required verification</dt>
<dd><t>check round state</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>delayed closure</t></dd>
<dt>Negative test</dt>
<dd><t>late evidence after close; refused</t></dd>
</dl>
</section>
<section anchor="t13">
<name>T13 Reuse of determinations across projects</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>reuse approval elsewhere</t></dd>
<dt>Capability</dt>
<dd><t>holds a determination</t></dd>
<dt>Trust boundary</dt>
<dd><t>determination scope</t></dd>
<dt>Invariant at risk</dt>
<dd><t>bound to task and round</t></dd>
<dt>Required verification</dt>
<dd><t>check scope binding</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>loose binding</t></dd>
<dt>Negative test</dt>
<dd><t>project A result used for B; refused</t></dd>
</dl>
</section>
<section anchor="t14">
<name>T14 Compromised persistent evaluator infrastructure</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>poison many tasks</t></dd>
<dt>Capability</dt>
<dd><t>controls platform</t></dd>
<dt>Trust boundary</dt>
<dd><t>platform to tasks</t></dd>
<dt>Invariant at risk</dt>
<dd><t>no automatic inheritance across tasks</t></dd>
<dt>Required verification</dt>
<dd><t>per-task mandates; platform compromise handling</t></dd>
<dt>Failure response</dt>
<dd><t>revoke all contexts</t></dd>
<dt>Residual limitation</dt>
<dd><t>platform key compromise</t></dd>
<dt>Negative test</dt>
<dd><t>platform compromised; all tasks revalidated</t></dd>
</dl>
</section>
<section anchor="t15">
<name>T15 Expired ephemeral evaluator credentials</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>reuse retired instance</t></dd>
<dt>Capability</dt>
<dd><t>holds old instance key</t></dd>
<dt>Trust boundary</dt>
<dd><t>instance lifetime</t></dd>
<dt>Invariant at risk</dt>
<dd><t>retired instance has no authority</t></dd>
<dt>Required verification</dt>
<dd><t>check instance state</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>retention of keys</t></dd>
<dt>Negative test</dt>
<dd><t>retired instance acts; refused</t></dd>
</dl>
</section>
<section anchor="t16">
<name>T16 Evaluator evidence-access escalation</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>read more evidence</t></dd>
<dt>Capability</dt>
<dd><t>evaluator process</t></dd>
<dt>Trust boundary</dt>
<dd><t>evidence access</t></dd>
<dt>Invariant at risk</dt>
<dd><t>access limited to mandate</t></dd>
<dt>Required verification</dt>
<dd><t>check evidence classes</t></dd>
<dt>Failure response</dt>
<dd><t>deny</t></dd>
<dt>Residual limitation</dt>
<dd><t>side channels</t></dd>
<dt>Negative test</dt>
<dd><t>evaluator requests unlisted class; denied</t></dd>
</dl>
</section>
<section anchor="t17">
<name>T17 Malicious instructions embedded in evidence</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>steer the evaluator</t></dd>
<dt>Capability</dt>
<dd><t>controls a source</t></dd>
<dt>Trust boundary</dt>
<dd><t>evidence to evaluator</t></dd>
<dt>Invariant at risk</dt>
<dd><t>evidence is data not instruction</t></dd>
<dt>Required verification</dt>
<dd><t>treat content as untrusted</t></dd>
<dt>Failure response</dt>
<dd><t>ignore instruction; flag</t></dd>
<dt>Residual limitation</dt>
<dd><t>model susceptibility</t></dd>
<dt>Negative test</dt>
<dd><t>embedded command; no behavior change</t></dd>
</dl>
</section>
<section anchor="t18">
<name>T18 Authenticated but false outcome reports</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>claim false success</t></dd>
<dt>Capability</dt>
<dd><t>holds valid keys</t></dd>
<dt>Trust boundary</dt>
<dd><t>source to evaluator</t></dd>
<dt>Invariant at risk</dt>
<dd><t>origin is not outcome truth</t></dd>
<dt>Required verification</dt>
<dd><t>require resource-origin evidence</t></dd>
<dt>Failure response</dt>
<dd><t>UNKNOWN</t></dd>
<dt>Residual limitation</dt>
<dd><t>dishonest source</t></dd>
<dt>Negative test</dt>
<dd><t>signed false success; not accepted as truth</t></dd>
</dl>
</section>
<section anchor="t19">
<name>T19 Correlated observations miscounted as independent</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>fake diversity</t></dd>
<dt>Capability</dt>
<dd><t>two reports one origin</t></dd>
<dt>Trust boundary</dt>
<dd><t>evidence independence</t></dd>
<dt>Invariant at risk</dt>
<dd><t>underlying provenance independent</t></dd>
<dt>Required verification</dt>
<dd><t>check provenance groups</t></dd>
<dt>Failure response</dt>
<dd><t>count once</t></dd>
<dt>Residual limitation</dt>
<dd><t>hidden provenance</t></dd>
<dt>Negative test</dt>
<dd><t>two APIs one source; counted once</t></dd>
</dl>
</section>
<section anchor="t20">
<name>T20 Stale policy generation</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>use old policy</t></dd>
<dt>Capability</dt>
<dd><t>holds old policy</t></dd>
<dt>Trust boundary</dt>
<dd><t>policy to evaluation</t></dd>
<dt>Invariant at risk</dt>
<dd><t>current generation applied</t></dd>
<dt>Required verification</dt>
<dd><t>compare generations</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>propagation delay</t></dd>
<dt>Negative test</dt>
<dd><t>old generation determination; refused</t></dd>
</dl>
</section>
<section anchor="t21">
<name>T21 In-flight messages after freeze</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>alter plan after freeze</t></dd>
<dt>Capability</dt>
<dd><t>queued messages</t></dd>
<dt>Trust boundary</dt>
<dd><t>communication edge</t></dd>
<dt>Invariant at risk</dt>
<dd><t>unaccounted messages not admitted</t></dd>
<dt>Required verification</dt>
<dd><t>pending-work inventory</t></dd>
<dt>Failure response</dt>
<dd><t>hold; reconcile</t></dd>
<dt>Residual limitation</dt>
<dd><t>unobserved path</t></dd>
<dt>Negative test</dt>
<dd><t>late message; snapshot invalidated</t></dd>
</dl>
</section>
<section anchor="t22">
<name>T22 Unfenced background work</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>act during hold</t></dd>
<dt>Capability</dt>
<dd><t>background worker</t></dd>
<dt>Trust boundary</dt>
<dd><t>worker to resource</t></dd>
<dt>Invariant at risk</dt>
<dd><t>workers fenced</t></dd>
<dt>Required verification</dt>
<dd><t>check fence generation</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>hidden workers</t></dd>
<dt>Negative test</dt>
<dd><t>worker acts during hold; rejected</t></dd>
</dl>
</section>
<section anchor="t23">
<name>T23 Unauthorized renegotiation</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>change a committed plan</t></dd>
<dt>Capability</dt>
<dd><t>one agent</t></dd>
<dt>Trust boundary</dt>
<dd><t>plan commitment</t></dd>
<dt>Invariant at risk</dt>
<dd><t>revision needs new consents</t></dd>
<dt>Required verification</dt>
<dd><t>check fresh recommitment</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>owner ambiguity</t></dd>
<dt>Negative test</dt>
<dd><t>unilateral patch; refused</t></dd>
</dl>
</section>
<section anchor="t24">
<name>T24 Next-actor impersonation</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>act as selected actor</t></dd>
<dt>Capability</dt>
<dd><t>stolen identity</t></dd>
<dt>Trust boundary</dt>
<dd><t>actor binding</t></dd>
<dt>Invariant at risk</dt>
<dd><t>passport bound to actor</t></dd>
<dt>Required verification</dt>
<dd><t>check actor identity</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>identity compromise</t></dd>
<dt>Negative test</dt>
<dd><t>wrong actor presents passport; refused</t></dd>
</dl>
</section>
<section anchor="t25">
<name>T25 Next-move passport replay</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>reuse a passport</t></dd>
<dt>Capability</dt>
<dd><t>captured passport</t></dd>
<dt>Trust boundary</dt>
<dd><t>consumption state</t></dd>
<dt>Invariant at risk</dt>
<dd><t>single use</t></dd>
<dt>Required verification</dt>
<dd><t>check consumed state</t></dd>
<dt>Failure response</dt>
<dd><t>reject; reconcile</t></dd>
<dt>Residual limitation</dt>
<dd><t>state loss</t></dd>
<dt>Negative test</dt>
<dd><t>second use refused</t></dd>
</dl>
</section>
<section anchor="t26">
<name>T26 Substitute-agent credential inheritance</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>inherit old credentials</t></dd>
<dt>Capability</dt>
<dd><t>new agent</t></dd>
<dt>Trust boundary</dt>
<dd><t>substitute to resource</t></dd>
<dt>Invariant at risk</dt>
<dd><t>no inheritance</t></dd>
<dt>Required verification</dt>
<dd><t>fresh authorization required</t></dd>
<dt>Failure response</dt>
<dd><t>deny</t></dd>
<dt>Residual limitation</dt>
<dd><t>weak revocation</t></dd>
<dt>Negative test</dt>
<dd><t>substitute uses old capability; denied</t></dd>
</dl>
</section>
<section anchor="t27">
<name>T27 Pending-work escrow substitution</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>swap escrowed work</t></dd>
<dt>Capability</dt>
<dd><t>controls escrow store</t></dd>
<dt>Trust boundary</dt>
<dd><t>escrow commit</t></dd>
<dt>Invariant at risk</dt>
<dd><t>escrow canonical and committed</t></dd>
<dt>Required verification</dt>
<dd><t>verify escrow digest or record</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>store compromise</t></dd>
<dt>Negative test</dt>
<dd><t>altered escrow; refused</t></dd>
</dl>
</section>
<section anchor="t28">
<name>T28 False simulation-as-completion assertion</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>claim simulation as done</t></dd>
<dt>Capability</dt>
<dd><t>controls reporter</t></dd>
<dt>Trust boundary</dt>
<dd><t>simulation to record</t></dd>
<dt>Invariant at risk</dt>
<dd><t>simulation not completion</t></dd>
<dt>Required verification</dt>
<dd><t>require native receipt</t></dd>
<dt>Failure response</dt>
<dd><t>reject completion</t></dd>
<dt>Residual limitation</dt>
<dd><t>fidelity gap</t></dd>
<dt>Negative test</dt>
<dd><t>simulated success; no completion</t></dd>
</dl>
</section>
<section anchor="t29">
<name>T29 Missing phase receipt</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>skip a phase</t></dd>
<dt>Capability</dt>
<dd><t>controls coordinator</t></dd>
<dt>Trust boundary</dt>
<dd><t>phase gate</t></dd>
<dt>Invariant at risk</dt>
<dd><t>phase needs receipt</t></dd>
<dt>Required verification</dt>
<dd><t>check receipt</t></dd>
<dt>Failure response</dt>
<dd><t>block advancement</t></dd>
<dt>Residual limitation</dt>
<dd><t>receipt delay</t></dd>
<dt>Negative test</dt>
<dd><t>advance without receipt; refused</t></dd>
</dl>
</section>
<section anchor="t30">
<name>T30 Network partition and split-brain coordination</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>two coordinators</t></dd>
<dt>Capability</dt>
<dd><t>partition</t></dd>
<dt>Trust boundary</dt>
<dd><t>coordinator to resource</t></dd>
<dt>Invariant at risk</dt>
<dd><t>one valid generation</t></dd>
<dt>Required verification</dt>
<dd><t>monotonic fencing</t></dd>
<dt>Failure response</dt>
<dd><t>fence older</t></dd>
<dt>Residual limitation</dt>
<dd><t>partition detection limits</t></dd>
<dt>Negative test</dt>
<dd><t>two leaders; stale one rejected</t></dd>
</dl>
</section>
<section anchor="t31">
<name>T31 Unknown native effect incorrectly retried</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>duplicate effect</t></dd>
<dt>Capability</dt>
<dd><t>lost ack</t></dd>
<dt>Trust boundary</dt>
<dd><t>retry path</t></dd>
<dt>Invariant at risk</dt>
<dd><t>unknown not retried automatically</t></dd>
<dt>Required verification</dt>
<dd><t>status inquiry first</t></dd>
<dt>Failure response</dt>
<dd><t>hold retry</t></dd>
<dt>Residual limitation</dt>
<dd><t>status unavailable</t></dd>
<dt>Negative test</dt>
<dd><t>lost ack; no blind retry</t></dd>
</dl>
</section>
<section anchor="t32">
<name>T32 Stale leader or supervisor during replacement</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>old leader acts</t></dd>
<dt>Capability</dt>
<dd><t>old host</t></dd>
<dt>Trust boundary</dt>
<dd><t>successor activation</t></dd>
<dt>Invariant at risk</dt>
<dd><t>generation-fenced</t></dd>
<dt>Required verification</dt>
<dd><t>check generation</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>delayed fence</t></dd>
<dt>Negative test</dt>
<dd><t>stale leader acts after replacement; refused</t></dd>
</dl>
</section>
<section anchor="t33">
<name>T33 Revocation propagation delays</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>act before revocation arrives</t></dd>
<dt>Capability</dt>
<dd><t>disconnected domain</t></dd>
<dt>Trust boundary</dt>
<dd><t>revocation to domain</t></dd>
<dt>Invariant at risk</dt>
<dd><t>no claim of instant revocation</t></dd>
<dt>Required verification</dt>
<dd><t>bounded freshness</t></dd>
<dt>Failure response</dt>
<dd><t>fail closed after bound</t></dd>
<dt>Residual limitation</dt>
<dd><t>irreducible delay</t></dd>
<dt>Negative test</dt>
<dd><t>disconnected domain; authority expires by lease</t></dd>
</dl>
</section>
<section anchor="t34">
<name>T34 Membership-vote inflation</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>gain votes</t></dd>
<dt>Capability</dt>
<dd><t>many agents</t></dd>
<dt>Trust boundary</dt>
<dd><t>membership voting</t></dd>
<dt>Invariant at risk</dt>
<dd><t>one organization one vote</t></dd>
<dt>Required verification</dt>
<dd><t>count organizations</t></dd>
<dt>Failure response</dt>
<dd><t>count once</t></dd>
<dt>Residual limitation</dt>
<dd><t>org identity</t></dd>
<dt>Negative test</dt>
<dd><t>ten agents one org; one vote</t></dd>
</dl>
</section>
<section anchor="t35">
<name>T35 Mandatory veto bypass</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>override an owner</t></dd>
<dt>Capability</dt>
<dd><t>majority</t></dd>
<dt>Trust boundary</dt>
<dd><t>veto</t></dd>
<dt>Invariant at risk</dt>
<dd><t>non-delegable veto</t></dd>
<dt>Required verification</dt>
<dd><t>check veto records</t></dd>
<dt>Failure response</dt>
<dd><t>block</t></dd>
<dt>Residual limitation</dt>
<dd><t>owner ambiguity</t></dd>
<dt>Negative test</dt>
<dd><t>majority over veto; blocked</t></dd>
</dl>
</section>
<section anchor="t36">
<name>T36 Incompatible cross-domain policies</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>merge unlike policies</t></dd>
<dt>Capability</dt>
<dd><t>policy mismatch</t></dd>
<dt>Trust boundary</dt>
<dd><t>policy compatibility</t></dd>
<dt>Invariant at risk</dt>
<dd><t>semantic compatibility verified</t></dd>
<dt>Required verification</dt>
<dd><t>compare semantics</t></dd>
<dt>Failure response</dt>
<dd><t>refuse</t></dd>
<dt>Residual limitation</dt>
<dd><t>semantic ambiguity</t></dd>
<dt>Negative test</dt>
<dd><t>same word different meaning; refused</t></dd>
</dl>
</section>
<section anchor="t37">
<name>T37 Leaderless causal-dependency spoofing</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>fake predecessor</t></dd>
<dt>Capability</dt>
<dd><t>forged receipt</t></dd>
<dt>Trust boundary</dt>
<dd><t>owner-side check</t></dd>
<dt>Invariant at risk</dt>
<dd><t>predecessor verified at owner</t></dd>
<dt>Required verification</dt>
<dd><t>verify receipt origin</t></dd>
<dt>Failure response</dt>
<dd><t>refuse</t></dd>
<dt>Residual limitation</dt>
<dd><t>origin compromise</t></dd>
<dt>Negative test</dt>
<dd><t>forged predecessor; refused</t></dd>
</dl>
</section>
<section anchor="t38">
<name>T38 Missing native predecessor effects</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>proceed without predecessor</t></dd>
<dt>Capability</dt>
<dd><t>coordinator</t></dd>
<dt>Trust boundary</dt>
<dd><t>owner gate</t></dd>
<dt>Invariant at risk</dt>
<dd><t>manifest enforced at owner</t></dd>
<dt>Required verification</dt>
<dd><t>check manifest</t></dd>
<dt>Failure response</dt>
<dd><t>block</t></dd>
<dt>Residual limitation</dt>
<dd><t>manifest gaps</t></dd>
<dt>Negative test</dt>
<dd><t>missing predecessor; blocked</t></dd>
</dl>
</section>
<section anchor="t39">
<name>T39 Selective evidence-disclosure abuse</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>extract extra data</t></dd>
<dt>Capability</dt>
<dd><t>evaluator</t></dd>
<dt>Trust boundary</dt>
<dd><t>disclosure gate</t></dd>
<dt>Invariant at risk</dt>
<dd><t>purpose and audience bound</t></dd>
<dt>Required verification</dt>
<dd><t>check purpose</t></dd>
<dt>Failure response</dt>
<dd><t>deny</t></dd>
<dt>Residual limitation</dt>
<dd><t>inference</t></dd>
<dt>Negative test</dt>
<dd><t>request outside purpose; denied</t></dd>
</dl>
</section>
<section anchor="t40">
<name>T40 Metadata leakage</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>infer from metadata</t></dd>
<dt>Capability</dt>
<dd><t>observer</t></dd>
<dt>Trust boundary</dt>
<dd><t>evidence exchange</t></dd>
<dt>Invariant at risk</dt>
<dd><t>metadata minimized</t></dd>
<dt>Required verification</dt>
<dd><t>review metadata</t></dd>
<dt>Failure response</dt>
<dd><t>minimize</t></dd>
<dt>Residual limitation</dt>
<dd><t>residual inference</t></dd>
<dt>Negative test</dt>
<dd><t>metadata in logs; flagged</t></dd>
</dl>
</section>
<section anchor="t41">
<name>T41 Forged evidence-inclusion proofs</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>fake inclusion</t></dd>
<dt>Capability</dt>
<dd><t>forger</t></dd>
<dt>Trust boundary</dt>
<dd><t>proof verification</t></dd>
<dt>Invariant at risk</dt>
<dd><t>proof verified against source</t></dd>
<dt>Required verification</dt>
<dd><t>verify against commitment</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>algorithm weakness</t></dd>
<dt>Negative test</dt>
<dd><t>forged proof; rejected</t></dd>
</dl>
</section>
<section anchor="t42">
<name>T42 False complete-session closure</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>claim closed</t></dd>
<dt>Capability</dt>
<dd><t>coordinator</t></dd>
<dt>Trust boundary</dt>
<dd><t>closure</t></dd>
<dt>Invariant at risk</dt>
<dd><t>all domains sealed</t></dd>
<dt>Required verification</dt>
<dd><t>check seals</t></dd>
<dt>Failure response</dt>
<dd><t>withhold closure</t></dd>
<dt>Residual limitation</dt>
<dd><t>missing domain</t></dd>
<dt>Negative test</dt>
<dd><t>DONE with open obligation; refused</t></dd>
</dl>
</section>
<section anchor="t43">
<name>T43 Residual effect-capable grants after task closure</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>reuse grants</t></dd>
<dt>Capability</dt>
<dd><t>holds old grant</t></dd>
<dt>Trust boundary</dt>
<dd><t>grant lifecycle</t></dd>
<dt>Invariant at risk</dt>
<dd><t>grants revoked or exhausted</t></dd>
<dt>Required verification</dt>
<dd><t>check grants</t></dd>
<dt>Failure response</dt>
<dd><t>revoke</t></dd>
<dt>Residual limitation</dt>
<dd><t>untracked grant</t></dd>
<dt>Negative test</dt>
<dd><t>grant used after closure; refused</t></dd>
</dl>
</section>
<section anchor="t44">
<name>T44 Incorrect global atomicity assumptions</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>assume atomic</t></dd>
<dt>Capability</dt>
<dd><t>designer</t></dd>
<dt>Trust boundary</dt>
<dd><t>composite plan</t></dd>
<dt>Invariant at risk</dt>
<dd><t>no global atomicity</t></dd>
<dt>Required verification</dt>
<dd><t>handle partial results</t></dd>
<dt>Failure response</dt>
<dd><t>report partial</t></dd>
<dt>Residual limitation</dt>
<dd><t>design error</t></dd>
<dt>Negative test</dt>
<dd><t>one domain commits; partial shown</t></dd>
</dl>
</section>
<section anchor="t45">
<name>T45 Partial native commitment</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>hide partial</t></dd>
<dt>Capability</dt>
<dd><t>coordinator</t></dd>
<dt>Trust boundary</dt>
<dd><t>composite status</t></dd>
<dt>Invariant at risk</dt>
<dd><t>partial effect reported</t></dd>
<dt>Required verification</dt>
<dd><t>per-domain evidence</t></dd>
<dt>Failure response</dt>
<dd><t>PARTIAL_EFFECT</t></dd>
<dt>Residual limitation</dt>
<dd><t>unobserved domain</t></dd>
<dt>Negative test</dt>
<dd><t>one commit missing; shown partial</t></dd>
</dl>
</section>
<section anchor="t46">
<name>T46 Compensation without owner consent</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>reverse an effect unilaterally</t></dd>
<dt>Capability</dt>
<dd><t>coordinator</t></dd>
<dt>Trust boundary</dt>
<dd><t>compensation</t></dd>
<dt>Invariant at risk</dt>
<dd><t>compensation is a new authorized act</t></dd>
<dt>Required verification</dt>
<dd><t>check owner authorization</t></dd>
<dt>Failure response</dt>
<dd><t>refuse</t></dd>
<dt>Residual limitation</dt>
<dd><t>urgent pressure</t></dd>
<dt>Negative test</dt>
<dd><t>unauthorized compensation; refused</t></dd>
</dl>
</section>
<section anchor="t47">
<name>T47 Irreversible effect treated as rolled back</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>claim rollback</t></dd>
<dt>Capability</dt>
<dd><t>coordinator</t></dd>
<dt>Trust boundary</dt>
<dd><t>history</t></dd>
<dt>Invariant at risk</dt>
<dd><t>irreversible effects stay in record</t></dd>
<dt>Required verification</dt>
<dd><t>reversibility classes</t></dd>
<dt>Failure response</dt>
<dd><t>mark irreversible</t></dd>
<dt>Residual limitation</dt>
<dd><t>classification error</t></dd>
<dt>Negative test</dt>
<dd><t>irreversible shown as rolled back; rejected</t></dd>
</dl>
</section>
<section anchor="t48">
<name>T48 Byzantine or dishonest resource-origin reports</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>lie at source</t></dd>
<dt>Capability</dt>
<dd><t>owner source</t></dd>
<dt>Trust boundary</dt>
<dd><t>source honesty</t></dd>
<dt>Invariant at risk</dt>
<dd><t>limits of source honesty stated</t></dd>
<dt>Required verification</dt>
<dd><t>cross-source checks</t></dd>
<dt>Failure response</dt>
<dd><t>UNKNOWN or DISPUTED</t></dd>
<dt>Residual limitation</dt>
<dd><t>source dishonesty</t></dd>
<dt>Negative test</dt>
<dd><t>contradicting sources; DISPUTED</t></dd>
</dl>
</section>
<section anchor="t49">
<name>T49 Protected-local controller compromise</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>subvert local state</t></dd>
<dt>Capability</dt>
<dd><t>local access</t></dd>
<dt>Trust boundary</dt>
<dd><t>controller</t></dd>
<dt>Invariant at risk</dt>
<dd><t>local trust stated</t></dd>
<dt>Required verification</dt>
<dd><t>detect inconsistency</t></dd>
<dt>Failure response</dt>
<dd><t>fail closed</t></dd>
<dt>Residual limitation</dt>
<dd><t>trusted base</t></dd>
<dt>Negative test</dt>
<dd><t>tampered local state; detected where possible</t></dd>
</dl>
</section>
<section anchor="t50">
<name>T50 Cryptographic algorithm or key compromise</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>forge signatures</t></dd>
<dt>Capability</dt>
<dd><t>breaks keys</t></dd>
<dt>Trust boundary</dt>
<dd><t>signature layer</t></dd>
<dt>Invariant at risk</dt>
<dd><t>algorithm agility optional</t></dd>
<dt>Required verification</dt>
<dd><t>rotate and revoke</t></dd>
<dt>Failure response</dt>
<dd><t>revoke</t></dd>
<dt>Residual limitation</dt>
<dd><t>unknown break</t></dd>
<dt>Negative test</dt>
<dd><t>revoked key signature; rejected</t></dd>
</dl>
</section>
<section anchor="t51">
<name>T51 Denial of service against mandatory evaluators</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>stall decisions</t></dd>
<dt>Capability</dt>
<dd><t>floods evaluators</t></dd>
<dt>Trust boundary</dt>
<dd><t>availability</t></dd>
<dt>Invariant at risk</dt>
<dd><t>unavailability never grants permission</t></dd>
<dt>Required verification</dt>
<dd><t>timeouts</t></dd>
<dt>Failure response</dt>
<dd><t>HOLD</t></dd>
<dt>Residual limitation</dt>
<dd><t>delay</t></dd>
<dt>Negative test</dt>
<dd><t>evaluator down; HOLD not approve</t></dd>
</dl>
</section>
<section anchor="t52">
<name>T52 Excessive reviewer or evidence requirements causing deadlock</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>block progress</t></dd>
<dt>Capability</dt>
<dd><t>designs impossible rules</t></dd>
<dt>Trust boundary</dt>
<dd><t>policy</t></dd>
<dt>Invariant at risk</dt>
<dd><t>rules satisfiable or escalate</t></dd>
<dt>Required verification</dt>
<dd><t>deadlock detection</t></dd>
<dt>Failure response</dt>
<dd><t>escalate to human</t></dd>
<dt>Residual limitation</dt>
<dd><t>policy design</t></dd>
<dt>Negative test</dt>
<dd><t>unsatisfiable quorum; escalates</t></dd>
</dl>
</section>
<section anchor="t53">
<name>T53 Resource-owner ambiguity</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>unclear owner</t></dd>
<dt>Capability</dt>
<dd><t>misconfigured</t></dd>
<dt>Trust boundary</dt>
<dd><t>owner resolution</t></dd>
<dt>Invariant at risk</dt>
<dd><t>actual owners resolved</t></dd>
<dt>Required verification</dt>
<dd><t>resolve ownership</t></dd>
<dt>Failure response</dt>
<dd><t>HOLD</t></dd>
<dt>Residual limitation</dt>
<dd><t>registry error</t></dd>
<dt>Negative test</dt>
<dd><t>two claimed owners; HOLD</t></dd>
</dl>
</section>
<section anchor="t54">
<name>T54 Time-source compromise</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>extend authority</t></dd>
<dt>Capability</dt>
<dd><t>controls clock</t></dd>
<dt>Trust boundary</dt>
<dd><t>time</t></dd>
<dt>Invariant at risk</dt>
<dd><t>expiry uses protected time or events</t></dd>
<dt>Required verification</dt>
<dd><t>protected time source</t></dd>
<dt>Failure response</dt>
<dd><t>fail closed</t></dd>
<dt>Residual limitation</dt>
<dd><t>time dependency</t></dd>
<dt>Negative test</dt>
<dd><t>clock rewind; authority not extended</t></dd>
</dl>
</section>
<section anchor="t55">
<name>T55 Replay after checkpoint restoration</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>reuse consumed state</t></dd>
<dt>Capability</dt>
<dd><t>restores checkpoint</t></dd>
<dt>Trust boundary</dt>
<dd><t>state restoration</t></dd>
<dt>Invariant at risk</dt>
<dd><t>consumption survives restore</t></dd>
<dt>Required verification</dt>
<dd><t>checkpoint lineage</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>backup practices</t></dd>
<dt>Negative test</dt>
<dd><t>restored checkpoint reuses grant; refused</t></dd>
</dl>
</section>
<section anchor="t56">
<name>T56 Cross-jurisdiction evidence disclosure</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>unlawful transfer</t></dd>
<dt>Capability</dt>
<dd><t>evaluator in other region</t></dd>
<dt>Trust boundary</dt>
<dd><t>evidence export</t></dd>
<dt>Invariant at risk</dt>
<dd><t>jurisdiction rules checked</t></dd>
<dt>Required verification</dt>
<dd><t>check constraints</t></dd>
<dt>Failure response</dt>
<dd><t>deny</t></dd>
<dt>Residual limitation</dt>
<dd><t>legal ambiguity</t></dd>
<dt>Negative test</dt>
<dd><t>evidence to forbidden region; denied</t></dd>
</dl>
</section>
<section anchor="t57">
<name>T57 Faulty recovery after a delayed valid receipt</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>double effect</t></dd>
<dt>Capability</dt>
<dd><t>late receipt</t></dd>
<dt>Trust boundary</dt>
<dd><t>recovery</t></dd>
<dt>Invariant at risk</dt>
<dd><t>late receipt reconciled</t></dd>
<dt>Required verification</dt>
<dd><t>re-inquire</t></dd>
<dt>Failure response</dt>
<dd><t>reconcile</t></dd>
<dt>Residual limitation</dt>
<dd><t>timing</t></dd>
<dt>Negative test</dt>
<dd><t>receipt arrives after recovery; reconciled not repeated</t></dd>
</dl>
</section>
<section anchor="t58">
<name>T58 Privacy-violating diagnostic logs</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>leak via logs</t></dd>
<dt>Capability</dt>
<dd><t>reads logs</t></dd>
<dt>Trust boundary</dt>
<dd><t>logging</t></dd>
<dt>Invariant at risk</dt>
<dd><t>logs minimized</t></dd>
<dt>Required verification</dt>
<dd><t>review logs</t></dd>
<dt>Failure response</dt>
<dd><t>redact</t></dd>
<dt>Residual limitation</dt>
<dd><t>residual info</t></dd>
<dt>Negative test</dt>
<dd><t>log with full evidence; flagged</t></dd>
</dl>
</section>
<section anchor="t59">
<name>T59 Incorrect use of a positive evaluator determination as an execution token</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>treat approval as authority</t></dd>
<dt>Capability</dt>
<dd><t>integrator</t></dd>
<dt>Trust boundary</dt>
<dd><t>determination to gate</t></dd>
<dt>Invariant at risk</dt>
<dd><t>determination non-executable</t></dd>
<dt>Required verification</dt>
<dd><t>gate ignores determinations</t></dd>
<dt>Failure response</dt>
<dd><t>reject</t></dd>
<dt>Residual limitation</dt>
<dd><t>integration error</t></dd>
<dt>Negative test</dt>
<dd><t>gate receives determination alone; refuses</t></dd>
</dl>
</section>
<section anchor="t60">
<name>T60 Uncovered alternate native effectuation paths</name>
<dl newline="true">
<dt>Attacker objective</dt>
<dd><t>bypass the gate</t></dd>
<dt>Capability</dt>
<dd><t>finds other path</t></dd>
<dt>Trust boundary</dt>
<dd><t>coverage</t></dd>
<dt>Invariant at risk</dt>
<dd><t>claims limited to covered paths</t></dd>
<dt>Required verification</dt>
<dd><t>coverage inventory</t></dd>
<dt>Failure response</dt>
<dd><t>state limitation</t></dd>
<dt>Residual limitation</dt>
<dd><t>unknown paths</t></dd>
<dt>Negative test</dt>
<dd><t>effect via uncovered path; limitation stated</t></dd>
</dl>
</section>
</section>
<section anchor="interop">
<name>Interoperability Requirements</name>
<t>No new wire protocol is proposed where existing standards suffice. Candidate carriers, with architecture requirements kept separate from candidate schemas and normative requirements:</t>
<table anchor="tbl-interop">
<name>Interoperability candidates</name>
<thead><tr><th>Mechanism</th><th>Possible use</th><th>Note</th></tr></thead>
<tbody>
<tr><td>HTTP and TLS</td><td>transport of requests and determinations</td><td>does not supply appointment semantics</td></tr>
<tr><td>OAuth 2.0 and token exchange</td><td>delegation of bounded evaluator access</td><td>token is not an effect authorization</td></tr>
<tr><td>JWT and CWT with COSE/JOSE</td><td>signed appointment and determination artifacts</td><td>signature is not outcome truth</td></tr>
<tr><td>Workload identity</td><td>evaluator and supervisor identity</td><td>identity is not independence</td></tr>
<tr><td>Remote attestation (RATS)</td><td>host and instance binding</td><td>attests state, not truthfulness of evidence</td></tr>
<tr><td>Distributed tracing</td><td>evidence lineage correlation</td><td>traces are claims, not receipts</td></tr>
<tr><td>Transparency logs</td><td>audit continuity</td><td>inclusion is not truth</td></tr>
<tr><td>Message queues and agent protocols</td><td>carrying requests</td><td>delivery is not admission</td></tr>
</tbody>
</table>
<ul>
<li>Evaluation <bcp14>SHOULD NOT</bcp14> require per-inference network calls to an external service; local or batched evaluation is a legitimate placement.</li>
<li>Cryptographic and protected-local alternatives are both admissible; post-quantum mechanisms are optional and not universally required.</li>
</ul>
</section>
<section anchor="industry">
<name>Industry Profiles</name>
<t>Hypothetical and non-endorsed. These descriptions identify potential research contexts only and were not verified against vendor documentation by the editor; readers must verify against public product documentation. Established product functionality is separated from proposed integration, and the alignment is AI-generated and must be cross-checked against the original filings. It is not an endorsement, validation, adoption claim or infringement statement.</t>
<dl newline="true">
<dt>Frontier AI platforms</dt>
<dd><t>Established: tool-call approval and human review steps exist in various platforms. Proposed: a gate placed at the tool-execution broker would reject an act whose appointment, round or determination is stale, and would require a separate owner-issued authorization before the call is released. Why it matters: model-approved text then cannot by itself cause an effect.</t></dd>
<dt>Enterprise and cloud</dt>
<dd><t>Established: policy decision and enforcement points, workload identity, audit logs. Proposed: an evaluator service whose appointment is derived from resource-owner consent and whose determinations are inputs to, not substitutes for, an existing PEP check at the resource API.</t></dd>
<dt>Financial</dt>
<dd><t>Established: dual control, maker-checker, reconciliation. Proposed: a reviewer appointed per transaction class, with minimal-disclosure outcome evidence, UNKNOWN on lost acknowledgments and compensation as a separately authorized act (Examples F, I, J, K).</t></dd>
<dt>Telecom</dt>
<dd><t>Established: orchestration and slice policies. Proposed: cross-domain evaluator for multi-operator changes, with owner-side predecessor checks and partial-outcome manifests when one domain does not commit.</t></dd>
<dt>Industrial</dt>
<dd><t>Established: safety interlocks and change management. Proposed: a persistent evaluator serving many maintenance projects with per-project mandates (Example B), with actuation only through the owner's existing controlled interface.</t></dd>
</dl>
</section>
<section anchor="examples">
<name>Worked Examples A-L</name>
<t>Each example gives initial authority, protected state, eligible evidence, determination, separate native authority and failure outcomes.</t>
<section anchor="ex-a">
<name>Example A: Two-company procurement</name>
<t>Initial authority: each company owns its purchasing and payment authority. Protected state: a committed, non-effective joint plan, frozen. Eligible evidence: inventory and price receipts from each owner. Determination: APPROVED_FOR_CONSIDERATION by an evaluator appointed by both. Native authority: each company's gate separately admits purchase and payment. Failures: stale price evidence yields HOLD; missing consent yields refusal; unknown payment outcome yields UNKNOWN.</t>
</section>
<section anchor="ex-b">
<name>Example B: Persistent industrial evaluator</name>
<t>Initial authority: one service, many projects, separate mandates. Protected state: per-project namespaces, rounds and validity. Evidence: sensor logs and work-order receipts. Determination: per-project, valid for a bounded interval. Native authority: plant controls admit changes. Failures: project A expiry gives no authority to B; platform compromise suspends all contexts.</t>
</section>
<section anchor="ex-c">
<name>Example C: Cross-domain evaluator committee</name>
<t>Initial authority: three owners set reviewer qualifications. Protected state: round with control-group map. Evidence: provenance-diverse reports. Determination: closed only with mandatory reviewer present and dissent preserved. Native authority: owners authorize separately. Failures: three accounts in one group count once; suppressed dissent rejects the round.</t>
</section>
<section anchor="ex-d">
<name>Example D: Communication freeze race</name>
<t>Initial authority: plan committed. Protected state: freeze epoch. Evidence: pending-work inventory. Determination: HOLD while a late message is unaccounted; snapshot invalidated. Native authority: unchanged, none granted. Failures: partial acknowledgment is not a freeze.</t>
</section>
<section anchor="ex-e">
<name>Example E: Failed agent substitution</name>
<t>Initial authority: old agent's consumed capability. Protected state: escrowed task data, source fenced. Evidence: neutrality screening. Determination: SUBSTITUTE recommended. Native authority: the substitute has no valid permission to continue consuming the old capability. Failures: credential inheritance denied.</t>
</section>
<section anchor="ex-f">
<name>Example F: Network partition</name>
<t>Initial authority: one authorized payment. Protected state: operation fingerprint and idempotency key. Evidence: none received. Determination: UNKNOWN. Native authority: retry needs new authorization. Failures: the coordinator must not repeat; a read-only status inquiry runs first.</t>
</section>
<section anchor="ex-g">
<name>Example G: Multi-organization membership change</name>
<t>Initial authority: epoch e with two owners. Protected state: electorate snapshot. Evidence: joiner consent; unresolved earlier operation. Determination: HOLD until the earlier operation settles. Native authority: local installation required. Failures: disconnected owner's veto not bypassed; activation withheld.</t>
</section>
<section anchor="ex-h">
<name>Example H: Leaderless cooperation</name>
<t>Initial authority: two independent owners. Protected state: predecessor manifest at each owner. Evidence: native receipt of operation one at owner two. Determination: convergence evidence only. Native authority: owner two admits operation two only if the receipt verifies. Failures: spoofed predecessor refused.</t>
</section>
<section anchor="ex-i">
<name>Example I: Minimal-disclosure outcome verification</name>
<t>Initial authority: bank controls records. Protected state: attribute commitments. Evidence: selective attributes for a stated purpose and audience. Determination: verified outcome attribute. Native authority: none gained. Failures: out-of-purpose request refused; metadata minimized.</t>
</section>
<section anchor="ex-j">
<name>Example J: Partial financial and inventory completion</name>
<t>Initial authority: two domain owners. Protected state: PREPARE records. Evidence: ledger committed, inventory not released. Determination: PARTIAL_EFFECT. Native authority: each domain separately. Failures: no global atomicity assumed; completion record not executable.</t>
</section>
<section anchor="ex-k">
<name>Example K: Reversible and irreversible compensation</name>
<t>Initial authority: a committed operation. Protected state: history record. Evidence: owner consent. Determination: compensation proposed. Native authority: compensation is a new native effect with its own receipt. Failures: irreversible effects cannot be recorded as rolled back.</t>
</section>
<section anchor="ex-l">
<name>Example L: Protected session termination</name>
<t>Initial authority: task apparently ended. Protected state: obligation inventory. Evidence: queued work and old grants found. Determination: closure withheld. Native authority: grants revoked or exhausted, then seals. Failures: DONE message alone is rejected.</t>
</section>
</section>
<section anchor="perf">
<name>Performance and Legacy Compatibility</name>
<t>The document makes no universal latency claim. Three paths are distinguished: a fast path (valid cached determination within validity), an evaluation path (new review), and a recovery path (partition, UNKNOWN, revocation). Legacy systems may adopt the architecture by placing the gate at an existing enforcement point, with the evaluator offline or batched.</t>
<t>Measurement plan: p50, p95, p99 and p99.9 latency; throughput; setup cost per appointment; reviewer count effect; revocation propagation time; partition recovery time; false acceptance and false denial rates; UNKNOWN frequency; and privacy leakage. No figures are asserted here. Source-reported test counts (see <xref target="xwalk"/>) come from local Python reference code only and are not performance or production evidence.</t>
</section>
<section anchor="prior">
<name>Prior Art and IETF Overlap</name>
<t>Each topic below is established prior art or practice in its own right. The applicant does not claim that any isolated primitive, signature, quorum, AI evaluator or distributed-transaction mechanism was invented by the applicant. The architecture's distinction, if any, lies in the particular cooperation of protected appointment, role independence, lifecycle separation, evidence eligibility, bounded determination, owner-specific authorization and native-effect admission. That statement is not a legal conclusion.</t>
<ul>
<li>Identity and access management: established; overlap and use is acknowledged.</li>
<li>Role-based and attribute-based access control: established; overlap and use is acknowledged.</li>
<li>Policy decision and enforcement points: established; overlap and use is acknowledged.</li>
<li>OAuth authorization and delegation: established; overlap and use is acknowledged.</li>
<li>Proof-of-possession tokens: established; overlap and use is acknowledged.</li>
<li>Distributed consensus: established; overlap and use is acknowledged.</li>
<li>Byzantine fault tolerance: established; overlap and use is acknowledged.</li>
<li>Quorum and multiparty approval: established; overlap and use is acknowledged.</li>
<li>Workflow orchestration: established; overlap and use is acknowledged.</li>
<li>Saga transactions: established; overlap and use is acknowledged.</li>
<li>Two-phase commit: established; overlap and use is acknowledged.</li>
<li>Idempotent operations: established; overlap and use is acknowledged.</li>
<li>Distributed fencing tokens: established; overlap and use is acknowledged.</li>
<li>Remote attestation: established; overlap and use is acknowledged.</li>
<li>Zero trust: established; overlap and use is acknowledged.</li>
<li>Agent tracing: established; overlap and use is acknowledged.</li>
<li>AI evaluation frameworks: established; overlap and use is acknowledged.</li>
<li>Human-in-the-loop approval: established; overlap and use is acknowledged.</li>
<li>Cryptographic audit logs: established; overlap and use is acknowledged.</li>
<li>Selective disclosure: established; overlap and use is acknowledged.</li>
<li>Federated identity and policy: established; overlap and use is acknowledged.</li>
<li>Distributed task coordination: established; overlap and use is acknowledged.</li>
</ul>
<section>
<name>Related IETF Work</name>
<t>Related individual drafts by the same author: <xref target="REL-FOUND"/>, <xref target="REL-CBEA"/>, <xref target="REL-CAUSAL"/>, <xref target="REL-IEV"/> (identifiers as supplied by the author; verify on the Datatracker). Others' work on agent delegation, handoff and requirements (<xref target="I-D.agentic-ai-usecases-requirements"/>) is acknowledged; the editor has not claimed a full survey.</t>
</section>
<section>
<name>Non-Normative Patent-Examination Methodology</name>
<ul>
<li>Identify specific claim limitations and read them as written.</li>
<li>Identify the closest prior-art combinations.</li>
<li>State the objective technical problem.</li>
<li>Assess functional interdependence of the elements.</li>
<li>Ask whether there was a motivation to combine, and a reasonable expectation of success.</li>
<li>Distinguish what a skilled person could combine from what the prior art would have motivated them to combine.</li>
<li>Avoid hindsight.</li>
<li>Distinguish new terminology from actual technical differences.</li>
</ul>
<t>This is not a legal conclusion on patentability.</t>
</section>
</section>
<section anchor="security">
<name>Security Considerations</name>
<t>The architecture reduces the chance that an evaluation result is mistaken for authority, but it depends on correct enforcement at participating boundaries. Non-bypassable native-effect control applies only to actual participating protected enforcement paths. See <xref target="threats"/> for 60 threat categories. Residual limits: a compromised trusted base, a dishonest resource owner, undetected correlated sources, algorithm compromise, and uncovered alternate effect paths. No implementation is assumed to guarantee globally atomic transactions, instantaneous revocation during disconnection, physically infallible source evidence, zero-latency evaluation or universal interoperability.</t>
</section>
<section anchor="privacy">
<name>Privacy Considerations</name>
<t>Evidence is disclosed only under authorized purpose and audience constraints and with minimal retention. Metadata, logs and traces can leak information and are minimized. Cross-jurisdiction evidence transfer follows applicable constraints; the document gives no legal advice.</t>
</section>
<section anchor="iana">
<name>IANA Considerations</name>
<t>This document requests no IANA actions.</t>
</section>
<section anchor="notices">
<name>Notices</name>
<dl newline="true">
<dt>Human-Originated Invention</dt>
<dd><t>The underlying technical invention and patent disclosure are attributed to human inventor Sangam Kumar Das. AI-assisted preparation of the draft, terminology mapping, diagrams, industry comparisons or explanatory text does not independently establish AI inventorship.</t></dd>
<dt>AI-Assisted Research and Verification</dt>
<dd><t>AI-assisted material can contain mistakes, omissions, inaccurate product references or misinterpretations of technical claims. All assertions must be checked against the original filed patent documents, verified RFCs, primary standards documentation and official industry publications.</t></dd>
<dt>IETF Status</dt>
<dd><t>The document is an individual technical proposal and does not constitute an endorsed IETF standard, approved RFC, adopted working-group specification or IETF consensus position. No IETF-assigned identifier is claimed.</t></dd>
<dt>Industry Neutrality</dt>
<dd><t>References to OpenAI, Anthropic, Google, Microsoft, NVIDIA, AWS, Alibaba, Huawei, SAP, IBM, Nokia, Ericsson, financial companies or other organizations identify potential research contexts only. No adoption, endorsement, affiliation, copying, infringement or commercial implementation is asserted.</t></dd>
<dt>Technical Limits</dt>
<dd><t>A positive evaluator determination does not independently establish a real-world effect or confer authority for such an effect. Non-bypassable native-effect control applies only to actual participating protected enforcement paths.</t></dd>
<dt>IPR Requirements</dt>
<dd><t>The author must comply with applicable IETF BCP 78 and BCP 79 requirements. Mentioning the patent inside a draft does not replace any required formal IETF IPR disclosure. No FRAND, royalty-free or other licensing commitment is made or implied here.</t></dd>
</dl>
</section>
<section anchor="ipr">
<name>Intellectual Property and Research Lineage</name>
<t>Current source disclosure: inventor Sangam Kumar Das; Indian Provisional Patent Application 202631131650; applicant-reported filing date 3 November 2026 (the source PDF is named with a 'NOV 4' label but contains a digital-signature date of 3 November; no 4 November filing is inferred). Supplied source: 1,905 pages, 27 embodiments, 540 numbered claims.</t>
<t>Lineage: PCT/IB2026/055615 filed 4 June 2026, published 16 July 2026 as WO 2026/150382 A2. Related Indian provisionals reported by the applicant: 202631134241 (8 November 2026), 202631136555 (15 November 2026). Later features are not assigned the June priority date automatically; entitlement to any priority date is a legal question outside this document.</t>
<t>Public research records: https://doi.org/10.5281/zenodo.24274259 and https://doi.org/10.5281/zenodo.24076231; Technical Disclosure Commons records 12542, 12934 and 12973. These were not all fetched by the editor; verify. An IETF IPR disclosure (reference 7610) exists for earlier drafts of the author; whether it covers this draft is not established. IPR will be disclosed at the appropriate time as per IETF guidelines.</t>
</section>
</middle>
<back>
<references><name>Normative References</name>
<reference anchor="RFC2119" target="https://www.rfc-editor.org/info/rfc2119"><front><title>Key words for use in RFCs to Indicate Requirement Levels</title><author initials="S" surname="Bradner"/><date year="1997" month="March"/></front><seriesInfo name="RFC" value="2119"/><seriesInfo name="BCP" value="14"/></reference>
<reference anchor="RFC8174" target="https://www.rfc-editor.org/info/rfc8174"><front><title>Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words</title><author initials="B" surname="Leiba"/><date year="2017" month="May"/></front><seriesInfo name="RFC" value="8174"/><seriesInfo name="BCP" value="14"/></reference>
</references>
<references><name>Informative References</name>
<reference anchor="REL-FOUND" target="https://datatracker.ietf.org/"><front><title>Execution Finality Foundations</title><author fullname="Sangam Kumar Das" initials="S." surname="Das"/><date year="2026" month="November"/></front><seriesInfo name="Internet-Draft" value="draft-das-execution-finality-foundations"/><refcontent>Identifier as supplied by the author; verify on the Datatracker. Not fetched.</refcontent></reference>
<reference anchor="REL-CBEA" target="https://datatracker.ietf.org/"><front><title>Consequence-Bound Execution Authority</title><author fullname="Sangam Kumar Das" initials="S." surname="Das"/><date year="2026" month="November"/></front><seriesInfo name="Internet-Draft" value="draft-das-consequence-bound-execution-authority"/><refcontent>Identifier as supplied by the author; verify on the Datatracker. Not fetched.</refcontent></reference>
<reference anchor="REL-CAUSAL" target="https://datatracker.ietf.org/"><front><title>Causal Diagnostics for Semantic Drift in Autonomous Agent Communications</title><author fullname="Sangam Kumar Das" initials="S." surname="Das"/><date year="2026" month="November"/></front><seriesInfo name="Internet-Draft" value="draft-das-causal-semantic-handoff-diagnostics"/><refcontent>Identifier as supplied by the author; verify on the Datatracker. Not fetched.</refcontent></reference>
<reference anchor="REL-IEV" target="https://datatracker.ietf.org/"><front><title>Receipt-Gated Independent Effect Verification (title as supplied; verify)</title><author fullname="Sangam Kumar Das" initials="S." surname="Das"/><date year="2026" month="November"/></front><seriesInfo name="Internet-Draft" value="draft-das-receipt-gated-iev"/><refcontent>Identifier as supplied by the author; verify on the Datatracker. Not fetched.</refcontent></reference>
<reference anchor="I-D.agentic-ai-usecases-requirements" target="https://datatracker.ietf.org/doc/draft-agentic-ai-usecases-requirements/"><front><title>Agentic AI Use Cases and Requirements</title><author initials="T." surname="Reddy"/><author initials="Z." surname="Sarker"/><author initials="K." surname="Yao"/><date year="2026" month="May"/></front><seriesInfo name="Internet-Draft" value="draft-agentic-ai-usecases-requirements"/></reference>
<reference anchor="PCT-055615" target="https://patentscope.wipo.int/"><front><title>PCT/IB2026/055615 (WO 2026/150382 A2)</title><author fullname="Sangam Kumar Das" initials="S." surname="Das"/><date year="2026" month="July" day="16"/></front><refcontent>Filed 4 June 2026; published 16 July 2026. Verify on PATENTSCOPE.</refcontent></reference>
<reference anchor="IN-131650" target="https://ipindia.gov.in/"><front><title>Indian Provisional Patent Application 202631131650</title><author fullname="Sangam Kumar Das" initials="S." surname="Das"/><date year="2026" month="November" day="3"/></front><refcontent>Date as reported by the applicant; not a certified registry record.</refcontent></reference>
<reference anchor="TDC-12542" target="https://www.tdcommons.org/dpubs_series/12542/"><front><title>Technical Disclosure Commons record 12542 (PCT file; title not verified)</title><author fullname="Sangam Das" initials="S." surname="Das"/><date year="2026"/></front><refcontent>Not fetched; verify.</refcontent></reference>
</references>
<section anchor="app-a">
<name>Appendix A: E01-E27 Profiles and Claim Crosswalk</name>
<t>The complete profiles for E01-E27 are in <xref target="e1"/> and <xref target="e2"/>; the crosswalk of all 60 independent claims (mapped to the embodiment of each) and the indexed dependent claims 61-540 is in <xref target="xwalk"/>. Independent claim titles are listed in each profile's Claim references field.</t>
</section>
<section anchor="app-b">
<name>Appendix B: Diagrams and State Machines</name>
<t>27 diagrams appear in the body; their captions are listed here for reference.</t>
<ul>
<li>1. Principal architecture (evaluation authority ends at the dotted line)</li>
<li>2. Four roles and their trust boundaries</li>
<li>3. Evaluator function and location-neutral deployment</li>
<li>4. Ephemeral evaluator lifecycle</li>
<li>5. Persistent evaluator lifecycle</li>
<li>6. Hybrid evaluator platform</li>
<li>7. Six independent lifetimes (not nested)</li>
<li>8. Owner and appointment resolution</li>
<li>9. Joint plan commitment and non-effective handoff</li>
<li>10. Freeze, pending-message reconciliation, quiescence</li>
<li>11. Causal lineage and snapshot invalidation</li>
<li>12. Evidence provenance and source-truth separation</li>
<li>13. Virtual consequence simulation and risk classification</li>
<li>14. Next actor, passport and automatic refreeze</li>
<li>15. Substitute selection and pending-work escrow</li>
<li>16. Phase receipt verification and feedback continuation</li>
<li>17. Reviewer independence, control groups, dissent</li>
<li>18. Partition, UNKNOWN outcome and fencing</li>
<li>19. Session closure, domain seals and tombstones</li>
<li>20. Supervisor installation and revocation</li>
<li>21. Trust and policy compatibility across domains</li>
<li>22. Leaderless coordination with owner-side predecessors</li>
<li>23. Membership, owner voting and reconstitution</li>
<li>24. Purpose-bound selective disclosure and audit</li>
<li>25. Multi-domain PREPARE, COMMIT and partial-effect reconciliation</li>
<li>26. Independently authorized compensation</li>
<li>27. Complete native-domain evidence verification</li>
</ul>
<t>Appointment state machine: proposed -&gt; owner-approved -&gt; active -&gt; (expired | revoked); evaluation round: open -&gt; admitting -&gt; (closed | disputed); determination: valid -&gt; (stale | withdrawn). Transitions to active require all mandatory owner consents; any lifecycle can end without ending the others (see <xref target="life"/>).</t>
</section>
<section anchor="app-c">
<name>Appendix C: Message and Evidence Examples</name>
<t>Illustrative records are in <xref target="model"/>. Example evidence chain: a resource-origin receipt (EvidenceRecord, class native_receipt) is admitted into a closed EvaluationRound, which yields a non-executable EvaluatorDetermination; owner policy separately issues a bounded authorization; the native gate admits or withholds the act and emits a NativeOutcomeReceipt.</t>
</section>
<section anchor="app-d">
<name>Appendix D: Negative Security Tests</name>
<t>The 60 negative-test categories, each with a concrete test, are in <xref target="threats"/> (T01-T60). Test results reported in the source are local reference-code tests and are not reproduced as evidence of security.</t>
</section>
<section anchor="app-e">
<name>Appendix E: Examiner Keyword and Phrase Index</name>
<t>Counts: 212 keywords and 115 phrases. The index serves searches on WIPO PATENTSCOPE, EPO Espacenet, Google Patents, IEEE, ACM, arXiv, the IETF Datatracker and general technical literature. Presence of a term does not imply novelty or infringement. Cross-references give the embodiment (E) and, through <xref target="xwalk"/>, the independent claims.</t>
<section>
<name>Evaluator architecture</name>
<ul>
<li>Location-Independent Evaluator (E04,E17,E21)</li>
<li>Neutral Evaluation Service (E04,E21)</li>
<li>Independent Evaluator (E04,E17)</li>
<li>Evaluation Function (E04)</li>
<li>Policy-Bounded Evaluator (E04,E06)</li>
<li>Trusted Local Evaluator (E04,E17)</li>
<li>Human-Assisted Evaluator (E04,E17)</li>
<li>Protected Decision Service (E04)</li>
<li>Single-Component Evaluator (E04)</li>
<li>Evaluator Role (E04)</li>
<li>Evaluator Software Identity (E17)</li>
<li>Decision Service Isolation (E04,E05)</li>
<li>Neutral Evaluator (E04,E13)</li>
<li>Review Function (E17)</li>
<li>Consequence Evaluator (E04)</li>
<li>Non-Effective Decision (E04)</li>
<li>Typed Determination (E04,E21)</li>
<li>Recommendation Versus Permission (E04,E07)</li>
<li>Evaluation Mandate (E04,E24)</li>
<li>Evaluator Roster (E21)</li>
</ul>
</section>
<section>
<name>Evaluator authority</name>
<ul>
<li>Protected Evaluator Appointment (E21,E24)</li>
<li>Task-Bound Mandate (E20,E21)</li>
<li>Appointment Completeness (E20,E21)</li>
<li>Authority Source Resolution (E20,E21)</li>
<li>Evaluator Role Qualification (E21)</li>
<li>Owner-Approved Review (E21)</li>
<li>Mandate Revocation (E24)</li>
<li>Scope-Limited Determination (E04,E21)</li>
<li>Appointment Expiry (E24)</li>
<li>Use-Count Limit (E24)</li>
<li>Authority Lease (E24)</li>
<li>Authority Generation (E24)</li>
<li>Owner-Derived Authority (E20,E21)</li>
<li>Mandatory Owner Resolution (E20)</li>
<li>Self-Appointment Prevention (E20)</li>
<li>Joint Supervisor Appointment (E20)</li>
<li>Supervisor Host Binding (E20)</li>
<li>Per-Call Authorization (E20)</li>
<li>Owner Withdrawal (E20,E24)</li>
<li>Non-Delegable Condition (E22)</li>
</ul>
</section>
<section>
<name>Lifecycle</name>
<ul>
<li>Ephemeral Evaluator (E21,E24)</li>
<li>Persistent Evaluator (E21,E24)</li>
<li>Hybrid Evaluator (E21,E24)</li>
<li>Evaluation Instance Lifetime (E24)</li>
<li>Appointment Lifetime (E24)</li>
<li>Review-Round Lifetime (E17,E21)</li>
<li>Determination Validity (E17,E24)</li>
<li>Evidence Retention (E19,E26)</li>
<li>Cross-Project Authority Isolation (E21,E24)</li>
<li>Runtime Retirement (E19,E24)</li>
<li>Stale Authority Reuse Prohibition (E24)</li>
<li>Context Migration (E14,E24)</li>
<li>Protected Continuation (E01,E16)</li>
<li>Round Closure (E17)</li>
<li>Closure Tombstone (E19)</li>
<li>Authority Retirement (E19)</li>
<li>Non-Reopenable Session (E19)</li>
<li>Determination Staleness (E15,E24)</li>
<li>Appointment Revocation (E24)</li>
<li>Platform Compromise Handling (E24)</li>
</ul>
</section>
<section>
<name>Governance</name>
<ul>
<li>Multi-Organization Governance (E20,E22,E25)</li>
<li>Mandatory Approver (E20)</li>
<li>Principal-Level Quorum (E25)</li>
<li>Independent Control Group (E17,E21)</li>
<li>Mandatory Veto (E06,E25)</li>
<li>Reviewer Independence (E17,E21)</li>
<li>Collusion-Resistant Review (E17)</li>
<li>Membership Epoch (E25)</li>
<li>Organizational Consent (E20,E25)</li>
<li>Threshold Governance (E25)</li>
<li>Eligible Electorate Snapshot (E25)</li>
<li>Joiner Consent (E25)</li>
<li>Old-Epoch Fencing (E25)</li>
<li>Membership Reconstitution (E25)</li>
<li>Principal-Independent Voting (E25)</li>
<li>Credential-Count Inflation (E25)</li>
<li>Reviewer Qualification (E17,E21)</li>
<li>Mandatory Reviewer Completeness (E17)</li>
<li>Dissent Preservation (E17,E21)</li>
<li>Dispute Containment (E21)</li>
</ul>
</section>
<section>
<name>Evidence</name>
<ul>
<li>Evidence Admission (E04,E17)</li>
<li>Source Role (E04,E16)</li>
<li>Evidence Provenance (E16,E17)</li>
<li>Source Authentication (E16,E26)</li>
<li>Outcome Truth (E15,E16)</li>
<li>Native Effect Receipt (E15,E27)</li>
<li>Correlated Evidence (E17)</li>
<li>Source Diversity (E17,E21)</li>
<li>Evidence Commitment (E26)</li>
<li>Selective Disclosure (E26)</li>
<li>Evidence Lineage (E26)</li>
<li>Freshness (E04,E15)</li>
<li>Source-Origin Receipt (E15,E27)</li>
<li>Resource-Origin Evidence (E15,E16)</li>
<li>Evidence Custodian (E26)</li>
<li>Evidence Audience (E26)</li>
<li>Attribute Commitment (E26)</li>
<li>Inclusion Proof (E26)</li>
<li>Contradictory Evidence (E17)</li>
<li>Unresolved Evidence (E17,E21)</li>
</ul>
</section>
<section>
<name>Agent workflow</name>
<ul>
<li>Multi-Agent Negotiation (E01)</li>
<li>Joint Plan Commitment (E01)</li>
<li>Communication Freeze (E02)</li>
<li>Distributed Quiescence (E02)</li>
<li>Protected Next Actor (E07)</li>
<li>Single-Use Next-Move Passport (E11)</li>
<li>Renegotiation (E08)</li>
<li>Task-State Migration (E14)</li>
<li>Pending-Work Escrow (E14)</li>
<li>Plan Revision (E08)</li>
<li>Corrective Patch (E08)</li>
<li>Role Reconfiguration (E09)</li>
<li>Bounded Delegation (E09)</li>
<li>Capability Transition (E09)</li>
<li>Substitute Agent (E13)</li>
<li>Single-Turn Authority (E10)</li>
<li>Automatic Refreezing (E10)</li>
<li>Inter-Phase Hold (E10)</li>
<li>Concurrent Move Arbitration (E07)</li>
<li>Causal Message Containment (E03)</li>
</ul>
</section>
<section>
<name>Recovery</name>
<ul>
<li>Unknown Outcome (E11,E18)</li>
<li>Distributed Partition (E18)</li>
<li>Durable Fencing (E18)</li>
<li>Split-Brain Prevention (E18)</li>
<li>Migration Epoch (E18)</li>
<li>Partial Completion (E27)</li>
<li>Authoritative Reconciliation (E18,E27)</li>
<li>Protected Compensation (E27)</li>
<li>Non-Reopenable Session Closure (E19)</li>
<li>Idempotency Key (E18)</li>
<li>Native-Effect Fingerprint (E18)</li>
<li>Stale Leader Fencing (E18)</li>
<li>Checkpoint Lineage (E18)</li>
<li>Lost Acknowledgment (E18)</li>
<li>Source-Status Inquiry (E18)</li>
<li>Retry Authorization (E18)</li>
<li>Recovery Authorization (E18)</li>
<li>PARTIAL_EFFECT (E27)</li>
<li>Reversible Versus Irreversible Effect (E27)</li>
<li>Successor Activation (E24)</li>
</ul>
</section>
<section>
<name>Native execution</name>
<ul>
<li>Effectuation Boundary (E12)</li>
<li>Finality Sink (E12)</li>
<li>Native Resource Admission (E12,E27)</li>
<li>Protected Local Commit (E27)</li>
<li>Owner-Controlled Effect (E12)</li>
<li>PREPARE/COMMIT (E27)</li>
<li>Predecessor Manifest (E23,E27)</li>
<li>Consequence-Bound Authorization (E12)</li>
<li>Pending Act Reconstruction (E12)</li>
<li>Alternate-Path Closure (E12)</li>
<li>External-Dispatch Outbox (E12)</li>
<li>Exact-State Authorization (E12)</li>
<li>One-Use Permission Consumption (E11,E27)</li>
<li>Composite Completion Record (E27)</li>
<li>Completeness Roster (E27)</li>
<li>Native Predecessor Verification (E27)</li>
<li>Resource-Side Admission (E12)</li>
<li>Point of Effect (E12)</li>
<li>Owner-Specific Commit (E27)</li>
<li>Non-Executable Completion (E27)</li>
</ul>
</section>
<section>
<name>Privacy and cross-domain policy</name>
<ul>
<li>Purpose-Bound Evidence (E26)</li>
<li>Audience-Bound Verification (E26)</li>
<li>Minimal Disclosure (E26)</li>
<li>Trust Compatibility (E22)</li>
<li>Policy-Semantic Compatibility (E22)</li>
<li>Federated Governance (E22)</li>
<li>Sovereign Resource Authority (E22)</li>
<li>Constraint Intersection (E22)</li>
<li>Policy Generation (E22)</li>
<li>Drift Containment (E22)</li>
<li>Metadata Leakage (E26)</li>
<li>Disclosure Audit (E26)</li>
<li>Evidence Redaction (E26)</li>
<li>Source Sequence Anchor (E26)</li>
<li>Joint Authority Formation (E22)</li>
<li>Leaderless Coordination (E23)</li>
<li>Causal Dependency Evidence (E23)</li>
<li>Remote-Status Lease (E23)</li>
<li>Owner-Side Admission (E23)</li>
<li>Local Supervision (E23)</li>
</ul>
</section>
<section>
<name>Risk and simulation</name>
<ul>
<li>Virtual Consequence Reconstruction (E05)</li>
<li>Non-Effective Simulation (E05)</li>
<li>Digital Twin Evaluation (E05)</li>
<li>Tail Exposure (E06)</li>
<li>Benefit-Loss Evaluation (E06)</li>
<li>Cumulative Exposure (E06)</li>
<li>Domain Threshold (E06)</li>
<li>Calibration Reassessment (E06)</li>
<li>Uncertainty-Aware Evaluation (E05)</li>
<li>Counterfactual Operational State (E05)</li>
<li>Risk-Constrained Actor Selection (E04,E07)</li>
<li>Adverse-Outcome Feedback (E06)</li>
</ul>
</section>
<section>
<name>Industry terminology</name>
<ul>
<li>Policy Decision Point (E04,E12)</li>
<li>Policy Enforcement Point (E12)</li>
<li>Workflow Orchestration (E15)</li>
<li>Saga Transaction (E27)</li>
<li>Two-Phase Commit (E27)</li>
<li>Fencing Token (E18)</li>
<li>Byzantine Fault Tolerance (E17,E25)</li>
<li>Remote Attestation (E20)</li>
<li>Workload Identity (E20)</li>
<li>Human-in-the-Loop Approval (E04,E17)</li>
<li>Zero Trust (E12)</li>
<li>Quorum Approval (E25)</li>
<li>Audit Log (E26)</li>
<li>Distributed Tracing (E03)</li>
<li>Agent-to-Agent Protocol (E01)</li>
<li>Model Context Protocol (E12)</li>
<li>Token Exchange (E11)</li>
<li>Proof-of-Possession Token (E11)</li>
<li>Lease (E24)</li>
<li>Idempotent Operation (E18)</li>
</ul>
</section>
<section>
<name>Search Phrases</name>
<t>Boolean phrases (quoted terms in practice); each lists the embodiments.</t>
<ol>
<li>location independent evaluator AND task appointment (E04,E21)</li>
<li>ephemeral evaluator AND protected mandate (E21,E24)</li>
<li>persistent evaluator AND project scoped authority (E21,E24)</li>
<li>hybrid evaluator AND task-specific context (E21,E24)</li>
<li>evaluator determination AND execution authorization (E04,E12)</li>
<li>evaluator decision validity AND appointment lifetime (E17,E24)</li>
<li>neutral evaluator AND administrative independence (E04,E21)</li>
<li>single component evaluator AND independence (E04,E17)</li>
<li>multi agent negotiation AND protected freeze (E01,E02)</li>
<li>joint plan AND non-effective AND evaluator (E01,E04)</li>
<li>distributed quiescence AND evaluation snapshot (E02,E03)</li>
<li>mandatory reviewer AND dissent AND authorization (E17,E21)</li>
<li>independent control groups AND threshold governance (E17,E25)</li>
<li>cross organization supervisor AND appointment (E20)</li>
<li>agent substitution AND pending work escrow (E13,E14)</li>
<li>source authenticated outcome AND independent evaluator (E16,E21)</li>
<li>unknown outcome AND durable fencing (E18)</li>
<li>leaderless coordination AND native effect (E23)</li>
<li>cross domain policy semantic compatibility (E22)</li>
<li>resource owner AND predecessor receipt (E23,E27)</li>
<li>partial effect AND authorized compensation (E27)</li>
<li>non executable determination AND native authorization (E04,E12)</li>
<li>purpose bound selective disclosure AND outcome evidence (E26)</li>
<li>protected session closure AND domain seal (E19)</li>
<li>membership reconstitution AND owner veto (E25)</li>
<li>evaluation authority AND effectuation authority (E04,E12)</li>
<li>evaluator AND appointment AND revocation AND lease (E24)</li>
<li>review round AND closure AND dissent (E17)</li>
<li>evidence admission AND source role AND task binding (E04,E17)</li>
<li>correlated evidence AND source independence (E17)</li>
<li>agent self-appointment AND evaluator (E20)</li>
<li>mandatory owner AND consent AND supervisor (E20)</li>
<li>host instance binding AND supervisor installation (E20)</li>
<li>next-move passport AND single use AND actor bound (E11)</li>
<li>bearer token AND non-bearer AND continuation (E11)</li>
<li>single turn authority AND refreezing (E10)</li>
<li>selective communication resumption AND hold (E10)</li>
<li>inter-phase hold AND evaluator (E10)</li>
<li>corrective patch AND committed plan AND recommitment (E08)</li>
<li>role reconfiguration AND capability generation (E09)</li>
<li>neutral substitute AND credential inheritance (E13)</li>
<li>pending work escrow AND source fencing (E14)</li>
<li>destination acceptance AND task migration (E14)</li>
<li>completion receipt AND phase advancement (E15)</li>
<li>actual outcome AND feedback AND continuation (E16)</li>
<li>untrusted feedback AND instruction containment (E16)</li>
<li>lost acknowledgment AND resource authoritative status (E18)</li>
<li>stale leader AND resource side fencing (E18)</li>
<li>checkpoint lineage AND migrated agent (E18)</li>
<li>session termination AND native obligation inventory (E19)</li>
<li>non-reopenable tombstone AND session (E19)</li>
<li>independent outcome evaluation AND cross organization (E21)</li>
<li>dispute appeal AND new round isolation (E21)</li>
<li>policy drift AND owner drift AND containment (E22)</li>
<li>constraint intersection AND joint authority (E22)</li>
<li>causal receipt AND remote status lease (E23)</li>
<li>predecessor manifest AND owner installed (E23,E27)</li>
<li>supervisor lease AND revocation convergence (E24)</li>
<li>successor activation AND partition (E24)</li>
<li>monotonic revocation serial AND tombstone (E24)</li>
<li>principal independent voting AND agent count (E25)</li>
<li>membership epoch AND old epoch fencing (E25)</li>
<li>selective evidence commitment AND source authenticated (E26)</li>
<li>audience bound release AND confidential outcome (E26)</li>
<li>minimal retention AND audit continuity (E26)</li>
<li>prepare commit AND native domain gate (E27)</li>
<li>composite completion AND no composite authority (E27)</li>
<li>compensation AND new native effect (E27)</li>
<li>partial composite effect AND controlled resume (E27)</li>
<li>virtual consequence AND non-effective scenario (E05)</li>
<li>simulation AND not proof of execution (E05)</li>
<li>tail exposure AND domain threshold AND veto (E06)</li>
<li>risk benefit loss AND organization policy (E06)</li>
<li>next actor selection AND bounded instruction (E07)</li>
<li>concurrent next moves AND arbitration (E07)</li>
<li>communication edge AND pending work AND freeze (E02)</li>
<li>late operation disposition AND admission fencing (E02)</li>
<li>causal message containment AND snapshot validity (E03)</li>
<li>negotiation state integrity AND evaluator (E03)</li>
<li>finality sink AND evaluator determination (E12)</li>
<li>alternate path closure AND external dispatch outbox (E12)</li>
<li>pending act reconstruction AND exact state (E12)</li>
<li>evaluator integrity AND provenance AND review (E17)</li>
<li>reviewer qualification AND control group (E17,E21)</li>
<li>evidence retention AND determination validity (E19,E24)</li>
<li>cross project isolation AND evaluator (E21,E24)</li>
<li>platform compromise AND evaluator contexts (E24)</li>
<li>context migration AND revalidation (E14,E24)</li>
<li>agent communication AND evaluation freeze AND quiescent (E02)</li>
<li>policy semantic mismatch AND refusal (E22)</li>
<li>non-delegable condition AND joint authority (E22)</li>
<li>owner local admission AND convergence evidence (E23)</li>
<li>dependency spoofing AND resource owner (E23)</li>
<li>supervisor revocation AND disconnected domain (E24)</li>
<li>bounded freshness AND revocation (E24)</li>
<li>eligible electorate AND snapshot (E25)</li>
<li>joiner consent AND mandatory veto (E25)</li>
<li>metadata leakage AND evidence exchange (E26)</li>
<li>inclusion proof AND source commitment (E26)</li>
<li>irreversible effect AND compensation record (E27)</li>
<li>native predecessor effect AND manifest (E27)</li>
<li>UNKNOWN AND not retried AND status inquiry (E18,E27)</li>
<li>partial effect AND honest reconciliation (E27)</li>
<li>HOLD AND UNKNOWN AND DISPUTED AND evaluator (E04,E17,E21)</li>
<li>agent AND tool AND evaluator AND approval AND gate (E04,E12)</li>
<li>multi-agent AND cross-organizational AND evaluator (E01,E21)</li>
<li>agent protocol AND evaluator appointment (E20,E21)</li>
<li>policy decision point AND evaluator lifecycle (E04,E24)</li>
<li>workflow engine AND review round AND closure (E15,E17)</li>
<li>saga AND compensation AND owner authorization (E27)</li>
<li>fencing token AND supervisor generation (E18,E24)</li>
<li>quorum AND organization AND agent count (E25)</li>
<li>human reviewer AND committee AND evaluator (E04,E17)</li>
<li>agent evaluation framework AND appointment (E04,E21)</li>
<li>audit log AND selective disclosure AND audience (E26)</li>
</ol>
</section>
<section>
<name>Boolean Search Strings</name>
<ul>
<li>("evaluator" AND "appointment") AND ("task" OR "mandate") AND ("agent")</li>
<li>("determination" AND "not executable") AND ("native" OR "resource owner")</li>
<li>("review round") AND ("dissent") AND ("control group" OR "independent")</li>
<li>("freeze" AND "quiescence") AND ("agent" OR "agents") AND ("evaluation")</li>
<li>("unknown outcome" OR "lost acknowledgment") AND ("fencing") AND ("recovery")</li>
<li>("supervisor" AND "lease") AND ("revocation") AND ("successor")</li>
<li>("prepare" AND "commit") AND ("native domain") AND ("partial effect")</li>
<li>("selective disclosure") AND ("purpose") AND ("audience") AND ("outcome evidence")</li>
</ul>
</section>
</section>
<section anchor="app-f">
<name>Appendix F: Industry Integration Mapping</name>
<t>See <xref target="industry"/>. The mapping is hypothetical, non-endorsed, AI-generated and unverified against vendor documentation; cross-check against the original filings before drawing any conclusion. It is not an endorsement, validation or infringement statement.</t>
<dl newline="true">
<dt>Gate placement</dt>
<dd><t>Where a tool broker, API gateway, payment rail adapter, network controller or actuator interface already enforces access, a gate that additionally requires a current owner-issued authorization could sit there.</t></dd>
<dt>What is checked</dt>
<dd><t>appointment, round, determination validity, policy generation, actor, operation and single-use state.</t></dd>
<dt>What is rejected</dt>
<dd><t>stale, replayed, wrong-actor or determination-only requests.</t></dd>
<dt>Why it matters</dt>
<dd><t>a model or reviewer approval cannot itself cause an effect.</t></dd>
</dl>
</section>
<section anchor="app-g">
<name>Appendix G: Patent and Research Cross-References</name>
<t>See <xref target="ipr"/>, <xref target="IN-131650"/>, <xref target="PCT-055615"/>, <xref target="TDC-12542"/>. Related drafts: <xref target="REL-FOUND"/>, <xref target="REL-CBEA"/>, <xref target="REL-CAUSAL"/>, <xref target="REL-IEV"/>. This document is distinguishable from them: causal diagnosis locates drift; execution finality governs admission of an effect; this document governs evaluator appointment, evidence eligibility, lifecycles, typed determinations and cross-domain evaluation governance, and uses E12, E15, E18 and E27 only in relation to those existing drafts.</t>
</section>
<section anchor="app-h">
<name>Appendix H: Source Verification and Limitations</name>
<section>
<name>Source Verification</name>
<ul>
<li>Source: 1,905-page PDF; page and embodiment map built from the extracted text; independent claims 1-60 and dependent ranges checked for gaps.</li>
<li>Filing date: the screenshot of the IP India eFiling paid-forms page shows reference 202631131650, C.B.R. No. 16058, CBR date 03/11/2026. This corroborates the applicant-reported date but is a payment record, not the official acknowledgment, which the editor has not seen.</li>
<li>The source file name says 'NOV 4'; it contains a digital-signature date of 3 November; no 4 November filing is inferred.</li>
<li>The Technical_Disclosure_1.json file was not available to the editor.</li>
</ul>
</section>
<section>
<name>Quality Checklist</name>
<ol>
<li>Actual source is the 1,905-page evaluator specification: partly verified (page count and embodiment map checked; read selectively).</li>
<li>All 27 embodiments represented: yes.</li>
<li>60 independent claims mapped: yes, titles checked against the filing.</li>
<li>Dependent-claim index covers 61-540: yes, by embodiment range.</li>
<li>Three deployment types preserved: yes.</li>
<li>Six lifecycles explained: yes.</li>
<li>Location independence not confused with control independence: yes.</li>
<li>Single-component evaluators not excluded: yes.</li>
<li>Neutrality not assumed from third-party hosting: yes.</li>
<li>Owner-derived appointment vs self-appointment: yes.</li>
<li>Vetoes, quorum, independence policy-specific: yes.</li>
<li>Control groups not account counts: yes.</li>
<li>Mandatory dissent preserved: yes.</li>
<li>Origin vs outcome truth: yes.</li>
<li>Evaluator decision vs owner authorization: yes.</li>
<li>Native admission independently enforced: yes.</li>
<li>Freeze accounts for in-flight operations: yes.</li>
<li>Simulation not proof of execution: yes.</li>
<li>Replacement agents inherit no stale authority: yes.</li>
<li>Unknown outcomes not retried: yes.</li>
<li>Source-side fencing and revocation modeled: yes.</li>
<li>Membership epoch/owner conditions: yes.</li>
<li>Leaderless cooperation not global authority: yes.</li>
<li>Partial effects honest: yes.</li>
<li>Compensation separately authorized: yes.</li>
<li>Closure needs native obligation reconciliation: yes.</li>
<li>Privacy-sensitive evidence purpose/audience bound: yes.</li>
<li>Cryptographic vs protected-local distinguished: yes.</li>
<li>Post-quantum optional: yes.</li>
<li>IETF and industry references verified: NO, partly unverified (see Appendix H).</li>
<li>Existing RFC functions and prior art credited: yes.</li>
<li>No invented performance figures: yes.</li>
<li>No company adoption inferred: yes.</li>
<li>IETF IPR requirements addressed: partly; formal disclosure pending.</li>
<li>Source and filing metadata reconciled: partly; acknowledgment not seen.</li>
<li>Every diagram technically meaningful: reviewed by editor.</li>
<li>Security, Privacy, IANA, References complete: yes.</li>
<li>Search terms cover inventor and industry language: yes.</li>
<li>Distinguishable from existing DAS drafts: yes.</li>
<li>Understandable without patent documentation: intended; reader feedback needed.</li>
</ol>
</section>
</section>
</back>
</rfc>
