<?xml version="1.0" encoding="utf-8"?>
<rfc xmlns:xi="http://www.w3.org/2001/XInclude" version="3"
     docName="draft-jackson-wimse-evaluation-00" category="info"
     ipr="trust200902" submissionType="IETF" xml:lang="en">
  <front>
    <title abbrev="WIMSE Verifier Evaluation Semantics">Verifier-Side Evaluation Semantics for Delegated Authority Chains</title>
    <author fullname="Wes Jackson">
      <address>
        <email>c.wesjackson@gmail.com</email>
      </address>
    </author>
    <date day="21" month="September" year="2026"/>
    <area>sec</area>
    <workgroup>wimse</workgroup>
    <keyword>workload identity</keyword>
    <keyword>delegation</keyword>
    <keyword>verifier</keyword>
    <keyword>evaluation</keyword>
    <abstract>
      <t>Delegation chain specifications describe the shape of conveyed
      authority. They leave the verifier's half of the exchange
      underdetermined. Two verifiers can check the same chain, both report
      success, and enforce different policy. This document states what a
      verifier MUST do: the explicit inputs evaluation depends on, and four
      rules that keep evaluation fail-closed. The rules are drawn from
      shipped specification text (<xref target="GAL"/> 0.2.6-draft,
      <xref target="PTC"/> 0.2.5-draft) and a
      public reference implementation.</t>
    </abstract>
  </front>
  <middle>
    <section anchor="sec-intro" numbered="true" toc="default">
      <name>Introduction</name>
      <t>A delegation chain is a claim about authority. A verifier turns that
      claim into a decision. Chain formats standardize the claim. This
      document standardizes the decision.</t>
      <t>The failure this prevents is silent divergence. A verifier that drops
      an authorization entry it cannot parse and still approves has enforced
      nothing while reporting success. A verifier that derives "now" from the
      records it is judging keeps expired authority alive during quiet
      periods. A verifier that lets one broken path poison a valid one hands
      any attacker a veto over legitimate authority. Each of these was
      observed in running code during WIMSE implementation work. Each is
      fixed here with a stated rule.</t>
      <t>This document does not define a chain format, a token envelope, or a
      policy language. It defines the evaluation function those formats are
      judged by. It is complementary to chain-format documents such as
      <xref target="DELEGATION-CHAIN"/> and <xref target="CONNECTED-FLIGHT"/>:
      those define what is conveyed, this defines how the receiver judges
      it.</t>
      <t>The rules in <xref target="sec-rules"/> are implemented in the public
      reference implementation <xref target="RI"/>. The key-bootstrapping
      mechanism discussed in <xref target="sec-security"/> is the known open
      item; it is named as a gap rather than specified.</t>
    </section>
    <section anchor="sec-terminology" numbered="true" toc="default">
      <name>Terminology</name>
      <t>The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
      "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and
      "OPTIONAL" in this document are to be interpreted as described in
      BCP 14 <xref target="RFC2119"/> <xref target="RFC8174"/> when, and only
      when, they appear in all capitals, as shown here.</t>
      <t><em>Frozen call:</em> the exact action a human was shown and
      approved: capability, arguments, recipient, and principal, as presented
      at approval time.</t>
      <t><em>Evaluation instant:</em> the instant against which validity,
      including expiry, is judged.</t>
      <t><em>On-behalf-of principal:</em> the principal the chain acts under,
      distinct from the agent acting at each hop.</t>
      <t><em>Principal:</em> the identity tuple a grant is issued to: agent
      identity, skill, user, and trust tier. The unit of authority is the
      (principal, action class) pair. The five attributes together name who
      acts, in what capacity, on whose behalf, at what trust, and over what
      class of action. The semantics of the trust tier are
      deployment-defined; this document standardizes the tier's position in
      the tuple, not its values.</t>
      <t><em>Issuer key:</em> the key that signs authority-raising
      records.</t>
      <t><em>Evaluator key:</em> the key that signs authority-lowering
      records, held by a deliberately separate identity.</t>
      <t><em>Delegation path:</em> one chain of delegations from a root of
      trust to the acting agent.</t>
    </section>
    <section anchor="sec-inputs" numbered="true" toc="default">
      <name>Evaluation Inputs</name>
      <t>A verifier's decision MUST be a function of explicit inputs. Four
      inputs are required. Anything the verifier needs that is not in this
      list MUST be supplied explicitly; the verifier MUST NOT guess.</t>
      <section anchor="sec-frozen-call" numbered="true" toc="default">
        <name>The Frozen Call</name>
        <t>The frozen call is the exact action the human was shown and
        approved. Evaluation and consumption bind to that call as presented,
        not to a re-serialization, a paraphrase, or an identifier that names
        it. (GAL-36, GAL Section 4.1)</t>
      </section>
      <section anchor="sec-instant" numbered="true" toc="default">
        <name>The Evaluation Instant</name>
        <t>The instant against which validity, including expiry, is judged is
        an explicit input to evaluation. It MUST NOT be derived from the
        timestamps of the records under evaluation. Deriving "now" from the
        latest record keeps expired authority alive through quiet periods,
        because time passes without producing records. (GAL-34)</t>
      </section>
      <section anchor="sec-principal" numbered="true" toc="default">
        <name>The On-Behalf-Of Principal</name>
        <t>Conveyed evaluation material MUST include the on-behalf-of
        principal. A verifier that does not share the issuer's log cannot
        recover the principal from the action chain, so material that omits
        it authenticates an action without saying whom it was taken for.
        (<xref target="PTC"/> Section 6.6)</t>
        <t>Evaluation is against the whole tuple, never the agent identity
        alone: a verifier that matches on agent identity while ignoring
        skill, user, or tier has evaluated a different grant. (GAL-36, <xref target="GAL"/>
        Section 5.1)</t>
      </section>
      <section anchor="sec-keys" numbered="true" toc="default">
        <name>Issuer and Evaluator Keys</name>
        <t>Authority-raising records (issuance, promotion, tightening) are
        signed by the issuer key. Authority-lowering records (demotion,
        lapse) are written by a deliberately separate evaluator identity and
        signed by the evaluator key. No identity SHALL hold both roles'
        signing keys. The separation is structural, not conventional: a
        grant-store write the agent can reach is a promotion bypass.
        (GAL-37, <xref target="GAL"/> Section 6.7.2, <xref target="GAL"/> Section 6.10)</t>
      </section>
    </section>
    <section anchor="sec-rules" numbered="true" toc="default">
      <name>Evaluation Rules</name>
      <section anchor="sec-every-entry" numbered="true" toc="default">
        <name>Process Every Entry or Reject</name>
        <t>A verifier MUST process every authorization_details entry
        <xref target="RFC9396"/> presented or reject the token. An entry of a type the verifier cannot
        evaluate is an entry it cannot process, and the token MUST be
        rejected. A verifier that silently drops an entry enforces an
        incomplete policy while reporting success. That is fail-open.
        (Aligned with the -02 direction of <xref target="DELEGATION-CHAIN"/>.
        A verifier advertises the types it supports with
        authorization_details_types_supported <xref target="RFC9728"/>.)</t>
      </section>
      <section anchor="sec-fail-closed-path" numbered="true" toc="default">
        <name>Fail Closed Per Path</name>
        <t>Where several delegation paths reach the same agent, the verifier
        MUST evaluate each path independently along its whole length. One
        fully valid path suffices for authority. A broken path MUST NOT
        neutralize a valid one, and a valid path MUST NOT excuse a broken
        one. Discovering which paths exist is the delegation mechanism's
        work; judging each presented path is the verifier's. (<xref target="GAL"/> 0.2.5
        changelog: lifecycle, which the grant standard owns, separated from
        path resolution, which it does not.)</t>
      </section>
      <section anchor="sec-consumption" numbered="true" toc="default">
        <name>Consumption Keyed to the Frozen Call</name>
        <t>An approval MUST be consumed only by release or rejection of the
        frozen call it binds, by the whole principal the call was frozen
        for. Consumption MUST NOT be inferred from the approval's identifier
        appearing in any record. Otherwise any party able to write a record
        can burn another party's approval without using it. The release
        executes the stored call and nothing re-sent. (GAL-36, <xref target="GAL"/>
        Section 4.1)</t>
        <t>This document does not standardize approval interaction. It
        standardizes what the verifier must check about consumption.</t>
      </section>
      <section anchor="sec-role-keys" numbered="true" toc="default">
        <name>Verify Records Under Role-Separated Keys</name>
        <t>A verifier MUST select the acceptable signing keys from the record
        type, MUST refuse a record signed by the other role's key, and MUST
        refuse a key resolvable under both roles. The key that grants
        authority MUST NOT be the key that judges it. (GAL-37)</t>
      </section>
    </section>
    <section anchor="sec-security" numbered="true" toc="default">
      <name>Security Considerations</name>
      <t>Clock skew: the evaluation instant is supplied by the evaluating
      party as an explicit input. When two parties do not share a clock, the
      party evaluating supplies the instant. The clock source and the skew
      bound are deployment details, and stating them is REQUIRED: a
      deployment MUST document which clock supplies the evaluation instant
      and the maximum skew it tolerates. This document sets no default
      bound.</t>
      <t>Key bootstrapping across boundaries: a receiving verifier needs the
      sending boundary's issuer key before any rule in <xref target="sec-rules"/>
      can run.</t>
      <t>Open issue: the mechanism by which the verifier obtains the sending
      boundary's issuer key is not specified in this document. The reference
      implementation resolves keys out of band. This document requires only
      that the verifier possess the key before any rule in
      <xref target="sec-rules"/> runs; the provisioning mechanism is left
      for the working group. This is the open question put to the WIMSE list
      alongside this draft.</t>
      <t>Structural separation: the agent MUST NOT hold signing keys and MUST
      have no write access to the grant store. Verification authenticates
      lineage; it does not clean taint and does not substitute for the
      receiver's own trust map. (<xref target="GAL"/> Section 6.7.2, <xref target="PTC"/> Section 6.6,
      <xref target="PTC"/> Section 6.7)</t>
      <t>Misconfigured verification, such as a configured but unusable key
      source, MUST fail closed loudly and MUST NOT silently degrade to no
      verification. (<xref target="PTC"/> Section 6.7)</t>
    </section>
    <section anchor="sec-iana" numbered="true" toc="default">
      <name>IANA Considerations</name>
      <t>This document makes no request of IANA.</t>
    </section>
  </middle>
  <back>
    <references>
      <name>References</name>
      <references>
        <name>Normative References</name>
        <reference anchor="RFC2119">
          <front>
            <title>Key words for use in RFCs to Indicate Requirement Levels</title>
            <author fullname="S. Bradner"/>
            <date month="March" year="1997"/>
          </front>
          <seriesInfo name="BCP" value="14"/>
          <seriesInfo name="RFC" value="2119"/>
          <seriesInfo name="DOI" value="10.17487/RFC2119"/>
        </reference>
        <reference anchor="RFC8174">
          <front>
            <title>Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words</title>
            <author fullname="B. Leiba"/>
            <date month="May" year="2017"/>
          </front>
          <seriesInfo name="BCP" value="14"/>
          <seriesInfo name="RFC" value="8174"/>
          <seriesInfo name="DOI" value="10.17487/RFC8174"/>
        </reference>
        <reference anchor="RFC9396">
          <front>
            <title>OAuth 2.0 Rich Authorization Requests</title>
            <author fullname="T. Lodderstedt"/>
            <author fullname="J. Richer"/>
            <author fullname="B. Campbell"/>
            <date month="September" year="2023"/>
          </front>
          <seriesInfo name="RFC" value="9396"/>
          <seriesInfo name="DOI" value="10.17487/RFC9396"/>
        </reference>
        <reference anchor="RFC9728">
          <front>
            <title>OAuth 2.0 Protected Resource Metadata</title>
            <author fullname="M. Jones"/>
            <author fullname="T. Lodderstedt"/>
            <date month="October" year="2024"/>
          </front>
          <seriesInfo name="RFC" value="9728"/>
          <seriesInfo name="DOI" value="10.17487/RFC9728"/>
        </reference>
      </references>
      <references>
        <name>Informative References</name>
        <reference anchor="GAL" target="https://github.com/wjatx/ptc-gal-standards/blob/main/GAL-SPEC.md">
          <front>
            <title>Grant Authority Ledger (GAL) Specification</title>
            <author fullname="W. Jackson"/>
            <date year="2026"/>
          </front>
          <refcontent>version 0.2.6-draft</refcontent>
        </reference>
        <reference anchor="PTC" target="https://github.com/wjatx/ptc-gal-standards/blob/main/PTC-SPEC.md">
          <front>
            <title>Policy Trust Chain (PTC) Specification</title>
            <author fullname="W. Jackson"/>
            <date year="2026"/>
          </front>
          <refcontent>version 0.2.5-draft</refcontent>
        </reference>
        <reference anchor="RI" target="https://github.com/wjatx/ptc-gal-reference">
          <front>
            <title>ptc-gal-reference</title>
            <author>
              <organization>wjatx</organization>
            </author>
          </front>
          <refcontent>Public reference implementation of GAL and PTC; the code this document's rules are drawn from</refcontent>
        </reference>
        <reference anchor="DELEGATION-CHAIN">
          <front>
            <title>Verifiable Attenuated Delegation Chains for AI Agents</title>
            <author fullname="R. Asor"/>
          </front>
          <seriesInfo name="Internet-Draft" value="draft-asor-wimse-agent-delegation-chain-01"/>
        </reference>
        <reference anchor="CONNECTED-FLIGHT">
          <front>
            <title>Connected Flight: A Model for Cross-Tenant Delegation</title>
            <author fullname="E. Seymour"/>
          </front>
          <seriesInfo name="Internet-Draft" value="draft-seymour-wimse-connected-flight-00"/>
        </reference>
      </references>
    </references>
    <section numbered="false" toc="exclude">
      <name>Author's Address</name>
      <t>Wes Jackson</t>
      <t>Email: c.wesjackson@gmail.com</t>
    </section>
  </back>
</rfc>
