AAuth: the substrate that grew the object
Evidence for the fundamental-versus-accidental split arrived from outside this family. AAuth is Dick Hardt’s proposed agent-native authorization protocol, an active individual draft (at revision -11, September 25, 2026, and renamed draft-hardt-oauth-aauth-protocol along the way) that departs from OAuth’s redirect model rather than extending it: conversational, signed-request-first, with the Person Server as its governing party. It set out to rebuild agent authorization from a clean sheet, and in an early 2026 revision it grew the object this handbook argues for: a first-class mission layer, with a mission proposal and approval flow, an AAuth-Mission header on the wire (by -11 the reference travels as the mission_s256 claim or parameter, paired with the approving Person Server), mission-aware token choreography, and Person Server governance endpoints.
This author mapped the Mission model onto AAuth before AAuth had a mission layer. AAuth’s author read that work, and the next revision adopted the concept. The handbook therefore counts AAuth as adoption and does not claim independent replication. A protocol designer rebuilding agent authorization from a clean sheet, with his own architecture and free to close the gap another way, built a mission reference and a mission lifecycle into his protocol’s core, and that is what the adoption is evidence for. AAuth made a different design choice on committed authority, subset derivation, and successor-on-widen, and the handbook reads those choices as a different design rather than as evidence against its own. AAuth’s clarification chat, where an approver questions a proposal before consenting, matches the interrogation channel that the approval part records in Consent Evidence.
The analysis works through where AAuth stands. AAuth answered where the mission layer lives. Its mission approval pre-authorizes no portable set of resource operations, and it defines no subset relation between successive or downstream scopes, so each resource decides in its own vocabulary what the mission’s work may do. The five laws are the yardstick to hold its model to as it matures, and they ask for portable containment and lifecycle completeness rather than mission correlation and governance hooks. The family’s AAuth binding carries the kernel that What Survives Without OAuth names at the Person Server with no new AAuth wire members, so the two proposals converge instead of forking. Its gating depends on access mode. The Person Server gates person-token issuance, PS authorization, and federated authorization, and the binding forbids claiming PS issuance gating for agent-identity and resource-managed access, where the agent calls the resource directly. The agent’s Supervisor, the person by default or a supervision server configured for the agent, reviews each proposal through the Person Server. An accepted AAuth update can broaden the work under the same mission reference, and a configured supervision server can accept one without a fresh human decision, so the binding offers no containment guarantee. This family keeps Narrowing per Mission by making every widening a successor with its own approval and authority_hash, and the predecessor terminates as superseded. AAuth keeps the reference and grows its approved version, which is where the two models diverge on the Narrowing law. The supervision behavior rests on an AAuth supervision draft that is not yet submitted, and the binding’s native auth token is Mission-referenced, not Mission-bound. The deep treatments are published: Mission Architecture on AAuth mapped the model onto the protocol before it had a mission layer, AAuth Now Has a Mission Layer re-ran the comparison after the revision, and Why Mission-Bound OAuth Might Be the Wrong Answer is the standing critique of the OAuth home itself.
With AAuth in view, the strategic position stays the same. OAuth is the adoption path because it is where the deployments are, AAuth may prove the cleaner native substrate for the agents that come next, and the model, the laws, and the gate are built to survive either outcome. Where this could be wrong gives the evidence that would falsify that position.
Where this could be wrong
An earlier post in this work’s ancestry asked openly whether Mission-Bound OAuth was the wrong answer. The handbook applies the same scrutiny to the six bets in its design most worth doubting, each with the evidence that would falsify it.
The admission-time bet. The model moves interpretation to admission, where an accountable authority, a human or a policy a human consented to, decides against committed inputs before any authority exists (who may approve). If agent work proves more emergent than that, if real tasks discover most of their authority mid-flight, then the discovery loop becomes the hot path instead of the exception, admission becomes the bottleneck no matter who staffs it, and deployments will choose between rubber-stamped ceilings and friction that drives them back to broad grants. Progressive authorization and policy adjudication within a human-consented ceiling are the hedges, and they are young because nobody yet knows the sustainable grain. The price per admission is falling. An approved expansion no longer costs a restart, because a harness may rebind the running work to the successor Mission under freshly derived credentials. The falsifying evidence: expansion and exception rates that stay high after templates and organizational priors mature. A quieter falsifier matters because rubber-stamping improves those metrics while losing the bet. It is approval telemetry showing that approvers stopped deciding: median time-to-approve collapsing toward the four-second reflex, disclosure interaction going to zero, and narrowing rates falling while approved breadth grows.
The issuer bet. The family puts the Mission at an issuer, the Authorization Server, the standalone MAS, or AAuth’s Person Server, because that is where approval, derivation, and revocation already live. But the work itself lives in harnesses and orchestrators, and the industry could consolidate governance there instead, with the runtime platform as the source of truth for the task and the identity stack reduced to credentials. The MAS binding hedges the deployment topology, not the ownership question. The falsifying evidence: harness vendors shipping proprietary task objects that win adoption without ever touching the token layer.
The revocation bet. The laws price Termination as non-negotiable, and much of the family’s weight, state-gated issuance, Status, the freshness bounds, the offline chain’s state check, is the cost of making revocation reach everything. A capability-native world could decide that short expiry with no refresh is enough, accept the bounded staleness, and skip the central state dependency entirely, trading the kill switch for autonomy and offline verification. If the market accepts that trade at scale, the category as this handbook defines it loses its fourth law to a cheaper approximation. The falsifying evidence: serious deployments running attenuable tokens with no state source and eating the staleness without incident.
The composition bet. The family bets that the approved task must be a first-class object. The alternative is composition, in which policy engines, workflow state, and richer token claims, wired together carefully, might deliver bounded, revocable, attributable agent work with no new object at all. Inside one administrative domain, a conventional stack does implement durable task state, fan-out joins, persistent narrowing, and audit locally, and the Architecture does not claim otherwise. The family’s answer is that composition without a shared root loses those properties where they cross bindings and trust domains: cross-audience revocation, cross-hop audit join, and integrity anchors every party reads the same way (what becomes possible only with a Mission), but that is an argument, not a deployment result. The falsifying evidence: production estates achieving task-bounded, task-revocable agent authority through composition alone, interoperably, without converging on a shared task object.
The portability bet. The family bets that Mission authority can travel, meaning that the Authority Set, the subset rule, and Mission state can be projected across authorization domains without losing their meaning. Cross-domain reality is harsher. The downstream domain may not speak the parent’s authority vocabulary, may not be able to verify the upstream approval’s semantics, may disagree about what counts as narrower, and may honor an action that is locally permitted and globally outside the Mission. Revocation propagation across administrative boundaries is the operational form of the same problem. Cross-Domain Projection and the portable Mandate are the constructive answers, and conservative refusal is the fallback where translation cannot be trusted. The falsifying evidence: cross-domain deployments that abandon authority translation and fall back to local re-approval at every boundary, making the Mission a per-domain object after all.
The classification bet. The runtime gate is priced by the consequential/non-consequential line, and that line is deployment policy with only an extreme-end floor. The bet is that deployments hold the line under latency and cost pressure. The quiet failure is reclassification, in which consequential classes drift below the gate because the PDP round-trip hurts while the claim name never changes. The falsifying evidence: enforcement-scope statements whose mediated class list shrinks release over release while the claim name holds, the classification twin of the admission bet’s rubber-stamp telemetry.
None of these is a reason to wait. The issuance-only floor, Baseline Issuance, needs the OAuth binding and published RFCs and runs on Resource Servers that need not be Mission-aware. A scope-only Resource Server operates within the Mission only where the AS establishes a safe scope projection for it, and the AS refuses issuance to it otherwise. The remaining dependency risk sits in the drafts the Runtime-Enforced bundle adds, which the standards map names. None of that risk blocks the floor, and the laws are cheap to hold even if the bindings move. A reader deciding how hard to bet should know which parts are invariants and which are wagers. The laws and the claim gate are the invariants. The admission grain, the issuer home, the price of Termination, the necessity of the object itself, the portability of its authority, and the classification line are the wagers, and deployment experience will settle them.
Where the framework leads
The estate picks the entry point. Under the OAuth binding the adoption order runs from the issuance-only floor to the Runtime-Enforced bundle where an action class needs a per-action, parameter-bound decision, and then to the Governed Agent and High-Assurance Agent bundles where unattended operation, delegation, or the high-consequence classes call for them. A standalone Mission Authority Server is a peer entry point with its own order: records first, enforcement as PEP coverage arrives, and issuance gating through the issuance grant. A deployment adopts the bundle its risk warrants and stops there, and a relying party compares a deployment’s assurance claims, not its level. Adopting Mission-Bound Authorization carries that build order, the ecosystem to compose with, the operational surfaces you will own, and the pieces the community still has to standardize.
The framework makes the laws portable. They are stated once and realized per substrate, and every competing proposal, this family’s own bindings included, is held to the same five. If a law is wrong, a wager is mispriced, or a binding is missing, raise it in the issues on the draft repository.