This family is the disciplined expansion lane of the corpus. It matters because public-account permissions and broad wallet delegation are poorly shaped for machines and agents. A robot, API client, research agent, or workflow assistant usually does not need unrestricted wallet power. It needs a bounded right: one access window, one compute budget, one task escrow, one relay allowance, one fee envelope, or one useful-work claim path.
The Core Thesis
The agentic and machine papers both keep returning to one architectural move: replace reusable account power with spendable capability objects.
| Object family | What it carries |
|---|---|
| Spending envelope | Task-scoped budget and fee capacity without full wallet control |
| Capability or service right | Access to data, compute, APIs, relays, rooms, routes, or other bounded services |
| Reward or claim object | A private payout or task result that becomes redeemable only after defined proof or attestation |
The same settlement logic still applies. Local use or acceptance may happen first. Canonical authority still arrives later through checkpointed reconciliation.
Why Public Permission Models Are Weak Here
Public systems often solve machine or agent commerce by delegation over an account, smart account, API key, or session key. The corpus argues that this creates two problems:
- too much authority is delegated if the agent or device is compromised;
- too much strategy becomes visible if the payment or permission graph is public.
Z00Z is a better fit when the right itself can carry the bound: provider scope, amount limit, expiry, one-time semantics, task category, or fee ceiling.
What FeeEnvelope Changes
Machine and agent rights only make sense if the system also explains who pays for later publication and settlement. That is why Agentic Offline Economy Whitepaper treats FeeEnvelope as a separate primitive.
| Right answers… | FeeEnvelope answers… |
|---|---|
| What action is allowed | Who pays for processing it |
| Which provider or policy scope applies | Which asset, sponsor, or limit funds the processing path |
| How much authority the holder has | How the later settlement lane remains operationally viable |
Without that split, the system falls back toward broad wallet control, which is exactly what this family is trying to avoid.
Representative Scenario Groups
| Scenario group | Why it fits |
|---|---|
| Offline machine economy | A device can present a bounded local right and reconcile later instead of waiting for live chain settlement |
| API, compute, or data credits | A user or agent can consume bounded service privately without creating a reusable public billing graph |
| Agent-to-agent commerce | One agent can buy or fund a task without receiving unrestricted account authority |
| Useful-work and payout claims | A reward can stay private until proof, authorization, and anti-replay conditions are satisfied |
These groups differ commercially, but the architecture is the same: a bounded private right becomes the economic primitive.
Current Versus Target Posture
| Surface | Current posture | Target direction |
|---|---|---|
| Rights-first conceptual fit | Strong in the current corpus | More object families can still land over time |
| Bounded agent and machine semantics | Clear and useful as a design language | Operational standards, providers, and wallet tooling still need to mature |
| Full multi-agent or machine ecosystem | Not a blanket live claim | Broader integrations, provider acceptance, and deployment evidence remain future-sensitive |
| Useful-work reward linkage | Coherent through the reward papers | Richer evaluator markets and large-scale adoption remain research-heavy |
This is why the use-cases paper places the family last. It is important, but it should widen the thesis only after the reader already accepts the cash and rights wedge.
What This Family Must Not Be Retold As
The corpus is especially sensitive to overclaiming here.
| Unsafe retelling | Safer retelling |
|---|---|
| “Z00Z is already a fully general hidden agent economy.” | “The corpus describes bounded private rights that can support agent and machine commerce as an expansion direction.” |
| “Agents get wallets.” | “Agents get scoped spending envelopes, capability rights, and fee support rather than broad wallet authority.” |
| “This proves a private VM.” | “This proves the usefulness of bounded rights and later settlement for service and machine workflows.” |
Those corrections keep the family strategically interesting without letting it become science fiction.
Agent Capability Wallet
Agents need rights rather than ordinary wallets because broad wallet authority is too dangerous. A wallet key can spend whatever it controls. A capability right can be scoped to a task, amount, service, time window, provider, or evidence requirement. A mandate tells the agent what it may attempt. A FeeEnvelope answers how publication or execution support is paid for. Useful-work evidence or service receipts explain why a later payout or settlement action is authorized. Linked liability defines what happens when autonomous or delayed execution conflicts with the allowed boundary.
This model is most useful when it stays bounded. An API agent may receive a prepaid right to call a service. A machine may receive a charging or access right that can be checked locally and reconciled later. A workflow agent may receive a budget envelope and a narrow mandate, not full treasury control. A useful-work contributor may receive a private reward only after WorkPackage evidence, challenge windows, and RewardAuthorization rules are satisfied. The corpus supports those patterns as a rights-first expansion lane. It does not prove a universal private AI economy, an autonomous treasury owner, or a hidden general-purpose VM.
Open Questions
Open questions remain around provider standards, capability schemas, revocation timing, fee sponsorship, evidence interoperability, model identity, agent reputation, and dispute routing. Those are not reasons to remove the use case. They are reasons to present it as an advanced scenario. The Agentic Offline Economy paper gives the strongest source for why machines and agents belong in the Z00Z family; the PoUW and DAO papers define the review and governance limits that stop the scenario from becoming uncontrolled automation.
For a real operator, the minimum safe question is “what exactly can this agent do without asking again?” The answer should be visible in the right object, mandate, fee envelope, expiration, provider scope, and evidence requirement. If that answer is vague, the docs should not describe the agent as production-ready.
Useful Work And Liability Boundary
Useful work and liability solve different parts of the agent problem. Useful-work evidence can show that a task was submitted, reviewed, challenged, and authorized for reward. Linked liability can make fraud, double execution, or overuse attributable under a bounded domain. Neither should become unrestricted surveillance. Honest-case privacy should remain preserved where the papers require it, while fraud-case or reward-case evidence becomes revealable only through the defined path.
The strongest agent-rights docs keep that split visible. They do not say “agents are trusted.” They say an agent receives a bounded right, spends inside a mandate, carries fee support, and can be reviewed, challenged, or penalized if the evidence path proves misconduct or unauthorized execution.
A final boundary is key custody. An agent should not be handed broad human wallet authority when a scoped capability right would do. That design choice keeps loss, compromise, model error, and prompt manipulation inside a smaller blast radius.
Read Next
Read Proof Of Useful Work for reward evidence, Linked Liability for delayed-execution accountability, and Governance before writing AI or treasury-control claims.
Evidence and Further Reading
- Agentic Offline Economy Whitepaper sections 3 through 8 are the primary source for spendable capability objects, rights wallets, machine rights, agent spending envelopes,
FeeEnvelope, and sequencing. - Use Cases Whitepaper section 9 is the source for ranking this family as the disciplined expansion lane rather than as the opening wedge.
- Proof-of-Useful-Work Whitepaper sections 4 through 8 are the companion source when the scenario depends on private reward claims,
WorkPackage,RewardAuthorization, or anti-replay payout discipline. - Linked Liability Whitepaper sections 3 through 6 are useful when the scenario needs bounded fraud accountability for delayed or autonomous execution.