> For the complete documentation index, see [llms.txt](https://docs.trueopen.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.trueopen.ai/protocol/v1/tasks/task-06-settlement.md).

# Settlement

Build one immutable accounting plan, finalize the Task, and release retained data.

Settlement begins only when no verification round is open and the challenge window cannot admit another round. The effective verdict is the verdict of the latest applicable closed round.

```mermaid
sequenceDiagram
    participant B as Task Builders
    participant P as Permissionless submitter
    participant K as Chain
    participant U as User
    participant S as Service providers

    K->>K: Open settlement window and freeze facts cutoff
    alt Ranked Task Builder submits in its grace
        B->>K: Submit settlement trigger
    else All Builder grace periods expire
        P->>K: Submit the same settlement trigger
    else Nobody submits
        K->>K: Bounded protocol runner triggers settlement
    end
    K->>K: Build settlement facts from accepted state
    K->>K: Apply pending objective effects
    K->>K: Credit earnings and reimbursements
    K->>K: Transfer maintenance to treasury
    K->>U: Return unused escrow
    K->>K: Record finality atomically
    K-->>S: Final earnings become claimable under maturity rules
    B->>K: Poll cleanup eligibility
    B->>B: Delete Task data after retention gates pass
```

## Trigger paths

Task Builders receive ranked opportunities to submit the settlement trigger. Only one valid submission is required; the Builders do not co-sign the final accounting plan.

After all Builder grace periods expire, the selected Worker, a Verifier, or any maintenance submitter may trigger the same transition. The submitter pays transaction fees but cannot change the verdict, recipients, amounts, or facts cutoff.

If nobody submits, a bounded protocol runner invokes the same settlement builder. Submission backlog may change the execution height, but not the frozen facts used for accounting.

## Build from accepted state

The chain derives the plan from immutable Task, round, receipt, result, gas-reimbursement, and responsibility state. The caller does not supply a bill or alternative participant list.

Before paying, settlement applies any confirmed objective effects whose protocol boundary is finality. It then:

* Pays an eligible Worker only under the effective successful verdict
* Pays eligible round-1 Verifier slots that were not later disproved
* Pays recorded eligible Task transaction-fee reimbursement
* Sends maintenance deductions to the treasury
* Returns all remaining Task escrow to the User

Challenge Verifiers were funded from the separate challenge budget and are not paid again from the User's Task escrow.

## Atomic finality

All balance transfers, earnings credits, refunds, fault effects, settlement facts, and the Task finality height are committed in one transaction. Any failed invariant rolls back the entire transition.

There is no “settled but still challengeable” state and no need to reverse an earlier payment. Participants disproved by a challenge are simply excluded from the only settlement plan ever applied.

The round-1 verdict can be queried before settlement, but it is not an account balance or claimable entitlement.

## Claims

Settlement credits service earnings to the rewards ledger. The applicable maturity rule determines when an operator may withdraw them. A claim does not recalculate the Task; it withdraws an already finalized entitlement.

## Data cleanup

Finality does not necessarily authorize immediate deletion. Task Builders and Cortex Nodes retain input, output, and evidence until every configured retention and responsibility gate passes.

Cleanup polling verifies that:

* The Task is final
* No round or economic effect remains open
* Challenge and evidence-retention margins have ended
* Required audit commitments remain preserved

Only then may large Task data be deleted. Cleanup never erases the minimal facts needed to explain the final verdict and accounting plan.

See [Challenges and settlement](/protocol/v1/tasks/challenges-and-settlement.md), [Economics](/protocol/v1/economics.md), and [Data and evidence](/protocol/v1/data-and-evidence.md).


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.trueopen.ai/protocol/v1/tasks/task-06-settlement.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.