> ## Documentation Index
> Fetch the complete documentation index at: https://niceeval.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Review and Accept Historical Attempts

> Get candidate locators from --dry, inspect immutable Attempts, explicitly accept selected results with niceeval accept, and verify the denominator and origin with a new Run ID.

Automatic carry can be rejected when configuration identity changes. If you have reviewed a historical Attempt and confirmed that it still holds under the new configuration, use `niceeval accept` to accept it explicitly.

Do not use this command to bypass a real change in evaluation conditions. Rerun and collect new evidence when the task, assertions, Agent behavior, or Sandbox starting point has changed.

## Get candidate locators from the plan

Preview the current target first:

```sh theme={null}
pnpm exec niceeval exp checkout --dry
```

Historical results blocked by an identity difference show their full Attempt locator and a runnable `accept` suggestion. The locator is the only scope authorized by this operation; the command never implicitly selects other results from the same Run or Experiment.

## Inspect the historical Attempt

Use the full locator to inspect the immutable result in the terminal:

```sh theme={null}
pnpm exec niceeval show @1K1P0VJAPVJ12
```

Review the Verdict, Assertions, usage, timing, and evidence. To review the source Run's full denominator, also use:

```sh theme={null}
pnpm exec niceeval view --run <source-run-id>
```

## Accept the results you need in one operation

```sh theme={null}
pnpm exec niceeval accept @1K1P0VJAPVJ12
```

When one Experiment has several results to accept, put their locators in the same command:

```sh theme={null}
pnpm exec niceeval accept @1K1P0VJAPVJ12 @1K1P0VJAPVJ13
```

NiceEval validates every input before it proceeds. If any locator is invalid, ineligible, or conflicts with another input, the command publishes no business result at all.

After success, every target slot points to the original Attempt with `reference` and records the `accepted` action. The source Attempt's facts, source Run, and locator are neither modified nor copied.

## Verify the new Run

Success feedback shows the new Run ID:

```sh theme={null}
pnpm exec niceeval view --run <accepted-run-id>
```

This Run retains the current Experiment's full expected-slot denominator. Positions selected in the command are `accepted`; sibling slots not selected are `not-dispatched`. When you accept only part of the positions, the Report accurately shows incomplete result coverage.

Every `accept` publishes a new Run. It never modifies or merges with a previous invocation. Put results in one command when they need to form the same acceptance boundary.

## Common failures

| Error                     | Next step                                                                                                                       |
| ------------------------- | ------------------------------------------------------------------------------------------------------------------------------- |
| `malformed-locator`       | Copy the complete uppercase locator from `--dry` or a fixed query result; do not use an old UUID or ambiguous prefix.           |
| `locator-not-found`       | Confirm that you are reading the same Record. When reading a copied SQLite file, pass `--record` explicitly to `show`.          |
| `accept-ineligible`       | Follow the message to check the Verdict, timeout, identity, or Observability. Rerun if you cannot prove the result still holds. |
| `duplicate-accept-member` | Keep one locator for every target slot and remove duplicate or conflicting input.                                               |

For the choice between automatic carry and forced reruns, see [Rerun and Carry Results](/docs/tutorials/rerun-and-cache).