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

# Reviewers

> The Reviewer-A / Reviewer-B split: independent, narrative-blind verification that gates Done. Worker agents cannot author their own final gate.

Verification is independent by construction. Two reviewers with different context packets gate completion, and neither is the worker who produced the artifact.

Inspect a run's reviews:

```bash theme={"theme":"github-dark"}
voss review            # latest run
voss review <run_id>   # a specific run
```

`voss review` is read-only — it reads persisted `.review.json` sidecars and constructs no live providers.

## Reviewer-A — authors the bar

Reviewer-A derives the verification bar from the **original human idea** — not the EM's acceptance criteria, tickets, or any prior review. It runs with a fresh memory per card so there is no cross-card bleed.

* For **code** cards it authors a test, runs it via `shell_run`, and the exit code is the verdict.
* For **AI** cards it authors a PASS/FAIL rubric, then judges against it.

Worker agents cannot author their own final gate.

## Reviewer-B — judges narrative-blind

Reviewer-B judges the artifact, diff, and Reviewer-A's verification summary using **only the card's own attributes** — original idea, acceptance criteria, artifact, diff, and A's summary. It sees no EM plan, tickets, or A's reasoning process, so it cannot be talked into a pass by the orchestrator's narrative.

It runs at two tiers: `fast` at intermediate gates, `strong` at the gate to `Done`. If Reviewer-A's verification has **diverged from what the original idea requires**, Reviewer-B can return `block` to abort the lineage.

## The verdict

Both reviewers emit a frozen `ReviewerVerdict`:

| Field             | Type                                          | Meaning                                        |
| ----------------- | --------------------------------------------- | ---------------------------------------------- |
| `conf`            | `float` 0–1                                   | Confidence                                     |
| `source`          | `"A"` \| `"B"`                                | Which reviewer authored it                     |
| `tier`            | `"fast"` \| `"strong"`                        | Model tier used                                |
| `verdict`         | `"pass"` \| `"fail"` \| `"block"`             | `block` aborts the lineage                     |
| `notes`           | `str`                                         | Reviewer text; appended to retry notes on fail |
| `evidence_refs`   | `tuple[str, ...]`                             | `file:line`, test names, eval refs             |
| `domain_inferred` | `"code"` \| `"ai"` \| `"docs"` \| `"unknown"` | Inferred work domain                           |

## Persisted artifacts

Each card writes a `<node_id>.review.json` sidecar (mode `0o600`) next to its session-tree node, under `.voss/sessions/<root_id>/`:

```json theme={"theme":"github-dark"}
{
  "a_verification": { "test_path_or_rubric": "...", "result": "pass", "notes": "..." },
  "b_verdict": { "conf": 0.91, "source": "B", "verdict": "pass", "...": "..." },
  "final_outcome": "Done"
}
```

These sidecars are the evidence the [board](/orchestration/board) gates Done on, and the input to [`voss review`](/orchestration/reviewers).