Export a decision's feedback as training data

Lists feedback with its claims and evaluated cases; reads only. The raw feedback on a decision, for a producer's training data (calibration, suggested versions): one item per submission, oldest first, with its claims (an expected_answer with the hash of the question it judged), the evaluation it judged (structure hash, engine profile, status, value, outcome, via and every aggregation's result, each question's System One answer under `answer` with its question hash, distribution and engine provenance, and content: {state, attribute_values?} when inputs were stored, otherwise content: null) and its group's split. Holdout feedback is never listed, not its claims, its evaluation or its content; it is only counted in `withheld.holdout`, together with feedback whose content's text belongs to a held-out group, so suggestions can still be tested on feedback their producer never saw. `withheld` counts submission rows, including withdrawal events, not distinct evaluations. It is counted on the first page only (no `cursor`) and is null on later pages. Counts per `structure_hash` filter can tell you that a version gained held-out feedback, never which item. It is raw, not a consensus: superseded submissions are listed and marked `superseded_by`, and how submissions become training data is the reader's call. `withdrawn_at` marks a claimless withdrawal event: resolve chain heads across pages, exclude withdrawn chains without reviving ancestors, and invalidate cached answers. On a spot check, `after_reveal` is false on the reviewer's first, blind answer and true on each answer that changed it after Moonbounce's call was revealed: pick the blind answer to measure accuracy, the latest to train. `channel` says where each submission came from; Slack's terms bar training an LLM on `slack` feedback. Scoped to the key's organization and environment, and feedback on deleted evaluations is left out. A decision with no feedback this key can see answers 200 with no items. `structure_hash` keeps only evaluations of that structure hash. `split.group` is an opaque id shared by submissions on near-identical items in the key's environment; it never names an item. Page forward by passing `next_cursor` as `cursor` while `has_more` is true. `next_cursor` is null on the last page. Keep `resume_cursor` for later polling; an empty poll retains it: reading from it later returns what was written after it, apart from the following. Items are in creation order, not commit order, and a submission's creation time is the start of the transaction that wrote it: any write that commits after a later one was read (a seed batch, or any concurrent submission) can land behind a cursor already handed out, and a holdout group that is burned later appears under its submissions' original creation time. A reason added later to a submission already read (a reviewer's `POST /v1/review/items/{review_item_id}/reason`) does not move it either. So a forward reader re-reads from the start now and then. Not an MCP tool, so bulk content stays out of coding agents' context. The web session may not read it (403).

GET /v1/decisions/{decision_id}/feedback

Export a decision's feedback as training data

cURL

curl --request GET \
  --url https://api.moonbounce.ai/v1/decisions/{decision_id}/feedback \
  --header 'Authorization: Bearer <token>'

200

{
  "environment": "production",
  "items": [
    {
      "id": "<string>",
      "source": "app_native",
      "channel": "slack",
      "actor": "<string>",
      "reason_pending": true,
      "unsure": true,
      "created_at": "2023-11-07T05:31:56Z",
      "claims": [
        {
          "claim": {
            "kind": "expected_answer",
            "question": "<string>",
            "value": true
          },
          "shown": {
            "structure_hash": "<string>"
          },
          "question_hash": "<string>"
        }
      ],
      "evaluation": {
        "id": "<string>",
        "item_id": "<string>",
        "assignment_mode": "primary",
        "created_at": "2023-11-07T05:31:56Z",
        "version": {
          "structure_hash": "<string>",
          "binding": 2,
          "engine_profile": "<string>"
        },
        "aggregations": {},
        "answers": {},
        "content": {
          "state": "<unknown>",
          "attribute_values": {}
        },
        "status": "computed",
        "value": 123,
        "via": "<string>",
        "outcome": "<unknown>",
        "missing": [
          "questions.asks_for_refund"
        ]
      },
      "split": {
        "split": "seed",
        "group": "<string>"
      },
      "reason": "<string>",
      "reason_added_at": "2023-11-07T05:31:56Z",
      "withdrawn_at": "2023-11-07T05:31:56Z",
      "after_reveal": true,
      "supersedes": "<string>",
      "superseded_by": "<string>"
    }
  ],
  "next_cursor": "<string>",
  "resume_cursor": "<string>",
  "has_more": true,
  "withheld": {
    "holdout": 1,
    "unassigned": 1
  }
}

Authorizations

Authorization · string · header · required

An API key, sent as Authorization: Bearer <key>. Each operation lists the scope its key needs: decisions:evaluate (evaluate and read evaluations), decisions:feedback (your app's feedback) or decisions:develop (everything else, including owner feedback and the review queue). An operation listing no scope takes any scoped key. A key with no scope, or without the operation's, gets 403 INSUFFICIENT_SCOPE.

Path Parameters

decision_id · string · required

Pattern: ^[a-z][a-z0-9_.-]{0,127}$

Query Parameters

structure_hash · string

Only feedback on evaluations of this structure hash.

Pattern: ^sha256:[0-9a-f]{64}$

limit · integer · default: 50

Required range: 1 <= x <= 100

cursor · string

The previous page's next_cursor, or its resume_cursor to poll later; retained on empty export polls.

Required string length: 1 - 1024

Response

200

application/json

A page of feedback, oldest first

environment · enum<string> · required

The environment of the key that made the call: production or test. Evaluations, feedback, reviews and suggestions are separate per environment; registered structures are shared.

Available options: production, test

items · object[] · required

Show child attributes

next_cursor · string | null · required

Next page, or null at the end.

resume_cursor · string | null · required

Export checkpoint for later polls, retained on empty pages. Pass as cursor on a later poll. Creation ordering requires periodic full rescans for late commits, holdout releases and updates; this is not a lossless change feed.

has_more · boolean · required

True when more items follow this page now.

withheld · object | null · required

Submission rows, including withdrawal events (not distinct evaluations), in scope (this decision, the key's environment, the same version filters) that the export leaves out of every page. Counted on the first page only (no cursor); null on later pages. Counts per filter can tell you that a version gained held-out feedback, never which item.

Show child attributes