Datasets API (Phase 5 E2)

All endpoints verified working via curl against a real dev server (real Postgres, a real OpenAI account behind the gateway connection — no mocked fetch). Document updated only after curl confirmation, per CLAUDE.md's API-reference rule.

Datasets hold examples used later by the evaluation/experiment domains (E3+): each example is an input (a variable bag to re-render against a candidate prompt template) plus an optional criteria (the rubric a judge checks against). Examples can be added manually (POST /:id/examples) or in bulk by selecting existing trace feedback rows (POST /from-feedback) — the latter only works for feedback whose source trace had captured payloads (x-capture-payloads: true on the originating gateway call), since it reads span_payloads.variables.

Auth: requireAnyAuth (session cookie or personal/team API key) on every route, no role restriction. Team-scoped throughout — a dataset outside the caller's team behaves as if it doesn't exist (404). DELETE /:id is a soft-delete: the row is excluded from GET / (list) and GET /:id returns 404 for it afterwards, same as a truly missing id.

POST /api/v1/datasets

Creates an empty dataset — no examples yet. overall_feedback is optional (a dataset-level rubric applied on top of each example's own criteria). Examples are added afterwards via POST /:id/examples or POST /from-feedback.

curl
TypeScript
Python

curl -X POST $ACRUXCORE_BASE_URL/datasets \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "manual-review-set", "overall_feedback": "Keep responses concise and polite"}'

const dataset = await hub.datasets.create({
  name: 'manual-review-set',
  description: 'Keep responses concise and polite',
});

dataset = await hub.datasets.create(
    'manual-review-set',
    overall_feedback='Keep responses concise and polite',
)

Response (status 201):

{
  "id": "04f757bb-c9a5-4400-8958-018defbd2fa1",
  "teamId": "8385d37a-c015-4787-b3da-0bf64d2378f0",
  "name": "manual-review-set",
  "overallFeedback": "Keep responses concise and polite",
  "createdBy": "4949b28b-cace-488d-9369-cdedac4fa4a2",
  "createdAt": "2026-07-07T08:40:51.145Z",
  "updatedAt": "2026-07-07T08:40:51.145Z",
  "exampleCount": 0
}

Error responses

Missing name (status 400):

curl -X POST $ACRUXCORE_BASE_URL/datasets \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"overall_feedback": "no name provided"}'

{ "error": { "code": "VALIDATION_ERROR", "message": "Required" } }

POST /api/v1/datasets/from-feedback

Builds a dataset from selected feedback rows in one call. Each eligible feedback row becomes one example: input = the captured prompt variables from the feedback's source trace, criteria = the feedback's comment. overall_feedback (optional) is a dataset-level rubric applied on top of each example's own criteria. Feedback rows are only eligible if their source trace has captured payloads (span_payloads.variables) — this example uses a feedback row from a real gateway completion made with x-capture-payloads: true and a prompt-ref body ({"prompt":{"name":"greeting-e2task4","alias":"production","variables":{"name":"Alice"}}}), then POST /traces/:id/feedback with {"rating":-1,"comment":"Use third person, do not say I"}.

The rows come from one of two mutually exclusive fields — send exactly one:

feedback_ids — an explicit list, at most 100 per request.
filter — the criteria that select them, using the same vocabulary as GET /api/v1/traces/feedback (prompt_id, rating, has_comment, tags, metadata, q, from/to, and the rest).

Either way at most 100 rows are processed — the build runs synchronously and reads several rows per id. A filter request also returns matched, the total the criteria selected, so a capped selection is visible rather than silently truncated. Split a larger selection into batches; each batch builds its own dataset.

curl
TypeScript
Python

curl -X POST $ACRUXCORE_BASE_URL/datasets/from-feedback \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "unhappy-greetings", "overall_feedback": "Always third person", "feedback_ids": ["44c64fa2-8189-4ab0-a08a-a28a14893a06"]}'

const result = await hub.datasets.buildFromFeedback({
  name: 'unhappy-greetings',
  overallFeedback: 'Always third person',
  feedbackIds: ['44c64fa2-8189-4ab0-a08a-a28a14893a06'],
});

result = await hub.datasets.build_from_feedback(
    'unhappy-greetings',
    ['44c64fa2-8189-4ab0-a08a-a28a14893a06'],
    overall_feedback='Always third person',
)

Response (status 201):

{
  "id": "4c979706-3e84-4e8f-a089-6ce4dd9f6e4a",
  "name": "unhappy-greetings",
  "overall_feedback": "Always third person",
  "example_count": 1,
  "skipped": []
}

The created example, as seen via GET /api/v1/datasets/:id (below):

{
  "id": "fe8244a4-e9c6-4a38-a061-4e27faa778f8",
  "datasetId": "4c979706-3e84-4e8f-a089-6ce4dd9f6e4a",
  "input": { "name": "Alice" },
  "criteria": "Use third person, do not say I",
  "sourceTraceId": "131cffd7-3729-4b69-95e1-7c073ecffdc9",
  "sourceFeedbackId": "44c64fa2-8189-4ab0-a08a-a28a14893a06",
  "sourcePromptVersionId": "41014d2e-9a06-4053-9514-afacc4b06676",
  "createdAt": "2026-07-07T08:31:14.773Z"
}

Mixed eligibility — one captured, one not (`skipped` populated)

A second gateway completion was made against the same prompt WITHOUT the x-capture-payloads header, so its trace has no span_payloads row. Feedback was posted on that trace too ({"comment": "no captured variables"}), then both feedback ids were passed together:

curl -X POST $ACRUXCORE_BASE_URL/datasets/from-feedback \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "mixed-eligibility", "feedback_ids": ["44c64fa2-8189-4ab0-a08a-a28a14893a06", "3ef48b8c-c731-441d-8c76-0019f2807b3d"]}'

Response (status 201) — one example built, the uncaptured one skipped with a reason:

{
  "id": "9330c7d0-1dc1-4642-815c-0bfbd0f56d97",
  "name": "mixed-eligibility",
  "overall_feedback": null,
  "example_count": 1,
  "skipped": [
    { "feedbackId": "3ef48b8c-c731-441d-8c76-0019f2807b3d", "reason": "the prompt is not stored in AcruxCore" }
  ]
}

Selecting rows by criteria instead of by id

"Every thumbs-down that carries a written comment" is one request, rather than a morning of ticking boxes. matched is how many rows the criteria selected in total; example_count is how many of them became examples.

curl -X POST $ACRUXCORE_BASE_URL/datasets/from-feedback \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "checkout-regressions", "filter": {"rating": "down", "has_comment": true}}'

Response (status 201):

{
  "id": "7c9aa45f-8411-45e6-9e73-edfb64963d43",
  "name": "checkout-regressions",
  "overall_feedback": null,
  "example_count": 3,
  "skipped": [
    {
      "feedbackId": "5c9baf1d-3a51-4e38-8dab-76de04ebf8c7",
      "reason": "the prompt is not stored in AcruxCore"
    }
  ],
  "matched": 4
}

Sending both feedback_ids and filter, or neither, is rejected before anything is read:

curl -X POST $ACRUXCORE_BASE_URL/datasets/from-feedback \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "x", "feedback_ids": ["d3cabcc9-2275-4b04-b1e6-41d1db0bef57"], "filter": {"rating": "down"}}'

{ "error": { "code": "VALIDATION_ERROR", "message": "Provide exactly one of feedback_ids or filter." } }

Error responses

All requested feedback ids are ineligible (no captured variables) — 422, same as "none eligible" below rather than a partial 201:

curl -X POST $ACRUXCORE_BASE_URL/datasets/from-feedback \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "empty", "feedback_ids": ["5c9baf1d-3a51-4e38-8dab-76de04ebf8c7"]}'

The message names the reason the rows were actually skipped. Here the run never rendered a prompt stored in AcruxCore, so there are no variables to replay:

{ "error": { "code": "UNPROCESSABLE", "message": "No eligible feedback rows — the prompt is not stored in AcruxCore." } }

When the rows failed for more than one reason, each is listed with its count: "No eligible feedback rows — 2 because the prompt is not stored in AcruxCore; 1 because no prompt variables were captured."

A feedback id that does not exist (status 422, same message/code as above — a nonexistent id is just another way to have zero eligible rows):

curl -X POST $ACRUXCORE_BASE_URL/datasets/from-feedback \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "missing", "feedback_ids": ["00000000-0000-0000-0000-000000000000"]}'

{ "error": { "code": "UNPROCESSABLE", "message": "No eligible feedback rows — feedback not found." } }

More than 100 feedback_ids (status 400) — the request is rejected before any row is read, so no dataset is created:

curl -X POST $ACRUXCORE_BASE_URL/datasets/from-feedback \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "too-many", "feedback_ids": ["<101 uuids>"]}'

{ "error": { "code": "VALIDATION_ERROR", "message": "Array must contain at most 100 element(s)" } }

Note: the all-eligible-plus-team-isolation scenarios (every requested feedback id eligible; a cross-team feedback id skipped/422) are covered by the automated test suite (apps/api/src/evaluations/datasets/datasets.test.ts) and were not separately re-curled here since the mixed-eligibility case above already exercises the same code path with real data.

Multi-turn feedback captures the prior-turn history

When the flagged trace's session_id links it to earlier traces in the same conversation, the example built from that feedback also carries a reconstructed history — every prior turn's user/assistant messages, tool calls, and tool results, in order. example_count is still 1: this is the same example row, with one extra field. This example reused the production alias trace from above, first replaying a prior turn in the same session (x-session-id header, not shown — captured payloads are enough for the reconstruction) and leaving feedback on the second turn:

curl -X POST $ACRUXCORE_BASE_URL/datasets/from-feedback \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "doc-history-demo-dataset", "feedback_ids": ["113ebbdf-ef46-41d4-8b93-1dd5f3f6aeef"]}'

The created example, as seen via GET /api/v1/datasets/:id:

{
  "id": "ef1d5b62-014f-43e4-be78-ed8c20c4a80a",
  "datasetId": "71e14e39-574d-49ee-a2e3-776994e7d563",
  "input": { "question": "And what is its population?" },
  "criteria": "Give the population as of the most recent census, with a year.",
  "history": [
    { "role": "user", "content": "What is the capital of France?" },
    { "role": "assistant", "content": "The capital of France is Paris." }
  ],
  "sourceTraceId": "11111111-1111-1111-1111-111111111112",
  "sourceFeedbackId": "113ebbdf-ef46-41d4-8b93-1dd5f3f6aeef",
  "sourcePromptVersionId": "51c52ee0-23db-454b-b044-a6a1a242f8e5",
  "createdAt": "2026-08-03T20:59:15.345Z"
}

A single-turn example (no session_id, or the first turn in a session) has history: null — see the first example at the top of this page, predating this field. history is also null when the prior turns' spans hold nothing readable to replay: reconstruction is extra context on the example, never the example itself, so it degrades to null instead of failing the build.

Prior turns recorded by the TypeScript or Python SDK's auto-trace are read the same way as ones the gateway recorded — the two write their captured payloads in different shapes and both are understood. When a turn was reported twice (the gateway writes its own span and the SDK self-reports one for the same call), it appears once in history.

Every eval run freezes an example's history at run-start time and replays it ahead of the new turn, so a candidate prompt is judged against the same conversation the model actually saw — see GET /api/v1/runs/:id/cells/:cellKey in the experiments reference.

GET /api/v1/datasets

Lists the team's non-deleted datasets, newest activity first. No pagination params observed in the controller (returns the full team list in data).

curl
TypeScript
Python

curl $ACRUXCORE_BASE_URL/datasets \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY"

const datasets = await hub.datasets.list();

datasets = await hub.datasets.list()

Response (status 200):

{
  "data": [
    {
      "id": "4c979706-3e84-4e8f-a089-6ce4dd9f6e4a",
      "teamId": "fcfb223d-2780-458b-adc5-ec53cf87f740",
      "name": "unhappy-greetings",
      "overallFeedback": "Always third person",
      "createdBy": "d0d70477-ecc2-4f1f-85a8-a382acf67e4d",
      "createdAt": "2026-07-07T08:31:14.772Z",
      "updatedAt": "2026-07-07T08:31:14.772Z",
      "exampleCount": 1
    }
  ]
}

GET /api/v1/datasets/:id

Fetches one dataset with its full example list (unlike the list endpoint, which only returns exampleCount).

curl
TypeScript
Python

curl $ACRUXCORE_BASE_URL/datasets/4c979706-3e84-4e8f-a089-6ce4dd9f6e4a \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY"

const dataset = await hub.datasets.get('4c979706-3e84-4e8f-a089-6ce4dd9f6e4a');

dataset = await hub.datasets.get('4c979706-3e84-4e8f-a089-6ce4dd9f6e4a')

Response (status 200):

{
  "id": "4c979706-3e84-4e8f-a089-6ce4dd9f6e4a",
  "teamId": "fcfb223d-2780-458b-adc5-ec53cf87f740",
  "name": "unhappy-greetings",
  "overallFeedback": "Always third person",
  "createdBy": "d0d70477-ecc2-4f1f-85a8-a382acf67e4d",
  "createdAt": "2026-07-07T08:31:14.772Z",
  "updatedAt": "2026-07-07T08:31:14.772Z",
  "exampleCount": 1,
  "examples": [
    {
      "id": "fe8244a4-e9c6-4a38-a061-4e27faa778f8",
      "datasetId": "4c979706-3e84-4e8f-a089-6ce4dd9f6e4a",
      "input": { "name": "Alice" },
      "criteria": "Use third person, do not say I",
      "history": null,
      "sourceTraceId": "131cffd7-3729-4b69-95e1-7c073ecffdc9",
      "sourceFeedbackId": "44c64fa2-8189-4ab0-a08a-a28a14893a06",
      "sourcePromptVersionId": "41014d2e-9a06-4053-9514-afacc4b06676",
      "sourcePrompt": {
        "promptId": "a290cf2f-a724-4389-8bea-da8b53229e00",
        "name": "support-reply",
        "versionNumber": 1,
        "lastUserMessage": "Reply to {{ name }}"
      },
      "createdAt": "2026-07-07T08:31:14.773Z"
    }
  ]
}

PATCH /api/v1/datasets/:id

Updates name and/or overall_feedback. Both fields optional; omitted fields keep their current value.

curl
TypeScript
Python

curl -X PATCH $ACRUXCORE_BASE_URL/datasets/4c979706-3e84-4e8f-a089-6ce4dd9f6e4a \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "unhappy-greetings-v2", "overall_feedback": "Always use third person, never first person"}'

const updated = await hub.datasets.update('4c979706-3e84-4e8f-a089-6ce4dd9f6e4a', {
  name: 'unhappy-greetings-v2',
  description: 'Always use third person, never first person',
});

updated = await hub.datasets.update(
    '4c979706-3e84-4e8f-a089-6ce4dd9f6e4a',
    name='unhappy-greetings-v2',
    overall_feedback='Always use third person, never first person',
)

Response (status 200):

{
  "id": "4c979706-3e84-4e8f-a089-6ce4dd9f6e4a",
  "teamId": "fcfb223d-2780-458b-adc5-ec53cf87f740",
  "name": "unhappy-greetings-v2",
  "overallFeedback": "Always use third person, never first person",
  "createdBy": "d0d70477-ecc2-4f1f-85a8-a382acf67e4d",
  "createdAt": "2026-07-07T08:31:14.772Z",
  "updatedAt": "2026-07-07T08:31:37.620Z",
  "exampleCount": 1
}

POST /api/v1/datasets/:id/examples

Adds one example manually — no source trace/feedback needed. input is a free-form object (the variable bag); criteria is optional. history is also optional — an array of chat messages (role, content, and/or tool_calls/tool_call_id) giving the prior turns to replay ahead of input during a run; capped at 20 messages / 32KB. Omit it for a single-turn example.

curl
TypeScript
Python

curl -X POST $ACRUXCORE_BASE_URL/datasets/4c979706-3e84-4e8f-a089-6ce4dd9f6e4a/examples \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"input": {"name": "Bob"}, "criteria": "Say hi politely, third person"}'

const example = await hub.datasets.addExample(
  '4c979706-3e84-4e8f-a089-6ce4dd9f6e4a',
  { input: { name: 'Bob' }, criteria: 'Say hi politely, third person' }
);

example = await hub.datasets.add_example(
    '4c979706-3e84-4e8f-a089-6ce4dd9f6e4a',
    input={'name': 'Bob'},
    criteria='Say hi politely, third person',
)

Response (status 201) — sourceTraceId/sourceFeedbackId/sourcePromptVersionId are null for a manually-added example, and so is history when the request didn't supply one:

{
  "id": "001d9f90-45fe-41f7-8f59-d1f833549fac",
  "datasetId": "4c979706-3e84-4e8f-a089-6ce4dd9f6e4a",
  "input": { "name": "Bob" },
  "criteria": "Say hi politely, third person",
  "history": null,
  "sourceTraceId": null,
  "sourceFeedbackId": null,
  "sourcePromptVersionId": null,
  "createdAt": "2026-07-07T08:31:37.634Z"
}

Supplying history explicitly:

curl -X POST $ACRUXCORE_BASE_URL/datasets/71e14e39-574d-49ee-a2e3-776994e7d563/examples \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"input": {"question": "What about Germany?"}, "criteria": "Answer in one sentence, include the year.", "history": [{"role": "user", "content": "What is the capital of France?"}, {"role": "assistant", "content": "Paris."}]}'

Response (status 201):

{
  "id": "b4563eb4-3add-44a0-84bc-73e673153cf0",
  "datasetId": "71e14e39-574d-49ee-a2e3-776994e7d563",
  "input": { "question": "What about Germany?" },
  "criteria": "Answer in one sentence, include the year.",
  "history": [
    { "role": "user", "content": "What is the capital of France?" },
    { "role": "assistant", "content": "Paris." }
  ],
  "sourceTraceId": null,
  "sourceFeedbackId": null,
  "sourcePromptVersionId": null,
  "createdAt": "2026-08-03T20:59:25.851Z"
}

A history longer than 20 messages is rejected (status 400):

curl -X POST $ACRUXCORE_BASE_URL/datasets/54fd9af3-c470-4d1a-a0eb-ea192b5b43ab/examples \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"input": {"q": "x"}, "history": [{"role": "user", "content": "m0"}, "… 21 messages …"]}'

{ "error": { "code": "VALIDATION_ERROR", "message": "Array must contain at most 20 element(s)" } }

POST /api/v1/datasets/:id/examples/from-feedback

Appends feedback rows to a dataset that already exists. The counterpart to POST /datasets/from-feedback, which always creates a new one — this is the call for the second and every later pass over the feedback list.

Each eligible row contributes the variables captured on its trace as input, its comment as criteria, and its reconstructed session history. A feedback id whose example is already in this dataset is reported in skipped, not inserted twice. At most 100 rows per call.

Takes the same two mutually exclusive selectors as POST /datasets/from-feedback: feedback_ids, or a filter. The filter form is what makes a dataset something you top up — save the criteria once, and every later pass picks up only the new traffic.

The status distinguishes the two outcomes: 201 when at least one example was created, 200 when none were. Neither is an error — "all three rows were already filed" is a normal answer.

curl only for now

Neither published SDK exposes this endpoint yet. Use the REST call directly.

curl -X POST $ACRUXCORE_BASE_URL/datasets/0981fbbf-51a0-419c-a9e4-4e9a08bb7dc9/examples/from-feedback \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"feedback_ids": ["055effc3-b3fe-4951-b24f-e9428aa9df09", "11157385-e9eb-4cc4-9bea-9851034493c5"]}'

Response (status 201):

{ "added": 2, "example_count": 2, "skipped": [] }

Running the same call again reports the rows as already present rather than duplicating them.

Response (status 200):

{
  "added": 0,
  "example_count": 2,
  "skipped": [
    { "feedbackId": "055effc3-b3fe-4951-b24f-e9428aa9df09", "reason": "already in this dataset" }
  ]
}

The same call driven by criteria rather than ids. Everything the filter matched this time was either already filed or ineligible, so nothing was added:

curl -X POST $ACRUXCORE_BASE_URL/datasets/7c9aa45f-8411-45e6-9e73-edfb64963d43/examples/from-feedback \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"filter": {"rating": "down", "has_comment": true}}'

Response (status 200):

{
  "added": 0,
  "example_count": 3,
  "skipped": [
    {
      "feedbackId": "5c9baf1d-3a51-4e38-8dab-76de04ebf8c7",
      "reason": "the prompt is not stored in AcruxCore"
    },
    { "feedbackId": "055effc3-b3fe-4951-b24f-e9428aa9df09", "reason": "already in this dataset" },
    { "feedbackId": "11157385-e9eb-4cc4-9bea-9851034493c5", "reason": "already in this dataset" },
    { "feedbackId": "f6548350-b0dc-4292-9948-385c4f546530", "reason": "already in this dataset" }
  ],
  "matched": 4
}

PATCH /api/v1/datasets/:id/examples/:exampleId

Edits one example's criteria — the per-example rubric the judge grades against. Send null to clear it; omit the key to leave it unchanged.

Only criteria is editable. input and history are the frozen record of what actually ran, so rewriting them would invalidate every past run that graded against them.

curl only for now

Neither published SDK exposes this endpoint yet. Use the REST call directly.

curl -X PATCH $ACRUXCORE_BASE_URL/datasets/0981fbbf-51a0-419c-a9e4-4e9a08bb7dc9/examples/99684fdf-2b1e-4b68-8a7a-896272dc3c32 \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"criteria": "Says plainly whether a date exists, names what is known, and gives one next step."}'

Response (status 200):

{
  "id": "99684fdf-2b1e-4b68-8a7a-896272dc3c32",
  "datasetId": "0981fbbf-51a0-419c-a9e4-4e9a08bb7dc9",
  "input": {
    "message": "Our API keys stopped working this morning — every call comes back 401 and nothing changed on our side. We have a customer demo at 2pm and I am getting nervous."
  },
  "criteria": "Says plainly whether a date exists, names what is known, and gives one next step.",
  "history": null,
  "sourceTraceId": "82a2f4d6-f6ce-4abd-b21b-beaff6b59fd2",
  "sourceFeedbackId": "11157385-e9eb-4cc4-9bea-9851034493c5",
  "sourcePromptVersionId": "4ccc4890-c7ca-4352-af8a-0a74896b1ff9",
  "sourcePrompt": {
    "promptId": "a290cf2f-a724-4389-8bea-da8b53229e00",
    "name": "support-reply",
    "versionNumber": 1,
    "lastUserMessage": "Reply to {{ name }}"
  },
  "createdAt": "2026-09-07T19:52:29.769Z"
}

DELETE /api/v1/datasets/:id/examples/:exampleId

Removes one example. Verified by re-fetching the dataset afterwards — exampleCount dropped from 2 back to 1 and the removed example is gone from examples.

curl
TypeScript
Python

curl -X DELETE $ACRUXCORE_BASE_URL/datasets/4c979706-3e84-4e8f-a089-6ce4dd9f6e4a/examples/001d9f90-45fe-41f7-8f59-d1f833549fac \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY"

await hub.datasets.removeExample(
  '4c979706-3e84-4e8f-a089-6ce4dd9f6e4a',
  '001d9f90-45fe-41f7-8f59-d1f833549fac'
);

await hub.datasets.remove_example(
    '4c979706-3e84-4e8f-a089-6ce4dd9f6e4a',
    '001d9f90-45fe-41f7-8f59-d1f833549fac',
)

Response (status 200):

{ "success": true }

DELETE /api/v1/datasets/:id

Soft-deletes a dataset.

curl
TypeScript
Python

curl -X DELETE $ACRUXCORE_BASE_URL/datasets/4c979706-3e84-4e8f-a089-6ce4dd9f6e4a \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY"

await hub.datasets.delete('4c979706-3e84-4e8f-a089-6ce4dd9f6e4a');

await hub.datasets.delete('4c979706-3e84-4e8f-a089-6ce4dd9f6e4a')

Response (status 200):

{ "success": true }

GET on the now-soft-deleted id behaves exactly like a nonexistent id:

curl $ACRUXCORE_BASE_URL/datasets/4c979706-3e84-4e8f-a089-6ce4dd9f6e4a \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY"

{ "error": { "code": "NOT_FOUND", "message": "Dataset not found." } }

Error responses

GET a dataset id that never existed (status 404, same message as a soft-deleted one — never distinguishes the two):

curl $ACRUXCORE_BASE_URL/datasets/00000000-0000-0000-0000-000000000000 \
  -H "Authorization: Bearer $ACRUXCORE_API_KEY"

{ "error": { "code": "NOT_FOUND", "message": "Dataset not found." } }

No Authorization header on GET /api/v1/datasets (status 401):

curl $ACRUXCORE_BASE_URL/datasets

{ "error": { "code": "UNAUTHORIZED", "message": "Authentication required." } }

POST /api/v1/datasets​

Error responses​

POST /api/v1/datasets/from-feedback​

Mixed eligibility — one captured, one not (skipped populated)​

Selecting rows by criteria instead of by id​

Error responses​

Multi-turn feedback captures the prior-turn history​

GET /api/v1/datasets​

GET /api/v1/datasets/:id​

PATCH /api/v1/datasets/:id​

POST /api/v1/datasets/:id/examples​

POST /api/v1/datasets/:id/examples/from-feedback​

PATCH /api/v1/datasets/:id/examples/:exampleId​

DELETE /api/v1/datasets/:id/examples/:exampleId​

DELETE /api/v1/datasets/:id​

Error responses​

POST /api/v1/datasets

Error responses

POST /api/v1/datasets/from-feedback

Mixed eligibility — one captured, one not (`skipped` populated)

Selecting rows by criteria instead of by id

Error responses

Multi-turn feedback captures the prior-turn history

GET /api/v1/datasets

GET /api/v1/datasets/:id

PATCH /api/v1/datasets/:id

POST /api/v1/datasets/:id/examples

POST /api/v1/datasets/:id/examples/from-feedback

PATCH /api/v1/datasets/:id/examples/:exampleId

DELETE /api/v1/datasets/:id/examples/:exampleId

DELETE /api/v1/datasets/:id

Error responses