Evaluation Sets Migration | Extend Documentation

What’s Changed

New evaluation set runs endpoints — POST /evaluation_set_runs to start an evaluation run and GET /evaluation_set_runs/{id} to retrieve it
Nested item paths — /evaluation_set_items → /evaluation_sets/{id}/items
New get item endpoint — GET /evaluation_sets/{id}/items/{itemId} (wasn’t available before)
Entity objects — processorId replaced with typed entity object

If you don’t use evaluation sets, you can skip this page.

Quick Start: Common Patterns

TypeScript

Python

Java

Before (2025-04-21)

// Evaluation Sets
const sets = await client.evaluationSet.list({ processorId: "dp_abc123" });
const set = await client.evaluationSet.get("ev_abc123");
// Evaluation Set Items
const items = await client.evaluationSetItem.list("ev_abc123");
await client.evaluationSetItem.update("evi_xyz789", { expectedOutput: {...} });

After (2026-02-09)

// Evaluation Set Runs (new!)
const evalRun = await client.evaluationSetRuns.retrieve("bpr_abc123");
// Evaluation Sets
const sets = await client.evaluationSets.list({ entityId: "ex_abc123" });
const set = await client.evaluationSets.retrieve("ev_abc123");
// Evaluation Set Items
const items = await client.evaluationSetItems.list("ev_abc123");
const item = await client.evaluationSetItems.retrieve("ev_abc123", "evi_xyz789");
await client.evaluationSetItems.update("ev_abc123", "evi_xyz789", { expectedOutput: {...} });

Endpoint Changes Summary

Old Endpoint	New Endpoint	Change
`GET /batch_processor_runs/{id}`	`GET /batch_processor_runs/{id}`	Deprecated but still works
—	`POST /evaluation_set_runs`	New for starting evaluation set runs
—	`GET /evaluation_set_runs/{id}`	New for evaluation set runs
`GET /evaluation_sets`	`GET /evaluation_sets`	Query param `processorId` → `entityId`
`POST /evaluation_sets`	`POST /evaluation_sets`	Request body `processorId` → `entityId`
`GET /evaluation_sets/{id}`	`GET /evaluation_sets/{id}`	Response has `entity` instead of `processorId`
`GET /evaluation_sets/{id}/items`	`GET /evaluation_sets/{id}/items`	Returns summaries (no `expectedOutput`)
—	`GET /evaluation_sets/{id}/items/{itemId}`	New for full item details
`POST /evaluation_set_items`	`POST /evaluation_sets/{id}/items`	Path changed, supports batch
`POST /evaluation_set_items/bulk`	`POST /evaluation_sets/{id}/items`	Consolidated into single endpoint
`POST /evaluation_set_items/{id}`	`POST /evaluation_sets/{id}/items/{itemId}`	Path changed
`DELETE /evaluation_set_items/{id}`	`DELETE /evaluation_sets/{id}/items/{itemId}`	Path changed

Request Changes

Evaluation Sets

Old	New	Notes
`processorId` (query/body)	`entityId`	Renamed

Evaluation Set Items

Old	New	Notes
`POST /evaluation_set_items` with `evaluationSetId` in body	`POST /evaluation_sets/{id}/items`	ID moved to path
Separate single/bulk endpoints	Unified endpoint (1-100 items)	Use `items` array

Response Changes

Response shape changes: Single object responses are now returned directly (no wrapper key), and list responses use { "object": "list", "data": [...] } format. See Simplified Response Shapes for details.

Key Differences

Old	New
`success: true`	(removed) — Use HTTP status codes
`{ "evaluationSet": {...} }`	`{...}` (object returned directly)
`processorId`	`entity` object with full details
`fileId`	`file` object with full details
List items returns full objects	List returns summaries (use retrieve for full)

Example: EvaluationSet Response

After (2026-02-09)

{
  "object": "evaluation_set",
  "id": "ev_abc123",
  "name": "Invoice Test Set",
  "entity": {
    "object": "extractor",
    "id": "ex_xyz789",
    "name": "Invoice Extractor"
  },
  "createdAt": "2024-03-21T15:30:00Z",
  "updatedAt": "2024-03-21T16:45:00Z"
}

SDK Method Reference

Old Method	New Method
`client.batchProcessorRun.get()`	`client.batchProcessorRuns.retrieve()` (deprecated)
—	`client.evaluationSetRuns.retrieve()` (new)
`client.evaluationSet.list()`	`client.evaluationSets.list()`
`client.evaluationSet.get()`	`client.evaluationSets.retrieve()`
`client.evaluationSet.create()`	`client.evaluationSets.create()`
`client.evaluationSetItem.list()`	`client.evaluationSetItems.list()`
—	`client.evaluationSetItems.retrieve()` (new)
`client.evaluationSetItem.create()`	`client.evaluationSetItems.create()`
`client.evaluationSetItem.createBatch()`	`client.evaluationSetItems.create()` (supports arrays)
`client.evaluationSetItem.update()`	`client.evaluationSetItems.update()`
`client.evaluationSetItem.delete()`	`client.evaluationSetItems.delete()`

Detailed Schema Changes

EvaluationSetRun Schema (New)

Property	Old (BatchProcessorRun)	New (EvaluationSetRun)	Change
`object`	`"batch_processor_run"`	`"evaluation_set_run"`	Value changed
`id`	Required `string`	Required `string`	No change
`evaluationSetId`	`sourceId` field	Required `string`	Renamed
`processorId`	Required	—	Removed, see `entity.id`
`entity`	—	Required	New
`entityVersion`	—	Required	New
`source`	Required `"EVAL_SET"`	—	Removed (implicit)
`runCount`	Required	—	Removed

EvaluationSet Schema

Property	Old	New	Change
`processorId`	Required `string`	—	Removed
`entity`	—	Required	New (replaces `processorId`)

EvaluationSetItem Schema

Property	Old	New	Change
`fileId`	Required `string`	—	Removed
`file`	—	Required `FileSummary`	New (replaces `fileId`)

EvaluationSetItemSummary Schema (New)

Returned by list endpoint instead of full items:

Property	Type
`object`	`"evaluation_set_item"`
`id`	Required `string`
`file`	Required `FileSummary`

Need Help?

If you encounter any issues while migrating, please contact our support team at support@extend.app.

Migration Guides

Guide	Migrating From	Migrating To
Overview	—	What’s new and how to upgrade
Extract Runs	`/processor_runs`	`/extract_runs` + `/extract`
Classify Runs	`/processor_runs`	`/classify_runs` + `/classify`
Split Runs	`/processor_runs`	`/split_runs` + `/split`
Parse Runs	`/parse`, `/parse/async`	`/parse_runs` + `/parse`
Edit Runs	`/edit`, `/edit/async`	`/edit_runs` + `/edit`
Extractors	`/processors`	`/extractors`
Classifiers	`/processors`	`/classifiers`
Splitters	`/processors`	`/splitters`
Files	`/files`	`/files` (breaking changes)
Evaluation Sets	evaluation endpoints	Updated evaluation endpoints
Workflow Runs	`/workflow_runs`	`/workflow_runs` (breaking changes)
Webhooks	`processor_run.*` events	`extract_run.`, `classify_run.`, etc.