Skip to main content
GET
/
v1
/
workflows
/
artifacts
/
{operation}
/
{artifact_id}
from retab import Retab

client = Retab()

step = client.workflows.runs.steps.get("run_abc123xyz", "hil-block-1")

if step.artifact:
    artifact = client.workflows.artifacts.get(step.artifact)
    print(artifact.operation)
    print(artifact.requires_human_review)
    print(artifact.matched_condition_ids)
    print(artifact.evaluations)

{
  "operation": "hil_evaluation",
  "id": "heval_abc123",
  "organization_id": "org_123",
  "workflow_run_id": "run_abc123xyz",
  "step_id": "hil-block-1",
  "requires_human_review": true,
  "selected_handles": ["wrong"],
  "matched_branch_id": "branch_wrong",
  "matched_condition_ids": ["condition_wrong_total"],
  "evaluations": [
    {
      "condition_id": "condition_wrong_total",
      "matched": true,
      "output_handle_id": "wrong",
      "left_value": 1200,
      "operator": "greater_than",
      "right_value": 1000
    }
  ],
  "reviewer_id": null,
  "review_decision": null,
  "review_notes": null,
  "reviewed_at": null,
  "created_at": "2026-03-12T09:00:04Z"
}

Documentation Index

Fetch the complete documentation index at: https://docs.retab.com/llms.txt

Use this file to discover all available pages before exploring further.

Dereference a workflow step artifact ref into its persisted record. Workflow steps return artifact as a small { operation, id } pointer. This endpoint follows that pointer and returns the backing record flattened with operation and id at the top level. Use this when you need details that do not belong in the generic step payload: HIL condition evaluations, conditional branch matches, function output, API-call attempts, while-loop termination data, or the full primitive result for extract/split/classify/parse/edit/partition blocks. 
from retab import Retab

client = Retab()

step = client.workflows.runs.steps.get("run_abc123xyz", "hil-block-1")

if step.artifact:
    artifact = client.workflows.artifacts.get(step.artifact)
    print(artifact.operation)
    print(artifact.requires_human_review)
    print(artifact.matched_condition_ids)
    print(artifact.evaluations)

{
  "operation": "hil_evaluation",
  "id": "heval_abc123",
  "organization_id": "org_123",
  "workflow_run_id": "run_abc123xyz",
  "step_id": "hil-block-1",
  "requires_human_review": true,
  "selected_handles": ["wrong"],
  "matched_branch_id": "branch_wrong",
  "matched_condition_ids": ["condition_wrong_total"],
  "evaluations": [
    {
      "condition_id": "condition_wrong_total",
      "matched": true,
      "output_handle_id": "wrong",
      "left_value": 1200,
      "operator": "greater_than",
      "right_value": 1000
    }
  ],
  "reviewer_id": null,
  "review_decision": null,
  "review_notes": null,
  "reviewed_at": null,
  "created_at": "2026-03-12T09:00:04Z"
}
Artifact operations currently include:
operationtypical produceruseful fields
extractionextractextracted output, consensus choices, likelihoods
splitsplitsplit groups and output documents
classificationclassifierselected class and consensus details
parseparseparsed pages/content
editeditedited document result
partitionfor_each_sentinel_startpartitioned loop inputs
conditional_evaluationconditionalevaluations, selected_handles, matched_condition_ids
hil_evaluationhilevaluation details plus human-review state
while_loop_terminationwhile_looptermination_reason and final evaluations
api_call_invocationapi_callretry attempts, response data, final error
function_invocationfunctionfunction inputs, output, duration_ms, final error

Authorizations

​
Api-Key
string
header
required

Path Parameters

​
operation
enum<string>
required
Available options:
extraction,
split,
classification,
parse,
edit,
partition,
conditional_evaluation,
hil_evaluation,
while_loop_termination,
api_call_invocation,
function_invocation
​
artifact_id
string
required

Query Parameters

​
access_token
string | null

Response

Successful Response

The response is of type Response Get Workflow Artifact V1 Workflows Artifacts Operation Artifact Id Get · object.