get_node_run
Read one node’s full output from a workflow run.
get_run_status deliberately withholds intermediate node outputs — they can be very large — and returns full output only for failed nodes and the workflow’s output node. Use get_node_run when you need exactly one node’s result: to explain what a workflow posted, or to inspect the step before a failure.
Parameters#
| Name | Type | Required | Default | Description |
|---|---|---|---|---|
run_id | string | yes | Run id (agent_run_…) | |
node_id | string | yes | Graph node id, from get_run_status → node_runs[].node_id |
Returns#
json
{
"run_id": "agent_run_xyz",
"node_id": "n_7",
"node_name": "Post summary",
"node_type": "tool_call",
"url": "https://app.adlyse.com/run/agent_run_xyz",
"node_runs": [
{
"node_run_id": "agent_node_run_abc",
"status": "completed",
"scope": null,
"output": { "message_ts": "1712345678.000100", "channel": "C123ABC456" },
"output_variables": [
{ "name": "message_ts", "type": "string", "value": "1712345678.000100" }
],
"events": [
{ "level": "success", "message": "Execution completed", "error": null, "created_at": "2026-04-22T14:00:07Z" }
],
"duration_seconds": 1.2,
"input_tokens": 0,
"output_tokens": 0,
"completed_at": "2026-04-22T14:00:07Z"
}
],
"node_run_count": 1
}| Field | Description |
|---|---|
node_runs | One entry per execution. A node inside a loop runs once per iteration |
node_runs[].scope | Loop-iteration key, or null outside a loop |
node_runs[].output_variables | The node’s declared output variables, resolved against that run’s output |
node_runs[].events | Execution log for that run (info, warn, error, success) |
node_runs[].error | Present only when that run’s status is failed |
total_node_runs | Present only when capped — the true number of iterations |
truncation_note | Present only when capped, at the top level and/or inside an oversized output |
Limits#
A looping node can run many times and each output can be large, so the response is bounded:
- at most 20 iterations, the earliest first (
total_node_runs+truncation_noteappear when there are more) - each
outputcapped at 20,000 characters; an oversized one is replaced by{ "truncated": true, "truncation_note": …, "output_head": … }
Both notes state that you are seeing the head of the data, not a summary of it — don’t total a truncated result and report it as the whole.
Errors#
error_type | When |
|---|---|
auth_error | Missing / invalid API key |
not_found | Run id doesn’t exist, belongs to another org, or the node never ran |
Example#
Request: { "run_id": "agent_run_xyz", "node_id": "n_7" }
Response: see “Returns” above.
See also#
get_run_status— the run’s overall status and node listget_workflow— the graph, to map a node id to its place in the workflow