# Execution trace

> Everything needed to debug one run offline: the trigger input, the run record, the logs, and the step-by-step thread.

Source: https://docs.generalinput.com/api/workflows/execution-trace



<Endpoint method="GET" path="/v1/workflows/:id/executions/:executionId/trace" />

<Access scope="workflows:manage" />

The bundle the CLI writes to disk with `geni workflow execution --download`. For an agent workflow, `thread` is the agent's conversation, every tool call and result.

## Response [#response]

```json
{
  "id": "exec_01HX…",
  "workflowId": "wf_01HX…",
  "status": "failed",
  "error": "Slack 429: rate limited",
  "startedAt": "2026-09-03T10:00:02.000Z",
  "completedAt": "2026-09-03T10:03:11.000Z",
  "input": { "days": 7 },
  "run": null,
  "logs": "…",
  "thread": [
    {
      "role": "assistant",
      "type": "tool-call",
      "content": { "name": "bash", "args": {} },
      "createdAt": "…"
    },
    { "role": "tool", "type": "tool-result", "content": "…", "createdAt": "…" }
  ]
}
```

| Field    | Notes                                                                                                           |
| -------- | --------------------------------------------------------------------------------------------------------------- |
| `input`  | The trigger payload that started the run.                                                                       |
| `thread` | Steps for a code run, the agent conversation for an agent run. `type` is `text`, `tool-call`, or `tool-result`. |