> ## Documentation Index
> Fetch the complete documentation index at: https://docs.anyreach.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Observability
> Inspect workflow runs, debug failures, and monitor production health.
Every workflow execution leaves a complete trace: per-step inputs and outputs, durations, and error messages. All of this is visible on the **Workflow Runs** page.
## The Workflow Runs page
Click the **Logs icon** in the top bar of the builder to open the Workflow Runs page.
The runs table shows:
| Column | Meaning |
| ---------------- | ----------------------------------------------------------- |
| **ID** | Unique execution identifier |
| **Version** | The workflow version that ran |
| **Created At** | When the execution started |
| **Completed At** | When the execution finished |
| **Duration** | Wall-clock time |
| **Status** | `success`, `failed`, `in_progress`, `waiting`, or `pending` |
| **Actions** | Click to view run details |
Use the **Search**, **Filters**, **Columns**, and **Refresh** controls at the top to narrow the list.
## Viewing run details
Click **Actions** on any run to open its detail view. For each step that ran, you can see:
* **Status** — whether the step succeeded, failed, or was skipped
* **Input** — the `ctx` snapshot that was passed into the step
* **Output** — what the step added to `ctx`
* **Duration** — how long the step took
* **Error** — full error message and traceback (for Code steps)
This is the primary debugging surface. Most workflow bugs are visible by inspecting the input and output of the failing step.
## API access
Programmatic access to runs:
```bash theme={null}
# List recent runs for a workflow
curl https://api.anyreach.ai/workflow/workflow-executions?workflow_id=$WORKFLOW_ID \
-H "Authorization: Bearer $ANYREACH_TOKEN"
# Get a specific run
curl https://api.anyreach.ai/workflow/workflow-executions/$EXEC_ID \
-H "Authorization: Bearer $ANYREACH_TOKEN"
```
The response includes the full context (every step's input/output) — the same data shown in the UI.
## Logging from steps
There's no separate log output facility. Use these patterns:
* **Code steps** — anything you `print()` is captured in the step's output and visible in the run detail
* **HTTP API steps** — the full request and response are captured automatically
* **Custom telemetry** — add a Code step late in the workflow to post metrics to your own monitoring system