Execution Logs
Inspect the input, output, and timing of every node in a workflow execution.
Opening the Logs
From the Activity page, click any row in the Activity Logs table. This opens the workflow in the editor with the Logs tab already active. You can also switch between the AI Builder and Logs tabs at the top of the editor at any time.
Log View
The Logs panel shows a vertical list of every node that ran during the execution, in order. Each row displays:
| Column | Description |
|---|---|
| Node icon | The node type (Flow Start, Code, Condition, Query Data, etc.) |
| Node name | The label you gave the node in the editor |
| Execution time | How long that specific node took to run (shown on the right in µs, ms, or s) |
Nodes that failed are marked with a red indicator. This makes it easy to scan the list and spot where things went wrong.
Inspecting a Node
Click any node row to expand it. You'll see two panels side by side:
- Input - The data that entered the node
- Output - The data the node produced
Each panel shows the number of lines in the payload (e.g. "8 lines", "6 lines").
View Modes
Both the Input and Output panels support three view modes, toggled by the icons in the top-right corner of each panel:
| View | Icon | Description |
|---|---|---|
| Raw JSON | { } |
The raw JSON payload, syntax-highlighted. Best for copying or reading the full structure. |
| Tree View | tree icon | Collapsible tree with expandable keys. Useful for navigating deeply nested objects without scrolling through raw text. |
| Table View | grid icon | Key-value table layout. Works well for flat objects or when you want a quick overview of field names and values. |
Copy to Clipboard
Each panel has a copy button (clipboard icon) next to the view mode toggles. Click it to copy the full raw JSON of that panel to your clipboard - useful for pasting into a bug report, sharing with a teammate, or testing in an external tool.
Reading the Data
Flow Start Node
The Input panel of the Flow Start node shows the original request that triggered the workflow. For an API-triggered workflow, this includes:
- The request body (parsed JSON)
- Headers (Content-Type, Accept, etc.)
- API configuration (method, custom path, rate limit, timeout)
The Output panel shows the same data reformatted as the variables available to downstream nodes.
Code Nodes
- Input - The variables and data available when the code ran
- Output - The
resultobject returned by the code
Condition Nodes
- Input - The condition expression and the values being compared
- Output - Which branch was taken (IF or ELSE)
Query Data / Write Data Nodes
- Input - The query filters, table name, and parameters
- Output - The rows returned (Query) or the write confirmation (Write)
HTTP Request Nodes
- Input - The URL, method, headers, and body sent
- Output - The response status, headers, and body received
Debugging Tips
Find the failing node fast
Scroll through the log list and look for the red indicator. Click it to see what input caused the failure and what error message was returned in the output.
Compare input vs output
If a Code node produces unexpected results, expand it and compare the Input (what went in) with the Output (what came out). This usually reveals whether the issue is in the incoming data or in the code logic.
Check execution time
If a workflow is slow, scan the execution times on the right side of each row. The node with the highest time is your bottleneck - usually an HTTP Request or Query Data node waiting on an external service.
Copy and test externally
Use the copy button on an HTTP Request node's Input to grab the exact URL, headers, and body. Paste it into curl or Postman to test the request in isolation.