Data Flow Spy Node
The Data Flow Spy Node lets you inspect, debug, and visualize the data flowing through your workflow at any point. It's a pure display node — no Run button, no output port — that reacts to its input in real time.
Output Dialog
Opening the node's output shows two tabs:
- Payload: the incoming
payload. - Tools Payload: the incoming
toolsPayload.
Each tab renders independently, in a readable Markdown/JSON format.
Supported Data Formats
Each tab accepts any data type:
- Strings: Displayed as plain text or Markdown.
- Numbers: Displayed as-is.
- Objects/Arrays: Rendered as formatted JSON blocks.
- Markdown: Rendered with full Markdown support (headings, lists, code, etc).
Example Usage
For example usage, see the Current Time workflow or Weather Dashboard workflow, which demonstrate using Data Flow Spy nodes to inspect intermediate results.
Common Use Cases
- Debugging: Inspect the output of any node in your workflow.
- Data Inspection: View raw or formatted data at any stage.
- Markdown Preview: Render Markdown output from LLMs or other nodes.
- Error Tracing: Quickly spot issues in data transformations.
- Workflow Development: Rapidly iterate by checking intermediate results.
Best Practices
- Place after key nodes: Add Data Flow Spy nodes after important processing steps to verify outputs.
- Use for troubleshooting: Temporarily insert to debug unexpected results.
- Preview Markdown: Use to check LLM-generated Markdown or JSON.
- Remove for production: Remove or disable unnecessary spies in finalized workflows for clarity.
Troubleshooting
Common Issues
- Unreadable output: Non-JSON objects are rendered as best-effort Markdown or code blocks.
- Markdown not rendered: Ensure your data is a valid Markdown string or JSON.
- Nothing in the Tools Payload tab: Not every upstream node sends a
toolsPayload— an empty tab usually just means none was provided.
Supported Data Types
- String (including Markdown)
- Number
- Boolean
- Object (including arrays)
- Null / Undefined
Performance Tips
- Avoid sending extremely large data blobs for inspection.
- Use JSON Reformatter nodes to trim or format data before spying.
- For large arrays/objects, consider inspecting only a subset.