Validate a pipeline
Understand graph validation and how to respond to configuration or structure errors.
Validation checks whether Ragify can accept a graph's supported structure and configuration. It is a useful design review step, but it is not the same as running the workflow.
When to validate #
Validate after adding or reconnecting nodes, changing a retrieval strategy, or modifying node settings. A graph that was valid before editing may become invalid after a structural change.
What validation can help identify #
The frontend validator can return structured feedback for supported graph and configuration problems, including missing or incompatible connections and invalid node arrangements. Messages should guide you back to the relevant part of the editor.
How to resolve an issue #
Read the message in full. Identify the affected node or connection. Open that node's settings or return to the canvas to repair the relationship, then validate again. Avoid changing unrelated stages until you understand the reported issue.
What a passing result does not establish #
Passing validation does not prove that source documents are accessible, model credentials are valid, embeddings match a live vector store, retrieval is useful, an API is authorized, or generated answers are correct. Those are runtime concerns beyond the current frontend.
If the graph seems correct but still fails #
Review connection direction, the linear-chain requirement, the chosen retrieval strategy, and node-specific required fields. See Troubleshooting.