n8n-validation-expert

Run validate_node with the runtime profile, read each error message carefully, fix errors one at a time, and re-validate. Expect 2-3 iterations; use get_node to check which fields are required before configuring.

Errors like missing_required or type_mismatch block workflow execution and must be fixed. Warnings like best_practice or performance do not block execution and may be acceptable depending on context, such as development workflows or low-volume APIs.

Use runtime for most pre-deployment validation since it balances coverage and noise. Use ai-friendly to reduce false positives by about 60 percent, minimal for quick checks, and strict only for production-critical workflows.

Most invalid_expression errors come from missing curly braces around expressions, typos in node names, or accessing webhook data without the .body prefix. Wrap expressions in {{}} and verify referenced node names exist in the workflow.

Yes, auto-sanitization runs on every workflow save and fixes operator structure issues, such as removing singleValue from binary operators or adding it to unary operators, plus IF and Switch metadata. It cannot fix broken connections or branch count mismatches.

Warnings are often acceptable for development workflows, internal APIs without rate limits, idempotent operations, and small known datasets. Always fix security warnings and document why any accepted warning is safe for your use case.