Router node

The Router node branches the flow. It evaluates conditions that live on its outgoing edges and follows exactly one matching branch.

How it works

1. The Router receives a payload.
2. Each outgoing edge carries a condition (path, operator, value).
3. The first edge whose condition matches is taken.
4. If none match, the unconditional/default edge is taken.
5. If there is no matching edge and no default, the run fails with a graph error.

Conditions

Conditions are configured in the edge inspector. Available operators:

Operator	Meaning
eq / neq	equals / not equals (coerced)
gt / gte	greater than / or equal
lt / lte	less than / or equal
contains	string contains
startsWith	string prefix
endsWith	string suffix
regex	matches a regular expression

The path is a dot-path into the payload, e.g. label or outputs.agent.label.

Example — route on a classification

An Agent outputs { "label": "refund", "confidence": 0.9 }. Wire it into a Router with two edges:

• Edge A: label eq "refund" → a Refund HTTP Request node.
• Edge B: label eq "support" → a Support Agent node.
• Edge C (no condition): fallback → an Output node.

Notes

• Numeric comparisons (gt/lt) coerce strings to numbers first, so "10" compares as 10, not lexicographically.
• Exactly one branch runs per execution — a Router never fans out.

Related

• Agent — structured output that a Router can branch on.
• HTTP Request — a common branch target.
• Use cases — a complete lead-scoring flow built on a Router.