AI
A Connection node calls Anthropic mid-decision with the credentials your own contract issued. Whatever comes back is data the rest of the flow can read, branch on, and keep in the decision's trace.
What a flow can call
/v1/messagesOne request, one answer. max_tokens is required and is a hard ceiling on the reply — if it is hit the response comes back truncated with stop_reason "max_tokens" rather than failing.
/v1/messagesSame endpoint, constrained to a JSON schema so the reply parses without a regex.
/v1/messages/count_tokensReturns the input token count for a request without running it.
/v1/modelsModel ids, context windows and capabilities as the account sees them.
Provider documentation: https://platform.claude.com/docs/en/api/messages
Where it sits in the decision
The call runs where the policy needs it — after the cheap checks that might make it unnecessary, before the rules that depend on it. Calls that do not depend on each other run together rather than in sequence.
Because it is part of the run, the response is part of the record. When someone asks why an applicant was declined months later, what Anthropic returned at the time is in the trace, not re-fetched from a service whose answer has since changed.
Ready when you are
Build the flow in Sandbox, connect your account, and watch the decision pull what it needs before it answers.