budget

Overview ¶

This sample demonstrates the experimental Budget middleware, which caps the tokens a generate run spends across its whole tool loop.

The model follows a chain of boxes to a treasure. Each box names the next, so every box costs a turn, and a small budget runs out before the end:

• huntFlow stops the run once it spends maxTokens.
• huntWithApprovalFlow pauses the run instead, and approves another maxTokens up to "approvals" times.

Run it:

go run .

Or with the Dev UI, to watch each turn and its usage at http://localhost:4000/traces:

curl -sL cli.genkit.dev | bash    # install the Genkit CLI, once
genkit start -- go run .

Or over HTTP:

curl -X POST 'http://localhost:8080/huntWithApprovalFlow' \
  -H "Content-Type: application/json" \
  -d '{"data": {"maxTokens": 3000, "approvals": 3}}'