CL
Claude handles this as a draft: Order facts, stated reason, policy clause, and a proposed amount for a person to approve. Do not call Stripe or refund.
Claude · Unverified · activepieces.com · 16 Oct 2026
Loading the catalog…
Claude handles this as a draft: Why escalated, verified facts, unknowns, recommended next action, and records used. Do not message the customer again asking them to repeat what you already have.
Include the records you already have and any failed tool call. The packet is for the person who will finish the job.
The draft should say why it escalated and why it did not take the next action.
You finish with the customer. The agent does not ask them to repeat what you already have, and it does not continue after the escalation flag.
These are editorial labels so you can compare playbooks. They are not measurements.
No. Status is unverified. The InstantDM page was checked on 16 Oct 2026. The prompt is an adaptation, not a vendor quote.
Finishing with the customer. The packet should not hide a failed tool call or ask the customer to repeat facts you already have.
Same InstantDM guide as the agent-run evaluator playbook. The page returned HTTP 200 on 16 Oct 2026. Prompt adapted for AI Playbook Gems; unverified.
Machine-readable copy: JSON
Claude handles this as a draft: Order facts, stated reason, policy clause, and a proposed amount for a person to approve. Do not call Stripe or refund.
Claude · Unverified · activepieces.com · 16 Oct 2026
Claude handles this as a draft: Quote the request. Do not decide the roadmap. Do not tell the customer it is planned.
Claude · Unverified · anthropic.com · 15 Oct 2026
Claude handles this as a draft: Who is blocked and the last timestamp. Do not escalate further.
Claude · Unverified · anthropic.com · 13 Oct 2026