Agents

Human Handoff Policy

When the agent should stop and escalate.

The Prompt

Design the escalation policy for an agent that [task].
Define:
1. HARD stops (always escalate): e.g., money movement, legal language, angry user — give your list
2. SOFT stops (ask once, then proceed if user insists)
3. The exact handoff message: what context to pass so the human doesn't re-ask everything
4. What the agent may do while waiting

 Fill in before using

[task]

How to use it

→ Paste the prompt into your model of choice — it works with any frontier model.
→ Replace every bracketed placeholder with your specifics. Concrete inputs beat vague ones every time.
→ If the model asks a clarifying question, answer it — these prompts are designed to invite that.
→ Iterate on the output, not the prompt. The structure is already tuned.

production agentssafety

More Agents prompts

Agent System Prompt Skeleton

A battle-tested structure for tool-using agents.

Tool-Use Critic

Audit an agent trace for wasted calls and loops.

Multi-Agent Task Decomposer

Split a job across agents without overlap or gaps.