SecondMind

← Community

Keep a human in the loop

ai-agents by Admin 3 min read 1 file

Decide where a model may draft and where it may not act — money, messages, deletes, anything a person cannot undo — instead of "the agent handles it." Use when wiring tools, before an agent sends or pays, or when a demo is about to become a product.

---
name: keep-a-human-in-the-loop
category: ai-agents
description: >-
  Decide where a model may draft and where it may not act — money, messages,
  deletes, anything a person cannot undo — instead of "the agent handles
  it." Use when wiring tools, before an agent sends or pays, or when a
  demo is about to become a product.
---

# Keep a Human in the Loop

A model that can **draft** is a writing aid. A model that can **act** is
an employee with no memory of last quarter's lawsuit. The skill is drawing
that line on purpose, per action, not as a vibe in the system prompt.

"Log everything" is not a human in the loop. A log is how you write the
incident. A human is someone who can stop the action **before** it
reaches a customer, a ledger, or a delete.

## Sort actions by blast radius

You are reading the opening of this skill.

The rest is open to any signed-in account — free included. Reading the library costs nothing.

Join the waitlist Sign in

Shared publicly by the author and open for any use. See the Terms.