Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

Human-in-the-loop

Human-in-the-loop (HITL) is a design in which an AI system pauses at chosen points for a person to approve, correct or reject its next step, so risky or uncertain actions are not taken automatically.

Also called: HITL, approval prompts

By Divya Sharma · Checked Nov 15, 2026 · How we review

An agent that can send email, spend money or delete files needs a check before it does so. Human-in-the-loop puts a person at those points.

How it works

The application decides which actions need approval. Common triggers are actions with side effects, such as running a shell command, writing outside a project folder or calling a payment API, and moments when the agent is stuck. At those points the agent stops and shows what it plans to do. The person approves, edits or rejects, and the agent continues.

Frameworks support this directly. LangGraph can pause execution, let a person inspect or edit state and then resume. Coding agents use permission rules: Claude Code, for example, lets teams set which tools and commands are allowed, which are denied and which must ask first. Those rules are covered under agent permissions.

Example

A support agent drafts a refund. Refunds under a set amount go through; larger ones wait for a staff member to click approve.

Limits

Too many prompts and people stop reading them and click yes by habit. A prompt also only helps if it shows enough to judge the action. Pair approvals with an agent sandbox, so that actions run without asking are contained, and remember that prompt injection can make a harmful request look routine.

Related terms

Agent sandbox
An agent sandbox is an isolated environment, such as a container, virtual machine or operating-system restriction, in which an AI agent runs code and commands so it cannot reach files, networks or data outside set limits.
Guardrails (AI)
Guardrails are runtime checks placed around a language model that inspect inputs and outputs and block, rewrite or redirect them, so an application stays on topic, avoids harmful content and follows rules the model alone would not reliably keep.
Agent permissions
Agent permissions are the rules that decide which tools, files, commands, networks and accounts an AI agent may use, and which actions need a human's approval first, so a mistaken or manipulated agent can do limited damage.
Prompt injection
Prompt injection is an attack where text fed to a language model, typed by a user or hidden in a web page, file or tool result, changes the model's behaviour in ways its developer did not intend.
Agent hooks
Agent hooks are user-defined commands or scripts that a coding agent runs automatically at set points in its work, such as before a tool call or after a file edit, to check, block or extend what it does.

Sources