Human in the Loop

Use when designing or verifying auditable human oversight, approval gates, escalation paths, and safe state transitions for AI agent workflows.

9 stars 3 forksAI, LLMs & Agents

Install

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add sandbaseai/workbuddy-skill --skill human-in-the-loop
Manual install
git clone --depth 1 https://github.com/sandbaseai/workbuddy-skill.git /tmp/workbuddy-skill
mkdir -p ~/.claude/skills
cp -r /tmp/workbuddy-skill/skills/human-in-the-loop ~/.claude/skills/human-in-the-loop

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# Human in the Loop

Place human judgment where it changes risk. A confirmation dialog alone is not
oversight: bind an authorized decision to an understandable, immutable action
and preserve evidence of what happened. Designing a gate never authorizes the
underlying action.

## Inventory and risk tiers

List every Agent action, target object, effect, reversibility, scale,
sensitivity, external visibility, value, time pressure, affected rights, and
containment option. Separate reading, drafting, previewing, and recommending
from sending, publishing, purchasing, deleting, granting access, executing
code, or other consequential effects.

Assign the lightest sufficient control:

- **Autonomous with audit:** bounded, reversible, low-impact actions.
- **Notify after action:** low-impact actions needing awareness.
- **Review before action:** consequential or externally visible actions.
- **Step-up approval:** a value, sensitivity, anomaly, confidence, or scope
  threshold is crossed.
- **Dual control:** critical, irreversible, privileged, or regulated actions;
  require distinct approver subjects by default.
- **Prohibited:** the action cannot be acceptably safe.

Do not gate trivial steps until users rubber-stamp everything. Never let the
model fabricate, infer, or impersonate human approval.

## Approval contract

Define predicates, eligible roles and distinct subjects, quorum, evidence,
expiry, timeout, escalation, audit-outage behavior, execution-time
Read the full skill on GitHub

Details

Skill name
human-in-the-loop
Path
skills/human-in-the-loop/SKILL.md
License
MIT
Repo updated
8 September 2026

FAQ

What does the Human in the Loop skill do?

Use when designing or verifying auditable human oversight, approval gates, escalation paths, and safe state transitions for AI agent workflows.

How do I install the Human in the Loop skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add sandbaseai/workbuddy-skill --skill human-in-the-loop. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Human in the Loop popular?

It lives in sandbaseai/workbuddy-skill, which has 9 GitHub stars and 3 forks.

Related skills

Browse the most popular Claude skills, all ai, llms & agents skills, or write your own with the skills generator.