skills.rest
cursor
cursorOFFICIAL
Explore all skills by this author →
WebsiteGitHub

principle-never-block-on-the-human

Guides agents to proceed on reversible work instead of pausing for human confirmation.

6.6k|542|Updated Jan 23, 2026
Search Tags:#confirmation#Workflow#agent autonomy#Code review#async supervision#decision principle#reversible actions
One-click install
npx skills add https://github.com/cursor/plugins --skill principle-never-block-on-the-human
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: principle-never-block-on-the-human
Source: https://github.com/cursor/plugins/tree/main/pstack/skills/principle-never-block-on-the-human
Command: npx skills add https://github.com/cursor/plugins --skill principle-never-block-on-the-human

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Agents that stop to ask "should I do X?" on every reversible decision stall pipelines and make the human the bottleneck. This Skill encodes a decision principle so agents keep working and let humans course-correct asynchronously.

Core Features & Use Cases

  • Proceed-then-present pattern: Do the work, show the result, and explain the reasoning instead of asking for permission first.
  • Reversibility-based boundaries: Distinguishes reversible actions (write code, edit notes, split tasks) from irreversible ones (force-push, delete production data, send external messages) that still require confirmation.
  • Use Case: During a long autonomous coding session, the agent encounters an ambiguous refactoring choice. Instead of blocking, it picks a reasonable approach, logs the decision, and presents the diff for later human review.

Quick Start

Apply the never-block-on-the-human principle so the agent proceeds on reversible work and only asks before irreversible actions.

Frequently Asked Questions about principle-never-block-on-the-human

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I stop an AI agent from asking too many questions?▼

Apply a proceed-then-present principle: the agent makes reasonable decisions on reversible work, shows the result, and explains its reasoning. Questions are reserved for genuine ambiguity where intent cannot be inferred from context.

When should an autonomous agent still ask for confirmation?▼

Confirmation is required for irreversible actions such as force-pushing, deleting production data, or sending external messages. Reversible actions like writing code, editing notes, or splitting tasks should proceed without blocking.

What is asynchronous human supervision in agent workflows?▼

Asynchronous supervision means the human reviews plans, diffs, and changes on their own schedule rather than approving each step live. Workflows are designed for review-after-the-fact so agents never sit idle waiting.

What are the limits of the never-block-on-the-human principle?▼

The principle only covers execution, not product direction, which still comes from the human. It also does not apply to irreversible or externally visible actions, which always require explicit confirmation before proceeding.

Related Skills

Explore More Software Engineering Skills→
TeamFlint-DevTeamFlint-Dev

Autonomy Boundaries Skill

Categorize autonomous agent tasks into Green, Red, or Yellow zones.

Official
Intermediate
tta-labtta-lab
23

sp-complete-design

Self-review a written design plan for executability and route it for review.

Official
Intermediate
kehwarkehwar

dev-log

Record development decisions, bugs, and insights as Markdown files under docs/development-log/.

Community
Basic
MoRen9527MoRen9527

Lobster Workflow

Execute multi-step workflows with approval checkpoints via a typed JSON-first pipeline.

Community
Intermediate