stepback

Reframe current work with one abstract step-back question and validate scope, side effects, and approach.

Updated Apr 14, 2026
One-click install
npx skills add https://github.com/sr-ai-dev/sr-harness --skill stepback-sr-ai-dev
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: stepback
Source: https://github.com/sr-ai-dev/sr-harness/tree/main/skills/stepback
Command: npx skills add https://github.com/sr-ai-dev/sr-harness --skill stepback-sr-ai-dev

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps you recover from drifting or misguided work by forcing an instant, higher-level reframing and quick validation of scope, side effects, and whether a better approach exists.

Core Features & Use Cases

  • One-shot perspective reset: Scans what you have been doing and generates exactly one abstract reframing question to restore focus on the right problem.
  • Three rapid checks: Verifies scope drift, potential side effects, and whether a more fundamental approach might be available before you continue.
  • Mid-work, low-friction: Designed for interruptions during active development without starting dialogue, spawning agents, or generating code.

Use it when you suspect you are “fixing symptoms,” expanding scope unintentionally, or committing to a path that might have a simpler root solution—e.g., after hours of debugging, refactoring, or implementing a feature while the overall direction feels off.

Quick Start

Type /stepback to instantly reframe your current work and validate scope, side effects, and approach in under 10 lines.

Frequently Asked Questions about stepback

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

FAQPage Schema
How do I catch scope drift and side effects during active code iterations?

To catch scope drift during active code iterations, trigger a one-shot perspective reset that validates scope, side effects, and alternative approaches in under 10 lines. It forces a higher-level reframing question to restore focus on the right problem without generating code or starting dialogue.

What is the best way to reframe my current work when fixing symptoms instead of root causes?

The best way to reframe current work when fixing symptoms is to apply a mid-work perspective reset that generates a single abstract step-back question. This validates whether a more fundamental solution exists before you continue down a misguided path.

Why does my developer workflow keep expanding scope unintentionally during refactoring?

Your developer workflow expands scope unintentionally when direction drifts without high-level validation. A rapid three-step check verifies scope drift, potential side effects, and whether a better approach exists, preventing you from committing to the wrong path.

Can I validate a better approach for a feature implementation without interrupting my flow?

Yes, you can validate a better approach without interrupting flow by running a low-friction mid-work check. It scans your current work and outputs under 10 lines, verifying scope drift, side effects, and alternative solutions without spawning agents or requiring manual prompting.

When should I not use a step-back perspective reset?

You should not use a step-back perspective reset when you need executable code, extended dialogue, or detailed agent execution, because it strictly produces a one-shot abstract reframing question and validation checks without generating code or starting conversations.

Does the stepback reframing tool require any dependencies or specific environment setup?

No, the stepback reframing tool requires no dependencies or specific environment setup. It operates as a standalone one-shot command to instantly reframe work and validate scope, side effects, and approach without external components.