spike

Validate the riskiest feature unknown with a throwaway implementation and SPIKE.md verdict.

2|Updated May 13, 2026

Search Tags:#proof of concept #feature planning #feasibility #spike #throwaway code #risk validation

Ready to run

npx skills add https://github.com/mujtaba3B/gstack-extensions --skill spike-mujtaba3b

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: spike
Source: https://github.com/mujtaba3B/gstack-extensions/tree/main/eng/skills/spike
Command: npx skills add https://github.com/mujtaba3B/gstack-extensions --skill spike-mujtaba3b

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you answer the single riskiest question in a feature before you commit time to planning, scaffolding, or production-quality implementation.

Core Features & Use Cases

Mechanic-level validation: Confirms whether one specific capability is even possible with the smallest possible throwaway code.
Structured experimentation: Guides a four-phase workflow for framing the outcome, isolating work, recording findings in SPIKE.md, and landing a verdict.
Escalation when blocked: Encourages a second opinion and user input instead of wasting time on repeated failed attempts.
Use case: You have an uncertain integration, UI mechanic, or API behavior and want to test feasibility before building the full feature.

Quick Start

Use the spike skill to test the riskiest part of this feature with the smallest throwaway implementation and record the verdict in SPIKE.md.

Frequently Asked Questions about spike

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

FAQPage Schema

What is a spike in software development and when should I use it?▼

A spike validates the riskiest unknown of a feature using the smallest throwaway implementation before committing to planning or production work. Use it to test uncertain integrations, API behaviors, or UI mechanics.

How do I run a feasibility check for a risky feature using throwaway code?▼

Define a one-line outcome, isolate your work in a dedicated spike branch or worktree, and record findings in a live SPIKE.md ledger. Complete the feasibility check by landing a final verdict before cleaning up the throwaway code.

What should I do when my feature spike or proof of concept gets blocked?▼

When your feature spike gets blocked, you should escalate by seeking a second opinion and requesting user input. This prevents wasting time on repeated failed attempts during risk validation.

Can I test a mechanic-level experiment without setting up a full production environment?▼

Yes, you can test mechanic-level experiments with the smallest throwaway implementation in an isolated spike branch or workwork. This avoids production setup overhead while confirming whether a specific capability is possible.

How does a spike differ from building a standard feature prototype?▼

A spike focuses solely on validating the single riskiest unknown with minimal throwaway code, whereas a prototype typically explores broader user flows. Spikes require a recorded verdict in a ledger before cleanup, ensuring focused risk validation.

Why do I need to record a verdict in SPIKE.md after a feasibility spike?▼

Recording a final verdict in SPIKE.md ensures the outcome of your risk validation is documented before cleanup. This structured ledger prevents lost findings and informs whether to proceed with full feature planning.