What happens to the code written during a spike?

It follows a 'Burn After Reading' mandate—it is either deleted or kept in an isolated branch marked 'DO NOT MERGE' to prevent experimental code from reaching production.

When is a spike considered complete?

A spike is finished the moment the specific objective or success criteria are reached, rather than when a specific time limit expires.

What is the main goal of a Technical Spike?

The primary goal is to answer a specific high-level question and reduce technical uncertainty, producing a knowledge artifact rather than production-ready code.

How does this differ from a standard feature task?

Spikes prioritize speed and learning over code quality. Spike code is 'throwaway' and intended to be deleted or abandoned once the answer is found, whereas feature tasks aim for production-grade code.

What kind of artifacts does this skill produce?

It generates structured markdown reports containing a final verdict, performance metrics, architectural drafts, or root-cause explanations.

Technical Spike

Name: Technical Spike
Author: johannesfritz

byjohannesfritz

0•

Productivity & Workflow

Conducts objective-driven investigations to reduce technical uncertainty and validate architectural decisions before implementation.

The Technical Spike skill empowers Claude to perform focused, objective-driven investigations that eliminate 'unknown-unknowns' in the development process. Unlike traditional time-boxed spikes, this skill follows a philosophy where the investigation concludes only once a specific question is answered, producing high-value knowledge artifacts like feasibility verdicts or architecture drafts rather than production-ready code. It is essential for teams needing to validate performance, explore library capabilities, or identify the root causes of complex bugs without cluttering the main codebase with experimental scripts.

Key Features

01Burn-after-reading protocol ensuring experimental code is never merged into main

02Objective-driven investigation framework that finishes when the answer is found

03Automated reporting templates for consistent documentation and recommendations

040 GitHub stars

05Classification system for technical (feasibility) vs. functional (requirements) spikes

06Standardized knowledge artifacts including feasibility verdicts and architecture drafts

Use Cases

01Determining the root cause of non-deterministic bugs or complex race conditions

02Validating architectural patterns and scaling assumptions before committing to a build

03Evaluating if a specific library or database can meet strict performance requirements

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add johannesfritz/autonomous-orchestration technical-spike

For use in Claude.ai and ChatGPT

Download Skill