/cad-spike

Spike - a time-boxed experiment resolving one unknown before you bet on it, risk-first, with a validated | invalidated | inconclusive verdict

Boost this skill

crenshawdev@crenshawdev

GitHub

From plugin

cadence

534 skills30 agents3 hooks

Install

$ npx -y skills add crenshawdev/cadence --skill cad-spike --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.
You can call itInvoke it directly when you want it.
Slash command/cad-spike

Context preview

The summary Claude sees to decide when to auto-load this skill.

Spike - a time-boxed experiment resolving one unknown before you bet on it, risk-first, with a validated | invalidated | inconclusive verdict

SKILL.md

cad-spike.SKILL.md

name: cad-spike
description: "Spike - a time-boxed experiment resolving one unknown before you bet on it, risk-first, with a validated | invalidated | inconclusive verdict"
argument-hint: "<the question or hypothesis to resolve>"
allowed-tools:
  - Read
  - Write
  - Edit
  - Bash
  - Grep
  - Glob
  - AskUserQuestion

<objective> Answer one unknown with evidence before a plan commits to it - "will this approach work", "is this library fast enough", "does this API do what we need". The discipline is what makes it useful: falsifiable criteria written BEFORE the experiment, the riskiest assumption tested first so a dead end dies fast, and an honest verdict. The code is throwaway; the SPIKE.md verdict is the deliverable. </objective>

<execution_context> @${CLAUDE_PLUGIN_ROOT}/cadence-core/workflows/spike.md </execution_context>

<process> Run the spike workflow end-to-end. Write the Given/When/Then criteria before building anything, order them risk-first, and stop the moment a risk-first criterion invalidates the approach. Never soften an invalidated result into "inconclusive" to save the idea - a spike that kills a bad approach is a success. </process>

Read it on GitHub ↗

Ships withcadence

Appearance is cheap. Verification is the work. Cadence is for developers using Claude Code on software they will still own after the session ends. Claude can write a convincing plan, produce working code, and tell you the job is finished.

Get the whole plugin

Stats

JavaScript Node Bash Ruby Git

+4 tags

GitHub Anthropic OpenAI Gemini

Stars

Forks

Active

Maintenance

JavaScript

Language

MIT