spike

Build throwaway prototypes to validate feasibility and report a structured verdict.

Updated Jun 19, 2026
One-click install
npx skills add https://github.com/AmirulAndalib/Vilvona-AI --skill spike-amirulandalib
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: spike
Source: https://github.com/AmirulAndalib/Vilvona-AI/tree/main/skills/spike
Command: npx skills add https://github.com/AmirulAndalib/Vilvona-AI --skill spike-amirulandalib

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Before committing to a real build, teams often need to know whether an idea is technically feasible or which of two approaches works better. This Skill runs small throwaway prototypes that answer those questions with evidence instead of guesswork. ## Core Features & Use Cases - Structured Spike Loop: Follows a five-step loop of question, research, build, stress test, and verdict to keep prototypes focused. - A/B and Multi-Spike Support: Splits large ideas into 2-5 independent questions, runs the riskiest first, and keeps inputs equal for fair comparisons. - Standardized Verdict Output: Produces a VALIDATED, PARTIAL, or INVALIDATED verdict with evidence, surprises, and a concrete next-step recommendation. - Use Case: When asked "is it possible to stream responses from this API", the Skill builds a minimal runnable script in a scratch workspace, tests one edge case, and reports a verdict with the exact command output. ## Quick Start Ask the assistant to spike whether a specific idea or approach is feasible before building it for real.

Frequently Asked Questions about spike

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

FAQPage Schema
How do I test if an idea is feasible before building it?

Run a spike: state the concrete feasibility question, research enough to pick a credible approach, build the smallest runnable artifact, stress one edge case, then record a verdict. This produces evidence without committing to production code.

How to compare two technical approaches with a prototype?

Split the comparison into an A/B spike where both variants receive equal inputs and are measured on the same dimensions. Build the riskiest question first, and ask before building all variants if the work exceeds a small prototype.

When should I not use a spike prototype?

Skip a spike when reading documentation or existing code can answer the question directly, or when the user clearly asked for a production implementation. Spikes are for uncertainty that only running code can resolve.

Where should spike prototype code be stored?

By default, spike artifacts go in a throwaway workspace like .tmp/openclaw-spikes/<slug>. If a tracked location is needed, use a repo-local spikes/<NNN-slug>/ directory with a README and minimal code.

Can spike code be merged into production?

No, spike code should not be merged into production directly. It is written for speed of learning, not quality, so the validated approach should be rewritten normally with proper engineering standards.