skills.rest
ruban-s
ruban-sCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

Science

Applies the scientific method as a structured problem-solving loop with falsifiable hypotheses and measured experiments.

Updated Sep 3, 2026
Search Tags:#debugging#problem solving#experiment design#iteration#hypothesis testing#a/b testing#scientific method
One-click install
npx skills add https://github.com/ruban-s/DevOS --skill science-ruban-s
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Science
Source: https://github.com/ruban-s/DevOS/tree/main/skills/Science
Command: npx skills add https://github.com/ruban-s/DevOS --skill science-ruban-s

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Most problem-solving is guessing: the first plausible idea gets built, success is never defined, and results are judged by feel rather than evidence. This Skill replaces ad-hoc trial-and-error with a disciplined cycle — define the goal, generate multiple falsifiable hypotheses, design experiments that can fail, measure honestly, and iterate. ## Core Features & Use Cases - Structured Workflows: Nine workflows covering the full cycle — DefineGoal, GenerateHypotheses, DesignExperiment, MeasureResults, AnalyzeResults, Iterate, FullCycle, plus QuickDiagnosis and StructuredInvestigation shortcuts. - Scale-Adaptive Protocol: The same loop runs at micro-scale (TDD cycles in minutes), meso-scale (feature validation in days), and macro-scale (MVP launches over weeks). - Use Case: When debugging a stale-data bug, run QuickDiagnosis to generate ranked hypotheses (timestamp format mismatch, cache issue, query filter), test the cheapest one first, and confirm the root cause with a direct query instead of flailing for 30 minutes. ## Quick Start Ask the agent to run a structured investigation on why your checkout abandonment rate is 40%, generating at least three testable hypotheses before any fix is attempted.

Frequently Asked Questions about Science

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

FAQPage Schema
How do I structure a debugging session using the scientific method?▼

Use the QuickDiagnosis workflow: state the goal, list at least three ranked hypotheses about the root cause, then test the cheapest one first with a direct check. Confirm the fix by verifying the original symptom is gone, typically within 15 minutes.

How to run controlled experiments on prompt variations?▼

Define eval criteria and a baseline score first, then generate multiple prompt variants as falsifiable hypotheses and compare them with an eval suite. The skill delegates prompt experiments to the Evals skill for position swapping, multi-judge panels, and statistical significance testing.

When should I use FullCycle instead of QuickDiagnosis?▼

Use FullCycle for systematic investigations spanning days or weeks, such as feature validation or MVP launches, where pre-registered criteria and documented methodology matter. QuickDiagnosis fits fast bug hunts where the full ceremony would slow you down.

Why does the skill require at least three hypotheses before testing?▼

Single-hypothesis testing invites confirmation bias, missed alternatives, and sunk-cost attachment to the first idea. Generating three or more distinct, falsifiable candidates forces you to consider orthogonal explanations and test the cheapest high-value option first.

What are the limitations of this problem-solving approach?▼

The structured cycle adds overhead that is wasteful for trivial tasks, and it is not designed for multi-angle lens passes, which belong to the IterativeDepth skill. It also requires discipline to avoid infinite iteration without shipping.

Related Skills

Explore More Software Engineering Skills→
open-platform-modelopen-platform-model

openspec-explore

Guide problem exploration and requirement clarification with clarifying questions and ASCII diagrams.

Official
Intermediate
Handy369Handy369

shipping-and-launch

Define release readiness, staged rollout, monitoring, and rollback triggers before deployment.

Community
Intermediate
flonatflonat
127

devils-advocate

Challenge research assumptions and identify argument weaknesses through adversarial critique.

Community
Advanced
larsboeslarsboes

Science

Structure problem solving through hypothesis-driven experimentation with pre-registered success criteria.

Community
Advanced