# Research Idea Stress Test

> Stress-test a research idea before major investment. Use to expose hidden assumptions, weak novelty, confounders, and the cheapest decisive experiment. Use before starting a project, writing a proposal, scaling experiments, or committing major compute/resources.

- Skill: `jurgendn/research-idea-stress-test` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add jurgendn/research-idea-stress-test`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jurgendn/research-idea-stress-test/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Product & Planning
- Author: jurgendn (https://skillmd.com/u/jurgendn)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/jurgendn/research-idea-stress-test

---


# Research Idea Stress Test

Assume the idea is false until it survives cheap attacks. The goal is not to
prove the idea correct; it is to decide whether it is meaningful,
distinguishable, and worth more time.

For a detailed stress-test question bank, read `references/stress-test-checklist.md`.

## Workflow

1. State the core idea in one sentence.
2. Separate method, hypothesis, and contribution.
3. List what must be true for the idea to matter.
4. Identify nearest existing work and baselines.
5. Detect novelty illusions.
6. Identify hidden advantages and confounders.
7. Predict failure modes.
8. Define the cheapest falsifying test.
9. Define the strongest fair baseline.
10. Define minimal success criteria.
11. State what evidence would change your mind.
12. Recommend pursue, narrow, reframe, or drop.

## Rules

- Do not protect the idea from the strongest objection.
- Do not confuse novelty of phrasing with novelty of contribution.
- Do not scale experiments before a cheap falsifier has run.
- If the decisive test is unavailable, say what proxy would reduce uncertainty.
- If nearest prior work is unknown, route to `literature-triangulation`.

## Output

Return:
- core idea;
- claimed contribution;
- why it might work;
- why it might fail;
- hidden assumptions;
- nearest baselines / related work;
- cheapest decisive experiment;
- evidence that would change your mind;
- recommendation: pursue / narrow / reframe / drop.