05 Experiment

Use when designing, rewriting or auditing experiments, results or benchmark comparisons. Connect each research claim to matched controls, metrics, evidence and limits.

Install

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add ChenXiangYpily1234/paper-storycraft-skills --skill 05-experiment
Manual install
git clone --depth 1 https://github.com/ChenXiangYpily1234/paper-storycraft-skills.git /tmp/paper-storycraft-skills
mkdir -p ~/.claude/skills
cp -r /tmp/paper-storycraft-skills/05-experiment ~/.claude/skills/05-experiment

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# Experiments as an Evidence-Based Argument

## Execute this skill

**Typical request:** “帮我检查实验是否真的支撑优于 baseline 的结论。”

**Procedure**

1. Build claim → comparator → data/split → metric/unit → budget → result → interpretation for each experiment.
2. Verify named models, benchmarks, datasets and citation provenance; distinguish external reported numbers from the author's runs.
3. Order evidence by inferential job, check fair comparisons and alternative explanations, then write bounded findings.

**Deliver:** Return a question-to-evidence matrix, revised experiment narrative (if requested), and P0/P1 validity gaps.

**Evidence boundary:** Never invent seeds, statistical significance, runs or missing baselines; label proposed controls as proposals.

Read [PRINCIPLES.md](../PRINCIPLES.md) first. Use for empirical, numerical and engineering evidence sections in STEM; it is optional for proof-only papers.

## Section-specific editorial judgement

**Editorial choice:** Lead with decisive, fairly controlled findings, not a row-by-row score diary. Reserve main-text space for core conclusions and claim-changing trade-offs; extensive secondary negative or exploratory results can appear in appendices or supplements, with main-text qualification where material. A metric selected post hoc must not be portrayed as a prespecified endpoint. Distinguish directly verified gains from inferred explanations.
Read the full skill on GitHub

Details

Skill name
05-experiment
Path
05-experiment/SKILL.md
License
MIT
Repo updated
20 October 2026

FAQ

What does the 05 Experiment skill do?

Use when designing, rewriting or auditing experiments, results or benchmark comparisons. Connect each research claim to matched controls, metrics, evidence and limits.

How do I install the 05 Experiment skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add ChenXiangYpily1234/paper-storycraft-skills --skill 05-experiment. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is 05 Experiment popular?

It lives in ChenXiangYpily1234/paper-storycraft-skills, which has 44 GitHub stars and 0 forks.

Related skills

Browse the most popular Claude skills, all research & science skills, or write your own with the skills generator.