experiment-structure
Standard experiment folder structure and templates. Reference for creating or validating experiment folders.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Standard experiment folder structure and templates. Reference for creating or validating experiment folders.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
Write up research experiments and findings as an interactive, self-contained HTML report published as a claude.ai Artifact, with figures and data exploration. Use when creating a research report from experiment results.
How to write a comprehensive research report for your supervisor summarizing experiment findings, hypothesis tests, and data exploration. Use when you need to communicate results from multiple experiments with proper analysis, visualizations, and interactive data exploration.
How to send notifications to the human supervisor via ntfy.sh. Use when you need input, hit a blocker, or update them on your progress.
Cost and latency optimization for Anthropic API usage. Covers prompt caching, batch API, and when to combine them.
Read model outputs, samples, or log entries by eye and report specific observations about what's in them. Use when asked for a "qualitative" read/analysis, or when the goal is to understand WHAT is in the data rather than count patterns. Not for numerical summaries, metric computation, or classifier-driven sweeps.
How to evaluate research samples using structured JSON output from claude -p. Covers criteria writing, the core judging pattern, and practical examples.
| name | experiment-structure |
| description | Standard experiment folder structure and templates. Reference for creating or validating experiment folders. |
| user-invocable | false |
Standard structure for experiments in this research project.
mkdir -p experiments/exp_NNN_name/{outputs,judgments,scratch,suggested_utils}
name: exp_NNN_name
description: |
[What this experiment tests]
# Parameters
# [Add experiment-specific parameters]
# Expected outputs
# [What files/data this should produce]
# Experiment Report: exp_NNN_name
## Experiment
[What you tested]
## Method
[How you ran it - commands, configs, tools used]
## Observations
[Raw results, verbatim outputs]
## Judgments
[Aggregated scores, patterns - if applicable]
## Anomalies
[Anything unexpected]
## Data
- **Outputs**: `outputs/`
- **Judgments**: `judgments/`
- **Reproduction**: `reproduce.py`
#!/usr/bin/env python3
"""Reproduce key results from this experiment."""
def main():
# TODO: Add reproduction steps
pass
if __name__ == "__main__":
main()
Before running an experiment, verify the folder has:
config.yaml with experiment parametersreport.md (can be empty template)outputs/ directoryjudgments/ directory (if using judges)scratch/ directory (for throwaway code)suggested_utils/ directory (for proposing reusable code)If anything is missing, create it using the templates above.