samuhssamuhs/mad-science/llms.txt

mad-science

llms.txt1 starsChanged 13 days agoMITView on GitHub

Hypothesis-driven data science and machine learning experiment skills (Agent Skills, SKILL.md format) for coding agents: Claude Code, Codex, GitHub Copilot, Cursor, Gemini CLI and any agent supported by skills.sh. The flow takes a raw idea to a falsifiable hypothesis, an experiment contract, tracked slices, expectation-first Jupyter notebooks, a review gate and, optionally, a scientific article. Install with npx skills add samuhs/mad-science, or as a plugin with claude plugin marketplace add samuhs/mad-science then claude plugin install mad-science@samuhs. Run /mad-science-setup once per…

In this repository15 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- critic-reviewSkill
- developing-data-science-notebooksSkill
- expectation-driven-scienceSkill
- experiment-graphSkill
- experiment-pipelineSkill
- exp-reviewSkill
- hypothesysSkill
- let-it-ripSkill
- literature-researchSkill
- mad-science-setupSkill
- notify-gchat-human-waitSkill
- to-articleSkill
- to-docSkill
- what-i-doSkill

The file

Create a free account to mount

Files

llms.txt

# mad-science

> Hypothesis-driven data science and machine learning experiment skills (Agent Skills, SKILL.md format) for coding agents: Claude Code, Codex, GitHub Copilot, Cursor, Gemini CLI and any agent supported by skills.sh. The flow takes a raw idea to a falsifiable hypothesis, an experiment contract, tracked slices, expectation-first Jupyter notebooks, a review gate and, optionally, a scientific article.

Install with `npx skills add samuhs/mad-science`, or as a plugin with `claude plugin marketplace add samuhs/mad-science` then `claude plugin install mad-science@samuhs`. Run `/mad-science-setup` once per repo, then start with `/hypothesys`. MIT licensed.

## Docs

- [README](https://github.com/samuhs/mad-science/blob/main/README.md): installation for every agent, the flow, the catalogue and the FAQ
- [what-i-do](https://github.com/samuhs/mad-science/blob/main/skills/what-i-do/SKILL.md): canonical skill catalogue and flow checks
- [skills.sh listing](https://skills.sh/samuhs/mad-science): install counts and per-skill pages

## Main flow

- [hypothesys](https://github.com/samuhs/mad-science/blob/main/skills/hypothesys/SKILL.md): interview that makes a data science hypothesis falsifiable, with premises, baseline and success metric
- [to-doc](https://github.com/samuhs/mad-science/blob/main/skills/to-doc/SKILL.md): turns the discussion into an experiment doc (the experiment contract)
- [experiment-pipeline](https://github.com/samuhs/mad-science/blob/main/skills/experiment-pipeline/SKILL.md): slices the doc into tracked experiments with status and dependencies
- [let-it-rip](https://github.com/samuhs/mad-science/blob/main/skills/let-it-rip/SKILL.md): implements each experiment in a Jupyter notebook, expectation-first
- [exp-review](https://github.com/samuhs/mad-science/blob/main/skills/exp-review/SKILL.md): reviews notebooks for standards, reproducibility, data leakage and security

## Reference skills

- [expectation-driven-science](https://github.com/samuhs/mad-science/blob/main/skills/expectation-driven-science/SKILL.md): hypothesis, prediction, verification; TDD for science
- [developing-data-science-notebooks](https://github.com/samuhs/mad-science/blob/main/skills/developing-data-science-notebooks/SKILL.md): production-grade notebook practices
- [literature-research](https://github.com/samuhs/mad-science/blob/main/skills/literature-research/SKILL.md): state-of-the-art research saved as Markdown
- [to-article](https://github.com/samuhs/mad-science/blob/main/skills/to-article/SKILL.md): writes or reviews a computer science article from the results
- [critic-review](https://github.com/samuhs/mad-science/blob/main/skills/critic-review/SKILL.md): critical review of the whole project against the state of the art

## Optional

- [mad-science-setup](https://github.com/samuhs/mad-science/blob/main/skills/mad-science-setup/SKILL.md): one-time repo setup
- [experiment-graph](https://github.com/samuhs/mad-science/blob/main/skills/experiment-graph/SKILL.md): interactive graph of the experiment pipeline
- [notify-gchat-human-wait](https://github.com/samuhs/mad-science/blob/main/skills/notify-gchat-human-wait/SKILL.md): Google Chat ping when the agent waits for a human

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

It worked
Partly
It didn't

What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free