Physics Intern Claude Plugin — Development skill for Claude Code

PhysicsIntern — Claude Code plugin

Bootstrap a **PhysicsIntern** research workspace in one command, from inside Claude Code.

PhysicsIntern is a methodology for using an AI coding agent to conduct theoretical physics and mathematics research. You write a question into `problem.md`; the agent then surveys the literature, drafts a plan, dispatches analytical derivations and numerical computations to fresh-context sub-agents, reviews each result, and synthesises an answer — committing after every step.

Install

/plugin marketplace add huggingface/physics-intern-claude-plugin
/plugin install physics-intern@physics-intern-claude

Restart Claude Code afterwards so the command registers.

Use

In a fresh, empty folder for your research problem:

claude
> /init-physics-intern        # scaffolds the workspace (asks for one permission approval)

Then:

1. Edit problem.md — fill in the problem setup and the main question.
2. Exit and restart Claude Code in the folder — so the workspace's CLAUDE.md, skills, and sub-agents load.
3. Run /survey to begin.

Files are durable state and the session is ephemeral: you can clear the context at any time and the agent resumes from `research_log.md` and `plan.md`.

What `/init-physics-intern` scaffolds

A self-contained workspace in the current folder:

• CLAUDE.md — the main-agent methodology (auto-loaded each session).
• .claude/skills/ — /survey, /research-plan, /derive, /compute, /review, /critique, /finalize, /autoresearch.
• .claude/agents/ — the sub-agent role definitions.
• problem.md, research_log.md, and artefact directories (derivations/, computations/, critiques/, notes/, references/, data/).
• An initial git commit.

The one permission approval is expected: the script writes `.claude/` and runs `git init`, which Claude Code's sandbox blocks by default.

Notes

• The plugin only adds the global /init-physics-intern command. The methodology