# Claude-Setup---Wyss-Members

**Repository Path**: econometric/Claude-Setup---Wyss-Members

## Basic Information

- **Project Name**: Claude-Setup---Wyss-Members
- **Description**: https://github.com/AndreaMentasti/Claude-Setup---Wyss-Members
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-04-15
- **Last Updated**: 2026-04-15

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Claude Code Setup — Wyss Academy for Nature

A ready-to-use Claude Code setup template for empirical research projects. Clone this, fill in the placeholders, and your team has a fully configured AI-assisted research workflow in minutes.

---

## Two Setup Paths

There are exactly two roles. Pick the one that matches you:

| You are... | Read this | Time |
|---|---|---|
| 🏗️ The **owner** creating a new project from this template (PI / lead) | **[OWNER_SETUP.md](OWNER_SETUP.md)** | ~30 min, **once** |
| 💻 A **collaborator** (or owner setting up your laptop for daily work) | **[ONBOARDING.md](ONBOARDING.md)** | ~20 min per machine |

The owner runs `OWNER_SETUP.md` once to customize the template, push to GitHub, and invite the team. Then everyone — owner included — runs `ONBOARDING.md` on their own machine to set up daily work.

---

## What's Included

| Component | Description |
|-----------|-------------|
| `.claude/rules/` | Non-negotiable project rules Claude follows automatically |
| `.claude/agents/` | Specialist review agents (Python, Stata, domain, proofreader, etc.) |
| `.claude/skills/` | Slash commands (`/commit`, `/data-analysis`, `/review-stata`, `/lit-review`, etc.) |
| `.claude/hooks/` | Automation hooks (verify after edits, protect raw data, session logging) |
| `CLAUDE.md` | Claude's operating manual for the project (loaded every session) |
| `OWNER_SETUP.md` | One-time guide for the project owner |
| `ONBOARDING.md` | Per-machine setup with one-shot prompt for collaborators |
| `COLLABORATION.md` | Git workflow guide for teams |
| `MEMORY.md` | Shared project memory — persists learnings across sessions |
| `code/py/config.py` | Two-folder path configuration (repo + shared data) |
| `code/stata/config_local.do.template` | Stata path configuration template |
| `templates/` | Templates for session logs, plans, quality reports |
| `quality_reports/` | Plans, session logs, merge reports |

---

## Design Principles

**Plan first.** Claude enters plan mode before non-trivial tasks and saves plans to `quality_reports/plans/`.

**Quality gates.** Nothing is committed below a score of 80/100. `/review-python` and `/review-stata` run automated quality checks.

**Two-tier memory.** `MEMORY.md` (committed) holds shared project knowledge. `.claude/state/personal-memory.md` (gitignored) holds machine-specific notes.

**Correlational framing.** The domain-reviewer agent flags causal language automatically. All observational findings use association language unless a valid identification strategy is in place.

**Raw data is sacred.** The `protect-files.sh` hook blocks writes to `data/rawdata/` at the tool level.

---

## Stack

Designed for research projects using **Python + Stata**, but adaptable to any combination of Python, R, and Stata. The `.claude/skills/` folder includes translators between all three.

---

## Questions?

Open an issue or contact the Wyss Academy for Nature research team.