# CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

## What this directory is

This is **not** a code repository. It is a personal CV archive for Giorgio Gilestro, containing:

- `gilestro_CV_full.odt` — the single master/editable source of the CV (OpenDocument Text, edited in LibreOffice).
- `PDFs/` — dated, immutable PDF exports of the CV over time.

There is no build system, no tests, and no dependencies here. The global `~/.claude/CLAUDE.md` rules about Python/venv/pytest/PLANNING.md/tasks/ do not apply in this directory.

This directory syncs with the git remote **https://git.lab.gilest.ro/giorgio/myCV.git** (Giorgio's self-hosted Gitea). Per Giorgio's preference, the primary branch is **`master`**, not `main`. Never push without an explicit instruction.

## PDF naming convention

Exported PDFs follow `gilestro_CV_YYYYMM.pdf` (e.g. `gilestro_CV_202510.pdf` for the October 2025 export). When producing a new export, use the current year+month — do not overwrite a prior month's file.

## Common operations

- **Export ODT → PDF** (headless, writes alongside the source):
  ```
  libreoffice --headless --convert-to pdf gilestro_CV_full.odt --outdir PDFs/
  ```
  Then rename the output to `gilestro_CV_YYYYMM.pdf`. The headless converter always writes `gilestro_CV_full.pdf`, so a rename step is required.

- **Read the current CV contents** without opening LibreOffice:
  ```
  unzip -p gilestro_CV_full.odt content.xml | sed 's/<[^>]*>/ /g' | tr -s ' '
  ```
  Useful when the user asks questions about CV content (publications, positions, dates) — the ODT is a zip; `content.xml` holds the text.

- **Open for editing**: `libreoffice --writer gilestro_CV_full.odt`

## Working rules specific to this directory

- **Never edit the `.odt` directly via binary/text tools.** It is a zipped XML bundle; hand-edits will corrupt it. Defer ODT edits to the user (in LibreOffice) unless explicitly asked to script a change via `python-docx`-style tooling against `content.xml`.
- **Never delete or overwrite existing dated PDFs in `PDFs/`** — they are the historical record.
- Treat the most recent PDF in `PDFs/` (by filename date) as the last "published" snapshot when comparing against the current ODT.