cv-creator

これはレビュー時に保存された抜粋です。完全で最新の内容は外部ソースを確認してください。

---
name: cv-creator
description: Generate, update, tailor, compile, and visually review a repo-local LaTeX CV/resume and cover letter. Use when the user wants to create or improve a CV, tailor a resume for a role or company, compile LaTeX CV files to PDF, review formatting, or set up a clean repo-local career CV structure under `memory/areas/career/` or `memory/areas/builder/career/`.
---

# CV Creator

Use this skill for repo-local career materials. Keep the workflow shared, but keep the person's actual career facts inside the current repo.

## Repo contract

Durable career memory and disposable tailored packets are kept in two different places on purpose:

- **Durable (tracked, committed):** the career root holds the source of truth — base templates, profile facts, and tailoring rules. The skill auto-detects the career root in this order based on existing files: `memory/areas/career/`, then `memory/areas/builder/career/`. Use `--career-root <path>` when a repo intentionally has multiple career areas.
- **Disposable (gitignored, throwaway):** per-company tailored packets are renderings of that canon for one application, not memory. The CLI writes them under the repo's `tmp/` so they are never committed and clear out when the application is sent. Do not move tailored packets back into the tracked career root.

Expect this layout unless the repo clearly documents a different one:

- `<career-root>/overview.md`
- `<career-root>/profile.md`
- `<career-root>/tailoring-guide.md`
- `<career-root>/job-tracker/`
- `<career-root>/cv/latex/.gitignore`
- `<career-root>/cv/latex/base/resume.tex`
- `<career-root>/cv/latex/base/cover-letter.tex`
- `<repo-root>/tmp/cv/tailored/<company>/job-description.md` *(disposable)*
- `<repo-root>/tmp/cv/tailored/<company>/resume.tex` *(disposable)*
- `<repo-root>/tmp/cv/tailored/<company>/cover-letter.tex` *(disposable)*

The CLI resolves these paths for you — call `cv.py init/build/review --company <slug>` and let it place files. `cv.py clean` removes the disposable `tmp/cv/` tree (and the render temp) when the work is done. If the durable contract (base templates, profile, tailoring guide) is missing, create that minimal structure first instead of improvising files in random places.

Read `references/structure-contract.md` when you need the exact folder expectations or migration pattern.

Read `references/template-patterns.md` when you need to understand or modify the LaTeX template's structure, preamble, custom commands (`\role`, `\labelrow`, `\sep`), spacing system, or section conventions. This is the canonical reference for "how does a base resume look in this skill" and "how do I add a new section without breaking the rhythm."

## Core workflow

1. Read `<career-root>/profile.md` first for source-of-truth career facts.
2. Read `<career-root>/tailoring-guide.md` if tailoring or summary changes are needed. Pay particular attention to the "Competitor awareness" section, which governs how to handle product-level references to the target company's rivals (e.g. do not send an Anthropic application full of Codex name-drops).
3. If a tailored packet is already in progress, read its `tmp/cv/tailored/<company>/job-description.md` first when reviewing or continuing it.
4. If the tailored JD snapshot does not exist yet, read the relevant JD from `<career-root>/job-tracker/` or the user-provided source, then store a frozen copy beside the tailored files as `tmp/cv/tailored/<company>/job-description.md` so later review is self-contained.
5. When starting a new tailored packet, initialize it with `cv.py init --company <slug>` instead of manually copying files. The CLI writes it into the disposable `tmp/cv/tailored/<company>/`.
6. Keep the base LaTeX files in `cv/latex/base/` clean and generic.
7. Tailored, company-specific versions live in the disposable `tmp/cv/tailored/<company>/`; never commit them into the tracked career root. Run `cv.py clean` when the application is sent.
8. Never invent experience, skil