agentsclimarketplace
Sign in

Cv clone

Skill xiehuacheng/skills/skills/career/cv-clone

Use when the user has a target resume or CV sample (PDF, screenshot, or visual reference) and wants to clone its visual layout and style into an editable LaTeX template. Triggers on phrases like "clone this resume layout", "match this CV design", "rebuild in this style", "reproduce this template", "make me a resume like this one".From its SKILL.md

Install
npx -y skills add xiehuacheng/skills --skill cv-clone

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

5.7 KB, ~1.3k tokens by cl100k_base, as published. Nobody here has run it

cv-clone — Resume Visual Cloning

Overview

Given a target resume sample (image/PDF) plus fresh content, produce a compilable LaTeX template matching the sample's visual style plus a rendered PDF preview. Ships template cloning only — content filling is gated behind user confirmation.

Style is cloned, not content. Never copy names, phones, or company claims from the source sample.

When to Use

Use when:

  • User has a target resume sample (image/PDF) and wants its visual style cloned for their own CV
  • User wants a compilable, editable LaTeX template (not just a static PDF)
  • User can install tectonic

Do NOT use when:

  • User has no sample — no LaTeX template can be cloned without one
  • User only wants content, no specific external visual style
  • Target is Word / PowerPoint / Canva — not supported

Defaults

  • Read-only mode. First deliverable is template + dry-render preview; no user data touched.
  • Photo mirrors the sample. Has a slot → template defines \cvphoto{path}. None → no macro. No silent injection.
  • Step 5 gate. After preview, ask: "Shall I fill this template with your content?" Affirmative unlocks Step 6. Silence = dry template.
  • Sample by path. Sources referenced by path, never copied.

Can do: extract a target resume's layout, typography, spacing, and color palette from a PNG/PDF; generate a compilable LaTeX template with \newcommand placeholders; compile the template with tectonic to produce a PDF preview; accept user content for each placeholder after showing the dry preview.

Cannot do (without explicit user approval): fill user-provided personal data into the template before Step 5 approval; overwrite an existing template.tex without confirmation; copy names, phones, or company claims from the source sample (style only).

Default behavior: all commands are read-only; fidelity bar defaults to "same vibe" (not pixel-perfect); the Step 5 dry-preview gate is mandatory before any content is filled.

Required environment

tectonic (xelatex-based LaTeX compiler) + pdftoppm (PDF→PNG). See references/install.md for cross-platform install and verification. If tectonic is missing, install before Step 4 and confirm with the user.

Workflow

Step 1 — Receive sample + intent

Ask once, in one message:

  1. Path to the target sample (image or PDF)
  2. Fidelity bar: pixel-faithful (slower, 2-3 iteration rounds) or same vibe (one pass, default)?
  3. Whether the photo slot should be enabled (only if the sample has one)

Do not start until all three are answered.

Step 2 — Extract style skeleton

Use the prompt in references/style-extraction-prompt.md with a vision model. Get back a YAML object covering name alignment, section decor, entry layout, colors, fonts, photo position, page size. Write it into the chat for the user to confirm before Step 3.

Step 3 — Emit LaTeX template

Read references/latex-template-skeleton.tex. For each YAML field, override the relevant \newcommand / \setCJKmainfont / \cvsection / \cventry / \definecolor. Save as template.tex in the user's chosen directory.

Constraints:

  • Self-contained: no external .sty beyond what tectonic pulls on demand
  • Every fillable slot exposed via \newcommand with a default
  • Top of file carries a comment block listing every placeholder (see skeleton's header)

Step 4 — Compile dry preview

cd <user-dir> && tectonic template.tex
pdftoppm -r 150 template.pdf template_preview -png

If compile fails, read the error and pick one fix: (a) suggest a font fallback, (b) toggle AutoFakeBold=2 to =1, (c) pin \setCJKmainfont{Noto Serif CJK SC}. Never swallow silently. See references/common-mistakes.md for the full failure-mode catalogue.

Step 5 — Show preview & ask to fill

Open template_preview-1.png for the user. Ask:

Template generated. Preview at <path>. Shall I fill this template with your content?

Branches: No → done. Yes, fill these fields: … → Step 6.

Step 6 — Fill (only after explicit "Yes")

For each \newcommand the user wants filled: confirm the exact value (echo back if parsed from earlier context), add a \renewcommand{\cvX}{...} block near the top, recompile with tectonic template.tex, show the new preview.

Common mistakes

The five that bite most:

  1. Missing AutoFakeBold=2 on macOS — \textbf looks identical to regular weight on PingFang SC. Always include it.
  2. pdflatex instead of xelatex — CJK renders as boxes. Always tectonic.
  3. Skipping the dry preview — style extraction is approximate. Compile before claiming success.
  4. Filling real data before preview approval — violates Step 5 gate. Dry preview first, always.
  5. Forgetting \usepackage{xcolor} before \definecolor — build fails. Diff against the skeleton.

Extended catalogue (#6, #7, #9, #10): references/common-mistakes.md.

References

  • references/latex-template-skeleton.tex — canonical skeleton the agent extends
  • references/install.md — tectonic + pdftoppm install per platform
  • references/style-extraction-prompt.md — vision-model prompt + YAML→LaTeX mapping
  • references/common-mistakes.md — extended failure-mode catalogue
  • references/limits.md — what is approximate vs guaranteed

What ships with it: 5 files

11.8 KB alongside SKILL.md

references/

Gives 0 of the 12 instructions most hr recruiting skills give in ~1.3k tokens

Counted across 356 of the 357 authors here whose files we hold, read 2026-08-07

  • Quantify achievements with specific metricsin 14 of 356, across 6 files
  • Keep the resume under two pagesin 14 of 356, across 6 files
  • Request the full job description if not providedin 12 of 356, across 4 files
  • Extract keywords and prioritize job requirementsin 12 of 356, across 4 files
  • Stop and ask for clarification if required inputs are missingin 12 of 356, across 5 files
  • Map candidate experience to job requirementsin 11 of 356, across 3 files
  • Ask if the user wants adjustmentsin 11 of 356, across 3 files
  • Provide strengths and gap analysis after the resumein 10 of 356, across 2 files
  • Request candidate background details if not providedin 10 of 356, across 2 files
  • Format experience bullets as action verb plus resultin 10 of 356, across 2 files
  • Ask for missing inputs before startingin 10 of 356, across 9 files
  • Use exact job description terminologyin 9 of 356, across 1 file

Said here and by no other author read

  • do not copy content from the source sample
  • ask for the sample path and fidelity bar
  • wait for answers before starting extraction
  • extract layout typography and colors
  • write the style skeleton for user confirmation
  • generate a compilable LaTeX template

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.