# resume-builder

Adopt this skill: https://proskills.md/skills/productivity/heisenberguwu-resume-builder-skill

Page: https://proskills.md/skills/productivity/heisenberguwu-resume-builder-skill
This document is the public skill bundle for that page. It contains SKILL.md and every helper file the catalog carries. No login is required.

## File: SKILL.md
size: 8200
upstream: https://github.com/heisenberguwu/resume-builder-skill/blob/HEAD/SKILL.md

````text
---
name: resume-builder
description: Build a polished single-file HTML resume from scattered historical data. Supports packing up materials from .eml emails, .msg Outlook files, .txt/.html reports, old resumes, class notes, lab reports, project docs, and photos into a single resume. For students, course notes and academic materials are auto-extracted into skills and project experience. Includes resume writing guides for multiple career paths (computer science, human resources, marketing) and swappable CSS style themes (apple, minimal, corporate). Avatar is auto-cropped and Base64-embedded for offline use.
version: 1.0.0
author: community
license: MIT
prerequisites:
  commands: [python3]
  pip_deps: [Pillow]
metadata:
  hermes:
    tags: [resume, HTML, multi-source, data-extraction, writing-guide, swappable-styles, productivity]
    category: productivity
---

# Resume HTML Builder — Multi-source Resume Generator

Build a modern, single-file HTML resume from scattered historical data (.eml reports, old resumes, photos, publications).

## When to Use

- User wants to rebuild a resume from scattered materials (weekly reports, old resumes, etc.)
- User asks to pack up their materials and generate a resume (zip or directory)
- Converting Markdown/PDF resume to a single-file HTML
- Need avatar cropping + Base64 embedding for offline viewing
- Multi-career resume writing guidance (tech, HR, etc.)
- Multi-style design choices (apple, minimal, corporate)
- Multi-source data extraction and merging
- Student course notes, lab reports, and project docs for skill extraction

## Typical User Prompts

These are the most common ways users trigger this skill:

| Prompt | What happens |
|--------|-------------|
| "把周报打包成一份简历" | Extract .eml/.txt, merge, render HTML |
| "把我的资料整理成简历" | Categorize files, extract text, draft resume |
| "用这些材料帮我生成一份简历" | Full pipeline: extract → draft → style → deliver |
| "用我的课程笔记和实验报告提取一份在校生简历" | Extract skills and projects from academic materials, render HTML |
| "把旧简历转成 HTML 格式" | Parse existing resume, apply chosen theme |

## Language Rule (IMPORTANT)

Match the user's input language throughout the entire workflow:

| User speaks | Guide | HTML template | Resume content |
|-------------|-------|---------------|----------------|
| English | `*-guide.md` | `resume-html.html` | English |
| 中文 | `*-guide-cn.md` | `resume-html-cn.html` | 中文 |

If the user doesn't specify, default to English. This applies to every step — guide selection, template selection, and all generated resume content.

## Workflow

```
Collect sources → Extract data → Deduplicate & merge → Draft → Render HTML → Embed avatar → Deliver
```

### Two Execution Modes

#### Mode A: Full Pipeline (Recommended)

Run `pipeline.py` to automate all script steps. Manual steps (drafting, Markdown) will print TODO prompts:

```bash
# From zip
python scripts/pipeline.py --zip "resume-sources.zip" --output-dir ./output

# From directory
python scripts/pipeline.py --dir /path/to/sources --output-dir ./output

# Provide pre-built HTML to skip draft step
python scripts/pipeline.py --dir /path/to/sources --output-dir ./output --html ./my-resume.html

# Skip specific steps
python scripts/pipeline.py --dir /path/to/sources --output-dir ./output --skip embed-avatar,html-to-pdf
```

Pipeline output flow:
```
[1/6] Extract & categorize sources    (AUTO)   → extracted_manifest.json
[2/6] Extract text from files          (AUTO)   → extracted_texts.json
[3/6] Draft resume + render HTML       (MANUAL) → read extracted_texts.json, draft, save resume-<name>-final.html
[4/6] Crop & embed avatar              (AUTO)   → overwrites HTML with embedded avatar
[5/6] Generate Markdown                (MANUAL) → convert HTML to resume-<name>-final.md
[6/6] Convert HTML to PDF              (AUTO)   → resume-<name>-final.pdf
```

**Agent workflow with pipeline:**
1. Run `pipeline.py` — it completes steps 1-2 then prints TODO for step 3
2. Read `output_dir/extracted_texts.json` to understand source material
3. Follow steps 3-4.6 below to draft resume content
4. Save HTML to `output_dir/resume-<name>-final.html`
5. Re-run pipeline with `--skip extract-sources,extract-text,draft-resume` to continue from step 4
6. Generate Markdown (step 5)
7. Re-run pipeline with `--skip extract-sources,extract-text,draft-resume,embed-avatar,generate-markdown` for final PDF

#### Mode B: Individual Scripts

Run each script manually for full control. See individual step sections below.

## Prerequisites

```bash
pip install Pillow          # avatar cropping (recommended)
pip install weasyprint      # PDF generation (recommended)
pip install msgreader       # .msg support (optional)
```

## Step 1: Extract and Categorize Source Data

User typically provides a zip archive. Supported file types:

| Extension | Content | How it's read |
|-----------|---------|---------------|
| `.eml` | Standard email (Thunderbird/Mail.app export) | MIME parser |
| `.msg` | Outlook email | `extract-msg` (pip install) |
| `.txt`, `.text` | Plain text reports | Direct read |
| `.html`, `.htm` | HTML emails/reports | Tag stripped |
| `.md`, `.sg`, `.pdf` | Old resume drafts, class notes, lab reports, project docs | Text/pdf parser |
| `.jpg`, `.png` | Avatar photos | Image picker |

```bash
# From zip
python scripts/extract_sources.py --zip "resume-sources.zip" --output extracted

# From directory
python scripts/extract_sources.py --dir /path/to/sources
```

Outputs categorized file lists + JSON. Auto-selects best avatar (prefers `avatar`/`photo`/`head` in filename, else largest file).

**Supported email formats:**
- `.eml` — standard MIME (Thunderbird, Mail.app, generic export) — no extra deps
- `.msg` — Outlook format — needs `pip install msgreader`
- `.txt` / `.html` — plain text or HTML saved from email clients — direct read

## Step 2: Extract Text from Reports

The script converts any file to readable text — it does **not** structure content.

```bash
# Single file
python scripts/extract_text.py report.eml

# Directory of reports
python scripts/extract_text.py reports/ --json

# Recursive
python scripts/extract_text.py reports/ -r --json
```

**Output format (JSON):**
```json
[{"filename": "report.eml", "content": "From: ...\nSubject: ...\n\nBody content..."}]
```

**Agent interprets the text:** Feed the `content` to yourself and use natural language understanding to extract dates, subjects, key achievements, and metrics. The script handles MIME decode, HTML tag stripping, and encoding — you handle meaning.

**Supported formats:**
- `.eml` — standard MIME email — no extra deps
- `.msg` — Outlook — needs `pip install msgreader`
- `.html` / `.htm` — auto-stripped to text
- `.txt` / `.md` / `.sg` / anything else — direct read

## Step 3: Extract from Old Resumes

Markdown: parse `#` section headers and group content under each.

PDF: use the `ocr-and-documents` skill (pymupdf or marker-pdf).

## Step 4: Select Career Path and Draft Resume Content

Pick a career path from `references/<career>/resume-guide.md` (EN) or `resume-guide-cn.md` (CN).

| Career Path | Focus | Guide path |
|-------------|-------|-----------|
| **Computer Science** (default) | FAB pattern, per-stack focus (backend/frontend/mobile/C++/Go/architecture), ATS keywords | `references/computer-science/` |
| **Human Resources** | Recruitment metrics, employee relations, HRBP alignment, labor compliance | `references/human-resources/` |
| **Marketing** | Growth metrics (CAC/ROI/LTV), brand campaigns, user ops, content marketing, GTM | `references/marketing/` |

**Language:** Use English guide if user speaks English, Chinese guide (`-cn.md`) if user speaks 中文. Default to `computer-science` if role is unclear.

**Adding a career path:** Create `references/<career>/resume-guide.md` + `resume-guide-cn.md`. Cover: key metrics to emphasize, per-role focus areas, ATS keywords, and concrete before/after writing examples.

**Key merge principles:**
- **Timeline alignment:** Match report dates
````

## File: README.md
upstream: https://github.com/heisenberguwu/resume-builder-skill/blob/HEAD/README.md

````text
# Resume Builder

[English](README-en.md)

把周报、邮件、课程笔记、旧简历和照片交给 AI，几分钟生成一份精美简历。

支持一键提取、自动排版，换风格、翻译、导出 PDF 都能一句话完成。

---

### 效果预览

<table>
<tr>
<td align="center">
  <img src="previews/apple.png" width="280"><br><br><b>Apple</b> — 暗色头部，大字体，卡片网格
</td>
<td align="center">
  <img src="previews/minimal.png" width="280"><br><br><b>Minimal</b> — 居中排版，左侧边框强调
</td>
</tr>
<tr>
<td align="center">
  <img src="previews/corporate.png" width="280"><br><br><b>Corporate</b> — 橙色强调，箭头项目符号
</td>
<td align="center">
  <img src="previews/heisenberg.png" width="280"><br><br><b>Heisenberg</b> — 头像并排，柔和蓝色，圆润卡片
</td>
</tr>
</table>

---

### 安装

选一个你用的工具：

**🚀 Hermes:**
```bash
git clone https://github.com/HeisenbergUwU/resume-builder-skill.git ~/.hermes/skills/resume-builder
```

**🦞 OpenClaw:**
```bash
git clone https://github.com/HeisenbergUwU/resume-builder-skill.git ~/.openclaw/skills/resume-builder
```

**💻 Cursor / Claude Code / 通用:**
```bash
git clone https://github.com/HeisenbergUwU/resume-builder-skill.git skills/resume-builder
```

---

### 怎么用

装好之后，把资料发给 Agent 就行了：

> 帮我把这些周报和邮件整理成一份简历
>
> 用我的课程笔记和实验报告提取一份在校生简历

生成后一句话就能调整：

> 换成 corporate 风格，看着正式一点
>
> 翻译成英文版
>
> 导出成 PDF

---

### 能力一览

| 能力 | 说明 |
|------|------|
| 多源提取 | .eml / .msg / .html / .txt / .md / .jpg / .png 全部支持 |
| 自动渲染 | 旧简历 Markdown 纯脚本拼接 HTML，无需大模型 |
| 在校生友好 | 课程笔记、实验报告自动提取技能和项目经历 |
| 职业指南 | 计算机科学 · 人力资源 · 市场运营 |
| 样式主题 | apple · minimal · corporate · heisenberg · pulse |
| 头像处理 | 自动发现、居中裁剪、Base64 内嵌 |
| 多格式输出 | HTML + Markdown + PDF |

---

### 自定义样式

内置主题不够？把 CSS + 模板放到 `assets/styles/<name>/` 下即可生效，欢迎 PR。
````
