casely-qa-skill

内容来源：README.md（说明文档） · 原始地址 · 查看安装指南

原始内容

# 🚀 Casely — AI QA Test Case Generator

<div align="center">

<img src="assets/opengraph-image.png" alt="Casely — AI QA Test Case Generator: PDF requirements to TestRail-ready Excel in 8 minutes" width="720">

**Turn messy PDF requirements into TestRail-ready test cases in 8 minutes.**  
*Free, open-source AI skill for Claude Code, Cursor, and any AI IDE.*

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
[![Version](https://img.shields.io/badge/version-1.5.0-blue.svg)](https://github.com/JohnWayneeee/casely-qa-skill/releases)
[![Python](https://img.shields.io/badge/Python-3.10+-blue.svg)](https://www.python.org/)
[![Stars](https://img.shields.io/github/stars/JohnWayneeee/casely-qa-skill?style=flat&logo=github)](https://github.com/JohnWayneeee/casely-qa-skill/stargazers)
[![Issues](https://img.shields.io/github/issues/JohnWayneeee/casely-qa-skill)](https://github.com/JohnWayneeee/casely-qa-skill/issues)
[![Casely Web](https://img.shields.io/badge/Hosted%20Version-casely.digital-ff6b6b?style=flat)](https://casely.digital/)

</div>

---

## The problem every QA engineer knows

You were hired to **find bugs**. Instead, you spend 40% of your week writing test cases.

Requirements scattered across 10 PDF files. Every project has different column names — manual reformatting every time. A single module takes 2 days to document. Then the TestRail import fails because the headers don't match.

> ❌ Requirements buried in PDF/DOCX/XLSX files with no structure  
> ❌ Each project reinvents column names — manual mapping every sprint  
> ❌ 50 test cases = 2–3 business days of repetitive writing  
> ❌ TestRail import breaks due to column mismatches  
> ❌ No coverage plan = missed edge cases and bugs reaching production  

**Every hour writing test cases is an hour not spent testing.**

---

## What Casely does

Casely is your **Virtual QA Lead**. It reads your requirements, learns your team's exact format, and writes the test cases — atomic, structured, and import-ready.

```
Requirements PDF → /parse → /style → /plan → /generate → /export → TestRail ✅
```

| Step | Command | What happens |
|------|---------|--------------|
| **Extract** | `/parse` | Docling OCR pulls tables and text from any PDF/DOCX |
| **Learn** | `/style` | Casely reads your existing Excel and clones your column structure |
| **Plan** | `/plan` | Generates a coverage map: "47 tests across 6 modules" |
| **Write** | `/generate` | Creates atomic `.md` test cases — one file per test |
| **Deliver** | `/export` | Batch-converts everything to TestRail-ready Excel |

Open `exports/functional_TC001_happy_path.xlsx` — it's a 1:1 match to your team's template, ready for immediate import.

---

## Why QA teams switch to Casely

| | Casely | Manual writing | Traditional tools |
|---|:---:|:---:|:---:|
| Parses any format (PDF/DOCX/XLSX) | ✅ | ❌ | ❌ |
| Matches **your** column structure | ✅ | ❌ | ❌ |
| Generates a test plan automatically | ✅ | ❌ | ❌ |
| 1 test case = 1 file (atomic, reviewable) | ✅ | ❌ | ⚠️ bulk only |
| TestRail / Qase ready out of the box | ✅ | ❌ | ⚠️ manual fix |
| Works with English **and** Russian docs | ✅ | ❌ | ❌ |
| Free, runs locally, no cloud lock-in | ✅ | ✅ | ❌ |

---

## ⚡ Quick Start

> **Prerequisites:** Python 3.10+ and [uv](https://github.com/astral-sh/uv)

### Option A — Skills CLI (recommended)

```bash
# with bunx
bunx skills add JohnWayneeee/casely-qa-skill

# or npx
npx skills@latest add JohnWayneeee/casely-qa-skill
```

### Option B — Clone & run

```bash
git clone https://github.com/JohnWayneeee/casely-qa-skill.git
cd casely-qa-skill
uv sync
```

Drop your files into the project folder and run the workflow:

```
/init my-project
/parse
/style
/plan
/generate functional AccountTransfer
/export
```

---

## 8-minute walkthrough

<details>
<summary><strong>Step 1 — Initialize</strong></summary>

```bash
/init my-project
```

Drop two files into `projects/my-project/input/`:
- `requirements.pdf` — your specification document
- `example.xlsx` — an existing test case file your team already uses

Casely learns your column structure from the example. No config needed.

</details>

<details>
<summary><strong>Step 2 — Parse & plan</strong></summary>

```bash
/parse    # extracts text and tables from your PDF via Docling OCR
/style    # reads your example.xlsx and clones the column structure exactly
/plan     # output: "Detected 6 modules. Recommended: 47 test cases."
```

You now have a full coverage strategy before writing a single test case.

</details>

<details>
<summary><strong>Step 3 — Generate</strong></summary>

```bash
/generate functional AccountTransfer
```

Casely writes 10+ atomic `.md` files — one per test case. Review and edit in any text editor or Git UI before exporting.

</details>

<details>
<summary><strong>Step 4 — Export</strong></summary>

```bash
/export
```

Every `.md` file becomes a separate Excel file matching your team's template exactly. Import into TestRail, Qase, or any TMS — no manual reformatting.

</details>

---

## Command reference

| Command | Action | Notes |
|---------|--------|-------|
| `/init [name]` | Scaffold project workspace | Run once per project |
| `/parse` | Extract from PDF/DOCX | High-fidelity OCR via Docling |
| `/style` | Clone your Excel column format | Reads your example — zero config |
| `/plan` | Build ISTQB-aligned coverage map | Shows module breakdown and estimates |
| `/generate [type]` | Write atomic test cases | Produces one `.md` per test |
| `/export` | Convert all `.md` to Excel | Batch output, import-ready |

Supported generation types: `functional`, `negative`, `integration`, `boundary`, `smoke`, `security`

---

## Under the hood

- **Docling Engine** — advanced OCR and table extraction for complex, multi-column PDFs including scanned documents
- **Atomic design** — 1 test case = 1 source file = 1 Excel; no monolithic spreadsheets to untangle
- **Style Guide System** — no hardcoded columns; Casely learns from your existing files and replicates the exact structure
- **Language agnostic** — generates test cases in English or Russian, matching the language of your documents

---

## Project structure

```
casely-qa-skill/
├── skill/casely/
│   ├── SKILL.md              # AI skill definition and full workflow
│   ├── scripts/
│   │   ├── casely_parser.py  # Document → Markdown converter (Docling)
│   │   └── export_to_xlsx.py # Markdown → Excel exporter
│   ├── references/           # Technical reference docs
│   └── evals/                # Evaluation test cases
├── docs/
│   └── hosted-web-version.md # Hosted version details
├── assets/                   # Images and branding
├── pyproject.toml            # Python dependencies
└── marketplace.json          # Skill marketplace metadata
```

---

## Hosted version for teams

This open-source skill runs locally in your AI IDE.

If you want a **browser UI, file uploads, team review flow, and zero local setup**:

**[casely.digital](https://casely.digital/)** — join the early access list

The hosted version is built for QA teams that want to turn requirements into review-ready test cases without writing code or running local scripts.

[Learn more about the hosted version →](https://github.com/johnwayneeee/casely-qa-skill/blob/HEAD/docs/hosted-web-version.md)

---

## FAQ

<details>
<summary>Does it work with scanned PDFs?</summary>

Yes. Casely uses Docling's OCR pipeline, which handles scanned documents, embedded tables, and mixed-format pages.

</details>

<details>
<summary>Can I use my own Excel column structure?</summary>

That's the core feature. Drop your existing template into `input/` and run `/style`. Casely reads your column names and replicates them exactly — no configuration file needed.

</details>

<details>
<summary>Does it support Russian-language requirements?</summary>

Yes. The parser and generator work with English and Russian documents. Column names in your Excel template are preserved as-is.

</details>

<details>
<summary>Which test management systems are supported for import?</summary>

Any TMS that accepts Excel import: TestRail, Qase, Zephyr, Xray, and plain Excel. Because Casely replicates your own column structure, the output matches whatever format your team already uses.

</details>

<details>
<summary>What's the difference between the skill and the hosted version?</summary>

The skill runs locally inside your AI IDE (Claude Code, Cursor, etc.) — full control, no cloud, free forever.  
The hosted version at [casely.digital](https://casely.digital/) adds a browser UI, team review workflows, and cloud storage — no local setup required.

</details>

---

## Contributing

Contributions are welcome. Please read [CONTRIBUTING.md](https://github.com/johnwayneeee/casely-qa-skill/blob/HEAD/CONTRIBUTING.md) before opening a pull request.

- 🐛 **Found a bug?** [Open an issue](https://github.com/JohnWayneeee/casely-qa-skill/issues/new?template=bug_report.md)
- 💡 **Have an idea?** [Request a feature](https://github.com/JohnWayneeee/casely-qa-skill/issues/new?template=feature_request.md)
- ⭐ **Did it help?** Star the repo — it helps other QA engineers find this tool

---

## ⭐ Star History

If Casely saved you a work week, a star helps others find it.

[![Star History Chart](https://api.star-history.com/svg?repos=JohnWayneeee/casely-qa-skill&type=Date)](https://star-history.com/#JohnWayneeee/casely-qa-skill&Date)

---

## License

[MIT](https://github.com/johnwayneeee/casely-qa-skill/tree/HEAD/LICENSE) — free to use, modify, and distribute.

---

<div align="center">

*Made for QA engineers who were hired to find bugs, not write documents.*

**[casely.digital](https://casely.digital/) — the hosted version for teams**

</div>

🚀 Casely — AI QA Test Case Generator

Turn messy PDF requirements into TestRail-ready test cases in 8 minutes.
Free, open-source AI skill for Claude Code, Cursor, and any AI IDE.

The problem every QA engineer knows

You were hired to find bugs. Instead, you spend 40% of your week writing test cases.

Requirements scattered across 10 PDF files. Every project has different column names — manual reformatting every time. A single module takes 2 days to document. Then the TestRail import fails because the headers don't match.

❌ Requirements buried in PDF/DOCX/XLSX files with no structure
❌ Each project reinvents column names — manual mapping every sprint
❌ 50 test cases = 2–3 business days of repetitive writing
❌ TestRail import breaks due to column mismatches
❌ No coverage plan = missed edge cases and bugs reaching production

Every hour writing test cases is an hour not spent testing.

What Casely does

Casely is your Virtual QA Lead. It reads your requirements, learns your team's exact format, and writes the test cases — atomic, structured, and import-ready.

Requirements PDF → /parse → /style → /plan → /generate → /export → TestRail ✅

Step	Command	What happens
Extract	`/parse`	Docling OCR pulls tables and text from any PDF/DOCX
Learn	`/style`	Casely reads your existing Excel and clones your column structure
Plan	`/plan`	Generates a coverage map: "47 tests across 6 modules"
Write	`/generate`	Creates atomic `.md` test cases — one file per test
Deliver	`/export`	Batch-converts everything to TestRail-ready Excel

Open exports/functional_TC001_happy_path.xlsx — it's a 1:1 match to your team's template, ready for immediate import.

Why QA teams switch to Casely

	Casely	Manual writing	Traditional tools
Parses any format (PDF/DOCX/XLSX)	✅	❌	❌
Matches your column structure	✅	❌	❌
Generates a test plan automatically	✅	❌	❌
1 test case = 1 file (atomic, reviewable)	✅	❌	⚠️ bulk only
TestRail / Qase ready out of the box	✅	❌	⚠️ manual fix
Works with English and Russian docs	✅	❌	❌
Free, runs locally, no cloud lock-in	✅	✅	❌

⚡ Quick Start

Prerequisites: Python 3.10+ and uv

Option A — Skills CLI (recommended)

# with bunx
bunx skills add JohnWayneeee/casely-qa-skill

# or npx
npx skills@latest add JohnWayneeee/casely-qa-skill

Option B — Clone & run

git clone https://github.com/JohnWayneeee/casely-qa-skill.git
cd casely-qa-skill
uv sync

Drop your files into the project folder and run the workflow:

/init my-project
/parse
/style
/plan
/generate functional AccountTransfer
/export

8-minute walkthrough

Step 1 — Initialize

/init my-project

Drop two files into projects/my-project/input/:

requirements.pdf — your specification document
example.xlsx — an existing test case file your team already uses

Casely learns your column structure from the example. No config needed.

Step 2 — Parse & plan

/parse    # extracts text and tables from your PDF via Docling OCR
/style    # reads your example.xlsx and clones the column structure exactly
/plan     # output: "Detected 6 modules. Recommended: 47 test cases."

You now have a full coverage strategy before writing a single test case.

Step 3 — Generate

/generate functional AccountTransfer

Casely writes 10+ atomic .md files — one per test case. Review and edit in any text editor or Git UI before exporting.

Step 4 — Export

/export

Every .md file becomes a separate Excel file matching your team's template exactly. Import into TestRail, Qase, or any TMS — no manual reformatting.

Command reference

Command	Action	Notes
`/init [name]`	Scaffold project workspace	Run once per project
`/parse`	Extract from PDF/DOCX	High-fidelity OCR via Docling
`/style`	Clone your Excel column format	Reads your example — zero config
`/plan`	Build ISTQB-aligned coverage map	Shows module breakdown and estimates
`/generate [type]`	Write atomic test cases	Produces one `.md` per test
`/export`	Convert all `.md` to Excel	Batch output, import-ready

Supported generation types: functional, negative, integration, boundary, smoke, security

Under the hood

Docling Engine — advanced OCR and table extraction for complex, multi-column PDFs including scanned documents
Atomic design — 1 test case = 1 source file = 1 Excel; no monolithic spreadsheets to untangle
Style Guide System — no hardcoded columns; Casely learns from your existing files and replicates the exact structure
Language agnostic — generates test cases in English or Russian, matching the language of your documents

Project structure

casely-qa-skill/
├── skill/casely/
│   ├── SKILL.md              # AI skill definition and full workflow
│   ├── scripts/
│   │   ├── casely_parser.py  # Document → Markdown converter (Docling)
│   │   └── export_to_xlsx.py # Markdown → Excel exporter
│   ├── references/           # Technical reference docs
│   └── evals/                # Evaluation test cases
├── docs/
│   └── hosted-web-version.md # Hosted version details
├── assets/                   # Images and branding
├── pyproject.toml            # Python dependencies
└── marketplace.json          # Skill marketplace metadata

Hosted version for teams

This open-source skill runs locally in your AI IDE.

If you want a browser UI, file uploads, team review flow, and zero local setup:

casely.digital — join the early access list

The hosted version is built for QA teams that want to turn requirements into review-ready test cases without writing code or running local scripts.

Learn more about the hosted version →

FAQ

Does it work with scanned PDFs?

Yes. Casely uses Docling's OCR pipeline, which handles scanned documents, embedded tables, and mixed-format pages.

Can I use my own Excel column structure?

That's the core feature. Drop your existing template into input/ and run /style. Casely reads your column names and replicates them exactly — no configuration file needed.

Does it support Russian-language requirements?

Yes. The parser and generator work with English and Russian documents. Column names in your Excel template are preserved as-is.

Which test management systems are supported for import?

Any TMS that accepts Excel import: TestRail, Qase, Zephyr, Xray, and plain Excel. Because Casely replicates your own column structure, the output matches whatever format your team already uses.

What's the difference between the skill and the hosted version?

The skill runs locally inside your AI IDE (Claude Code, Cursor, etc.) — full control, no cloud, free forever.
The hosted version at casely.digital adds a browser UI, team review workflows, and cloud storage — no local setup required.

Contributing

Contributions are welcome. Please read CONTRIBUTING.md before opening a pull request.

🐛 Found a bug? Open an issue
💡 Have an idea? Request a feature
⭐ Did it help? Star the repo — it helps other QA engineers find this tool

⭐ Star History

If Casely saved you a work week, a star helps others find it.

License

MIT — free to use, modify, and distribute.

Made for QA engineers who were hired to find bugs, not write documents.

casely.digital — the hosted version for teams