Video.mp4
An open-source CLI tool for academic paper writing, powered by your choice of AI model. Supports Claude, DeepSeek, GLM, Qwen, Moonshot, ERNIE and more — bilingual (English & Chinese), multiple journal templates, and major citation formats.
npm install -g academic-paper-skillSet any one of the following API keys and the tool auto-detects it:
macOS / Linux
export DEEPSEEK_API_KEY=your-key # DeepSeek (recommended for CN users)
export GLM_API_KEY=your-key # GLM — glm-4-flash tier is FREE
export DASHSCOPE_API_KEY=your-key # Qwen (通义千问)
export MOONSHOT_API_KEY=your-key # Moonshot (Kimi)
export QIANFAN_API_KEY=your-key # ERNIE (文心一言)
export ANTHROPIC_API_KEY=your-key # Claude (international)Windows PowerShell
$env:DEEPSEEK_API_KEY = "your-key"
$env:GLM_API_KEY = "your-key"
$env:DASHSCOPE_API_KEY = "your-key"
$env:MOONSHOT_API_KEY = "your-key"
$env:QIANFAN_API_KEY = "your-key"
$env:ANTHROPIC_API_KEY = "your-key"Windows tip: The above only lasts for the current session. To save permanently, run:
[System.Environment]::SetEnvironmentVariable("DEEPSEEK_API_KEY", "your-key", "User")
Then run:
aps draft "Your Research Topic"aps providersSupported providers:
✓ deepseek DeepSeek (深度求索) [DEEPSEEK_API_KEY is set] ← auto-selected
✗ glm GLM (智谱清言) [GLM_API_KEY not set] (glm-4-flash is free)
✗ claude Claude (Anthropic) [ANTHROPIC_API_KEY not set]
...
| Command | Description |
|---|---|
aps draft <topic> |
Generate a paper outline or full draft |
aps polish [text] |
Polish academic language and writing style |
aps cite <reference> |
Format references in any citation style |
aps abstract [text] |
Generate abstract from a full paper |
aps review [references] |
Generate literature review from a reference list |
aps caption <figure|table> <description> |
Generate figure or table caption |
aps providers |
List all providers and their API key status |
# Generate an IEEE paper outline (auto-detects provider)
aps draft "Federated Learning for Medical Image Analysis" --template ieee
# Use a specific provider
aps draft "Your Topic" --provider deepseek
aps draft "Your Topic" --provider glm --model glm-4-plus
# Generate a Chinese journal paper outline
aps draft "基于深度学习的医学图像分割" --template chinese --mode outline
# Polish a paragraph
aps polish "Our method is way better than all the other methods."
aps polish --file my_draft.md
cat section.md | aps polish
# Format citations
aps cite "LeCun Y et al. Deep learning. Nature. 2015;521:436-444." --style apa
aps cite "张伟等. 深度学习综述. 计算机学报. 2023." --style gbt7714
# Generate abstract from file
aps abstract --file paper.md
aps abstract --file paper.md --format structured # medical/life science
# Literature review
aps review --file refs.txt --style ieee
# Figure/table captions
aps caption figure "accuracy comparison of 5 models with error bars" --template ieee --number 3
aps caption table "三个数据集上各模型性能对比" --template chinese --number 2
# LaTeX output
aps draft "Topic" --template ieee --latex
# Full draft (not just outline)
aps draft "Topic" --template acm --mode full| Provider | Env Variable | Default Model | Notes |
|---|---|---|---|
| DeepSeek | DEEPSEEK_API_KEY |
deepseek-chat |
Affordable, strong reasoning |
| GLM | GLM_API_KEY |
glm-4-flash |
glm-4-flash is free |
| Qwen | DASHSCOPE_API_KEY |
qwen-plus |
Alibaba Cloud |
| Moonshot | MOONSHOT_API_KEY |
moonshot-v1-8k |
Long context |
| ERNIE | QIANFAN_API_KEY |
ernie-4.5-8k |
Baidu |
| Claude | ANTHROPIC_API_KEY |
claude-sonnet-4-6 |
International |
All Chinese providers support the OpenAI-compatible API — no extra dependencies needed.
Auto-detection: If multiple keys are set, the first one in the table above wins. Override with --provider.
Journal Templates: IEEE · ACM · Nature · Chinese Core Journals · Chinese Sci (理工类) · Chinese Social (社科类) · Chinese Science Series
Citation Styles: APA · MLA · Chicago · IEEE · ACM · GB/T 7714-2015
Output Formats: Markdown (default) · LaTeX (--latex)
Each command loads a crafted prompt from prompts/ and sends it to your chosen model with streaming — output appears token by token. Language is auto-detected from your input (Chinese input → Chinese output with GB/T 7714; English → APA).
Prompt improvements in prompts/ are the most impactful contributions. See CONTRIBUTING.md.
MIT © xiaoshuntian