⚒️ CaseForge

Don't just read papers. Build knowledge from them.

An extensible AI workflow for extracting structured case studies from academic literature.

Who is this for?

If you read 3+ papers a week and want to turn them into searchable, reusable knowledge assets — CaseForge is built for you.

---

Why CaseForge?

Most AI tools summarize papers.

CaseForge transforms papers into structured knowledge that can be searched, compared, and reused.

Instead of asking:

"What is this paper about?"

CaseForge asks:

"What knowledge can we extract from this paper?"

---

Philosophy

CaseForge is not another paper chatbot.

It is a workflow engine for turning academic literature into structured knowledge.

Small. Simple. Extensible. Reusable.

---

What It Does

  📄 Paper (PDF / TXT / MD)
        │
        ▼
  📖 Reader      ← pdfplumber / markitdown
        │
        ▼
  🧠 AI Extract  ← 10-field deep extraction
        │
        ▼
  📊 Export      ← HTML / Markdown / JSON / Word

---

Demo

python main.py --demo

---

Quick Start

git clone https://github.com/modusensus/CaseForge.git
cd CaseForge
pip install -r requirements.txt
cp config.example.py config.py     # add any one API key
python main.py --demo

---

Features

• ✅ 8-field structured case extraction
• ✅ Export to HTML / Markdown / JSON / Word
• ✅ Prompt-driven workflow (swap disciplines without changing code)
• ✅ Multi-provider LLM support (5 APIs, including free tier)
• ✅ Local TF-IDF fallback when API fails
• ✅ Dedup cache (never pay twice for the same paper)
• ✅ Academic search (Semantic Scholar / OpenAlex / CORE)

---

Example Output

## 案例标题

🌍 研究背景
城市湿地公园面临生态保护与开发的结构性张力...

🎯 研究对象/目的
分析杭州西溪湿地公园的开发争议与协调机制...

🧪 研究方法
案例研究法，数据来源包括政策文件、生态监测数据...

📈 核心发现
围栏式保护不可持续，需利益共享+多主体协商...

💡 创新点
将社区协调机制与规划管控相结合进行系统分析...

✨ 一句话总结
社区利益共享与协商机制需前置到规划阶段。

---

Architecture

prompts/           ← Discipline-specific extraction templates
    │
    ▼
api_client.py      ← Unified interface for 5 LLM providers
    │
    ▼
exporters.py       ← HTML / Markdown / JSON / Word output
    │
    ▼
search.py          ← Semantic Scholar / OpenAlex / CORE

---

Extensible by Design

Add a new discipline without touching any code:

prompts/
├── urban_design.md    ← default
├── education.md       ← contributed
├── medicine.md        ← coming soon
└── law.md             ← coming soon

---

Roadmap

• ✅ Markdown / JSON / Word / HTML export
• ✅ Multi-provider LLM support
• ✅ Prompt plugin system
• ✅ Academic search (3 databases)
• □ Prompt marketplace
• □ MCP / Skills integration
• □ Web UI

---

Contributing

We welcome contributions — especially:

• Prompt templates for new disciplines
• Readers for new input formats
• Exporters for new output formats
• Tests and documentation

See CONTRIBUTING.md.

---

MIT License · Changelog