Readme.md

metadata

title: Problem Decoder
emoji: 🧠
colorFrom: violet
colorTo: indigo
sdk: docker
pinned: false
license: mit
short_description: >-
  Paste a transcription or upload a PDF — get problem analysis and brainstormed
  ideas powered by Claude.

🧠 Problem Decoder

An AI-powered agent that helps interns understand real-world problems and brainstorm creative solutions from meeting transcriptions or PDF documents.

Features

• Text input — paste any transcription, meeting notes, or problem description
• PDF upload — upload a text-based PDF doc (specs, reports, research)
• Problem analysis — clearly articulates the core problem, stakeholders, context
• Root cause analysis — breaks down why the problem exists
• Brainstorming — generates 5–8 diverse, creative solution directions
• Next steps — gives the intern 3 concrete actions to take immediately
• Streaming output — response streams in real time

Deploying to HuggingFace (Docker Space)

1. Create a new Space

Go to huggingface.co/new-space and choose Docker as the SDK.

Name it: problem-decoder

2. Upload these files

problem-decoder/
├── app.py
├── requirements.txt
├── Dockerfile
└── README.md

3. Add your Anthropic API key as a secret

Go to Settings → Repository secrets and add:

Name:  ANTHROPIC_API_KEY
Value: sk-ant-xxxxxxxxxxxxxxxx

4. Done — the Space builds and launches automatically

The build takes ~2 minutes on first deploy.

---

Local development

# Build the image
docker build -t problem-decoder .

# Run (replace with your actual key)
docker run -p 7860:7860 -e ANTHROPIC_API_KEY=sk-ant-xxx problem-decoder

Then open http://localhost:7860

---

Stack

Layer	Technology
UI	Gradio 4
LLM	Anthropic Claude Sonnet (streaming)
PDF parsing	pypdf
Container	Python 3.11 slim