📁 Prompt Scope

Turn your ChatGPT history into something you can actually explore.

Prompt Scope is a local-first Streamlit app that lets you import, analyze, categorize, and extract insights from your ChatGPT conversations — without sending your data anywhere.

---

✨ What It Does

Prompt Scope transforms raw ChatGPT exports into a structured, searchable, analyzable dataset.

• 📥 Import your ChatGPT conversations.json
• 🗂️ Organize chats with categories
• 🔍 Search across titles + full content
• 📊 Generate insights about how you use ChatGPT
• 🧠 Detect topics, keywords, and patterns
• 🔎 Analyze message-level behavior (words, phrases, trends)
• 📤 Export everything back to JSON

All processing happens locally on your machine.

---

🧠 Why This Exists

ChatGPT stores your conversations, but it doesn’t help you understand them.

Prompt Scope answers questions like:

• What do I actually use ChatGPT for?
• What topics come up over and over?
• Which conversations go deep vs. quick hits?
• What patterns exist in how I write or prompt?

It turns your chat history into something closer to:

• a dataset
• a personal archive
• a reflection tool

---

🏗️ How It Works

1. Import

Upload your ChatGPT export JSON.

Prompt Scope:

• Normalizes different export formats
• Extracts messages (even messy nested structures)
• Stores everything in a local SQLite database

---

2. Explore

Browse your conversations with:

• Search (title + content)
• Category filters
• Sorting (newest, oldest, A→Z)
• Inline previews

---

3. Analyze

🔹 Insights Tab

• Topic detection (rule-based, no AI calls)
• Keyword extraction
• Conversation depth analysis
• Global summaries of your usage

Powered by deterministic logic (no external APIs)

---

🔹 Patterns Tab

Message-level analysis:

• Most common words
• Bigrams / trigrams
• Conversation frequency
• Monthly usage trends

Includes:

• Code filtering
• Speaker filtering (user vs assistant)
• Token cleaning

---

4. Organize

• Create categories
• Assign chats to multiple categories
• Filter by combinations

Everything is stored locally in SQLite.

---

5. Export

Export your enriched dataset:

{
  "schema_version": 1,
  "exported_at": "...",
  "categories": [...],
  "chats": [...]
}

---

🔐 Privacy First

• No API calls
• No tracking
• No cloud sync
• No external dependencies

Your data never leaves your machine.

---

⚡ Performance Design

Prompt Scope is optimized for large chat histories:

• Content truncation for fast UI rendering
• Cached analysis results
• Lazy loading (analysis only runs when requested)

Key limits:

• MAX_ANALYSIS_CHARS = 20,000
• MAX_PREVIEW_CHARS = 4,000

→ see

---

🧱 Tech Stack

• Frontend/UI: Streamlit
• Storage: SQLite (local file)
• Data processing: Pandas
• Visualization: Plotly / Altair
• Language: Python 3

Dependencies: → see

---

🚀 Getting Started

1. Clone the repo

git clone https://github.com/yourusername/prompt-scope.git
cd prompt-scope

---

2. Create a virtual environment

python -m venv .venv
.venv\Scripts\activate   # Windows

---

3. Install dependencies

pip install -r requirements.txt

---

4. Run the app

streamlit run app.py

---

📂 Project Structure

prompt-scope/
├── app.py                      # Main Streamlit app
├── promptscope.db              # Local SQLite database
├── conversation_insights.py    # Topic + insight engine
├── chat_patterns.py            # Message-level analysis
├── requirements.txt

---

🧩 Key Features (Deep Dive)

🧠 Deterministic Topic Detection

No AI, no embeddings — just rule-based scoring.

• Keyword pattern matching
• Multi-hit thresholding
• Topic confidence scoring

---

🔍 Robust Chat Parsing

Handles messy real-world exports:

• Nested JSON structures
• mapping-based exports
• Transcript-style [user] / [assistant] logs
• Mixed formats

---

🧼 Noise Reduction

Removes:

• JSON structural tokens (content, parts, etc.)
• Code blocks (optional)
• Stopwords

So your analysis reflects actual human language.

---

⚡ Local-First Architecture

• SQLite for persistence
• No server required
• Works offline
• Instant startup after import

---

🧪 Use Cases

• Personal knowledge mining
• Prompt engineering analysis
• Writing pattern discovery
• AI usage reflection
• Research / journaling analysis
• Dataset generation for future tools

---

🔮 Future Ideas

• Timeline visualizations
• Conversation clustering
• Prompt quality scoring
• Export to CSV / embeddings-ready formats
• GitHub-style diff for conversations

---

🤝 Contributing

Contributions are welcome — especially:

• Better topic rules
• New analysis modules
• UI improvements
• Export formats

---

📜 License

MIT

---

💭 Final Thought

ChatGPT remembers everything.

Prompt Scope helps you understand it.