ContextBox

Capture and organize your working context from the command line

Documentation • Installation • Quick Start

---

What is ContextBox?

ContextBox captures, organizes, and queries your working context. Take screenshots, extract web content, and ask AI questions about everything you've captured.

Key Features:

Feature	Description
Screenshot Capture	Capture screen with automatic OCR text extraction
Web Extraction	Pull content from web pages, Wikipedia, YouTube
LLM backend	GitHub Models client for the Python API (contextbox ask is not implemented yet)
Search	Text search across captured contexts
Cross-Platform	Works on macOS, Linux, Windows

Installation

Using pip

Not on PyPI yet, so install from GitHub:

pip install git+https://github.com/jayhemnani9910/contextbox.git

From source

git clone https://github.com/jayhemnani9910/contextbox.git
cd contextbox
pip install -e ".[all]"

Optional dependencies

Run these from the cloned source folder:

# LLM features (GitHub Models)
pip install -e ".[llm]"

# OCR support
pip install -e ".[ocr]"

# YouTube extraction
pip install -e ".[youtube]"

# Everything
pip install -e ".[all]"

Quick Start

1. Capture your screen

contextbox capture

This takes a screenshot, extracts text via OCR, and stores it in the local database.

2. List captured contexts

contextbox list

3. Ask questions (not implemented yet in the CLI)

export GITHUB_TOKEN="your_github_token"
contextbox ask "What was I working on?"

4. Generate summaries (not implemented yet in the CLI)

contextbox summarize --all-contexts

CLI Commands

Command	Description
contextbox capture	Capture screenshot and extract context
contextbox list	List all stored contexts
contextbox search <query>	Search through contexts
contextbox ask <question>	Ask questions about context (AI), not implemented yet in the CLI
contextbox summarize	Generate context summaries (AI), not implemented yet in the CLI
contextbox stats	Show database statistics
contextbox export	Export contexts to file

Configuration

GitHub Token (for AI features)

export GITHUB_TOKEN="ghp_xxxxxxxxxxxx"

Supported AI Models

All models are free via GitHub Models:

Model	Best For
gpt-4o-mini	Fast responses (default)
gpt-4o	Complex reasoning
meta-llama-3.1-70b-instruct	Open-source alternative

Python API

from contextbox import ContextBox

app = ContextBox()

# Capture and store
context_id = app.capture_context({"active_window": {"title": "Terminal"}, "clipboard": "copied text"})

# Search and retrieve
results = app.database.search_contexts("keyword")
context = app.get_context(context_id)

Requirements

Platform	Screenshot Tool	OCR
macOS	Built-in	Tesseract
Linux	scrot / flameshot	Tesseract
Windows	pyautogui	Tesseract

License

MIT License

---

Documentation • Issues • Contribute