Knowledge Management

ContextBox

CLI-first personal knowledge assistant. Captures screenshots, extracts text via OCR, embeds with sentence-transformers, and enables semantic search + AI Q&A — all from the terminal.

Source Docs

ClickSQLiteSentence-TransformersTesseract

contextbox search

Search by concept...

Fig. 1 — Semantic search demo. Queries matched against embedded contexts using cosine similarity.

CLI

Commands

contextbox capture— Screenshot + OCR + store to SQLite

contextbox search <query>— Semantic similarity search over stored contexts

contextbox ask <question>— AI Q&A powered by GitHub Models (GPT-4o)

contextbox summarize— AI-generated summaries of captured contexts

contextbox list— List all stored contexts with metadata

contextbox export --format json— Export contexts as JSON or CSV

Capabilities

What It Does

▸Cross-platform screenshot capture — Wayland (grim), X11 (scrot, maim), macOS, with graceful fallback

▸Tesseract OCR for text extraction from screenshots and images

▸Semantic search via sentence-transformers embeddings (all-MiniLM-L6-v2, 384-dim) stored as SQLite BLOBs

▸LLM-powered Q&A and summarization via GitHub Models (GPT-4o, Llama, Mistral — free tier)

▸Content extractors for URLs, YouTube transcripts, Wikipedia articles, and code files

▸Rich terminal UI with formatted tables, progress bars, and colored output

Content Sources

URLsYouTube transcriptsWikipediaCode filesOCR textClipboard

Tech Stack

PythonClickRichSQLiteSentence-TransformersTesseractGitHub Models

jayhemnani9910/contextbox

27 Modules6 ExtractorsMIT License