Productivity
Everywhere
An on-screen aware AI assistant for your desktop that sees your current app and acts across apps.
★ 6.3k starsBUSL-1.1
View guide →A proactive, context-aware AI partner that builds context from your screen and documents.
MineContext is an open-source productivity project by volcengine (volcengine/MineContext on GitHub). It has 5,537 GitHub stars and 414 forks, is released under the Apache-2.0 license, and was last updated on GitHub on 7 May 2026.
Decide which sources matter — screen activity, documents, or both — and review the privacy implications before enabling capture.
Clone the repo and sync Python dependencies with uv.
Set your LLM, vision-language and embedding providers (OpenAI or Doubao are supported) in the configuration and prompt templates.
Start the backend server, then run the Electron frontend to see your context timeline and proactive suggestions.
Instead of waiting for questions, it builds context continuously and surfaces relevant help, in the style of ChatGPT Pulse.
Context acquisition modules monitor screenshots and documents through an extensible capture interface.
Chunking strategies, entity extraction and normalization, and context merging/deduplication turn raw captures into usable memory.
Uses vision-language models and vision embeddings (the reference setup uses Doubao models) to understand screen content, not just text.
Stores context in SQLite and ChromaDB behind a unified storage interface for similarity search over your history.
An Electron/React desktop client talks to a FastAPI backend with WebSocket support for real-time updates.
Clone MineContext and enter the directory.
git clone https://github.com/volcengine/MineContext.git
cd MineContextInstall uv if needed and sync the Python dependencies (this creates the virtual environment).
curl -LsSf https://astral.sh/uv/install.sh | sh
uv syncActivate the environment and build/run the app as the README's development section describes.
source .venv/bin/activate
./build.sh # see README for the exact commandMineContext is maintained by volcengine, ByteDance's Volcano Engine team.
Yes — it is open source under the Apache-2.0 license. You pay for any hosted model APIs you configure.
The reference configuration uses hosted LLM/VLM/embedding services (OpenAI or Doubao), so you need an account with your chosen provider.
Those focus on personal screen history and search; MineContext is an open-source, self-hosted take that adds a full context-processing pipeline and proactive, LLM-driven assistance.
Clone the repo, uv sync the backend, configure your model providers, and run the server and desktop frontend.
An on-screen aware AI assistant for your desktop that sees your current app and acts across apps.
Model Context Provider designed for Amazing Marvin productivity app.
Free productivity tool that 🗓️ a caldav model context protocol (mcp) server to expo.
Your own AI news radar — daily briefings in English and Chinese from RSS, Hacker News, Reddit, X and more.
Open Brain (OB1): one database, one AI gateway, one capture channel — shared memory for every AI you use
Deta Surf — personal AI notebooks that organise files and webpages and generate notes from them.
The most-starred other Productivity projects in the LibreAI catalogue — the realistic shortlist if MineContext is not the right fit.
Real-time global intelligence dashboard: AI news aggregation, geopolitics and infrastructure tracking
Memos — a personal timeline for quick notes: capture short memos in seconds, find them later by search, tag or date; open source and self-hosted
Joplin — the privacy-focused note-taking app: Markdown notes and to-dos on every platform, offline-first, with end-to-end encrypted sync you control
Open-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, mult…
Logseq — a privacy-first platform for knowledge management: an outliner over your local Markdown files, with a graph, journals and spaced repetition built in
Free and open — grab the source, star the repo, and start creating.