Summary

MineContext is an open-source, proactive, context-aware AI partner designed to enhance clarity and efficiency for knowledge workers, content creators, and learners. By capturing and comprehending a user's digital environment through screenshots and content analysis, it actively delivers synthesized insights, summaries, and to-do lists, aiming to transform scattered information into actionable intelligence. Its local-first approach prioritizes user privacy while offering a customizable, cost-effective alternative to cloud-based AI assistants.

How It Works

MineContext functions as a proactive, context-aware AI partner by capturing and comprehending a user's digital world through screenshots and content analysis, with planned expansion to multimodal sources like documents and videos. It leverages a contextual engineering framework to actively deliver synthesized information, such as daily summaries, to-do lists, and insights, directly to the user's interface. This approach aims to bring clarity and efficiency by intelligently resurfacing relevant context during creative tasks without overwhelming the user. Its local-first data storage prioritizes user privacy and security, distinguishing it from cloud-dependent AI assistants.

Quick Start & Requirements

Installation involves downloading the latest release (a Mac app is available) and potentially disabling the quarantine attribute via sudo xattr -d com.apple.quarantine. Backend setup requires cloning the repository, creating a Python virtual environment, and installing dependencies (pip install -r requirements.txt). Users must configure API keys for supported LLM providers (Doubao, OpenAI, with Ollama planned) and grant screen recording permissions. Initial setup and value generation may require some time. Links to Discord and WeChat/Lark groups are provided for community engagement.

Highlighted Details

• Comprehensive Digital Context: Captures entire digital workflow via screen screenshots and planned multimodal sources, unlike single-conversation AI.
• Privacy-First & Local Storage: All data processed and stored locally, ensuring security without cloud reliance.
• Proactive & Diverse Insights: Delivers auto-generated content like daily summaries, todos, and activity reports, going beyond reactive chat assistance.
• Open-Source & Customizable: Allows inspection, modification, and building upon the codebase.
• Cost-Effective API Usage: Enables use of personal API keys, avoiding expensive premium subscriptions.

Maintenance & Community

Community support is available via a Discord Group and WeChat/Lark/Red Note Group. Users can report errors and issues through GitHub Issues or contact via Email Support.

Licensing & Compatibility

This repository is licensed under the Apache 2.0 License, which permits commercial use and integration into closed-source projects.

Limitations & Caveats

Current context capture primarily relies on screen screenshots, with broader multimodal support (documents, images, video, code, external app data) planned for future development. Initial backend setup may take several minutes, and the system requires time to generate meaningful insights from collected data. The Mac application requires specific steps to disable the quarantine attribute before execution.