mycontext: A Local Layer That Helps AI Remember Your Work Context

Eliminate the hassle of explaining background knowledge to AI every time. MyContext unifies scattered work data from messengers, documents, meeting minutes, and more to create personalized working memory. It enables LLMs and agents to access this layer, understand the user's work context, and provide relevant answers.

Data is stored exclusively on the user's device. All information is kept in a local SQLite-based repository and never leaves the device. For retrieval, it combines full-text search, semantic search, and knowledge graphs to present answers with precise sources. Agents reference the original data when generating responses, ensuring reliability.

Currently supports instant messengers, documents, and meeting records, with planned expansion to calendars and email in the future. The 'Digital Self' feature learns the user's tone and past conversation history to draft replies. However, critical actions such as sending messages require explicit user approval.

This is an open-source project licensed under Elastic License 2.0. It can be freely used and modified for personal or internal corporate use, but providing it as a hosted service to others is prohibited. It is delivered as a desktop application built with Electron and React, currently in developer preview with frequent updates.