SharedContext

概要

SharedContext functions as a local-first memory layer, tackling the problem of fragmented and ephemeral AI assistant memory. It consolidates the context from various AI clients like Cursor, Claude Code, and Codex into a single, encrypted, and portable layer that you own. By running as an MCP server, it allows agents to store and recall structured facts and conversation history, which is then encrypted client-side using AES-256-GCM and persistently synced to Arweave. This ensures your AI's memory follows you across different machines and sessions, recoverable with a simple 12-word phrase, thereby eliminating the need to re-explain project decisions or coding preferences repeatedly.

主な機能

• Stores and recalls structured facts (preferences, decisions, architecture) across sessions.
• Allows sharing specific conversations via encrypted links for secure collaboration.
• Encrypts all data locally with AES-256-GCM before secure upload to Arweave.
• Retrieves past conversations from any AI client to continue discussions.
• Enables full context restoration on any machine using a 12-word recovery phrase.
• 25 GitHub stars

ユースケース

• Maintaining consistent project context and architectural decisions for AI coding assistants across sessions and tools.
• Seamlessly restoring an AI agent's full memory and past conversations when switching development environments or machines.
• Securely sharing specific AI conversation threads or learned facts with other developers or for documentation purposes.