Persistent AI memory across Claude Desktop & Cursor IDE
Developers use a pile of AI assistants — but each one lacks full project context. Vernon fixes that by becoming the universal context layer: a single source of truth shared across Claude, Copilot, Zed, Cursor, and more. Install once, and every AI tool understands your project instantly.
Context Sync offers read-only Notion integration via notion() tool with search and read actions. Users can query project documentation stored in Notion directly within their AI coding session, keeping context in one place.
Context Sync provides structure() and search() tools that let AI agents explore the codebase systematically, and read_file() for focused inspection. This reduces the need for users to manually paste file contents or explain directory layouts.
The remember() tool lets users log decisions, constraints, and caveats with type classification. These are stored persistently and recalled in future sessions, ensuring the AI maintains awareness of project-specific rules and prior choices.
Context Sync installs optional Git hooks (post-commit, pre-push, post-merge, post-checkout) that automatically capture context. The git tool then provides status, hotspots, coupling, and blame analysis, giving AI agents awareness of recent changes and code relationships.
Context Sync stores project identity, tech stack, active decisions, and constraints in a local SQLite database, making this information retrievable via MCP tools across sessions. Users call recall() at session start to recover prior context instead of re-briefing the AI model.