[Feature Request] Shared memory between Claude Code and Claude Chat sessions in Desktop app

Resolved 💬 2 comments Opened Apr 14, 2026 by ceoimperiumprojects Closed Jun 2, 2026

Feature Request

Is your feature request related to a problem?

Claude Code (Desktop/Web) and Claude Chat (Cowork tab) currently maintain completely separate memory systems. Even when both surfaces live inside the same Claude Desktop app, they cannot access each other's memory. This means:

  • Context from casual chat conversations (preferences, project goals, decisions, emotional state) is invisible to Code sessions
  • Technical discoveries and project context from Code sessions don't carry over to Chat
  • Users have to manually repeat context when switching between Code and Chat tabs

Describe the solution you'd like

A unified memory layer that is shared across Claude Code sessions and Claude Chat sessions within the same Desktop app (and ideally claude.ai web as well). Specifically:

  • Claude Code should be able to read memories saved by Claude Chat (and vice versa)
  • The existing Memory system (Settings → Memory) should be accessible from both surfaces
  • This would create a seamless experience where Claude truly "knows you" regardless of which tab you're in

Describe alternatives you've considered

  • Manually copy-pasting context between Code and Chat sessions (current workaround)
  • Using external memory systems (Obsidian, custom MCP plugins) as a shared layer — works but requires extra setup and doesn't leverage the built-in Memory feature

Additional context

The new Desktop redesign is incredible — having Code and Cowork in one app is a game changer. The missing piece is shared memory between these surfaces. When a user tells Claude in Chat "I'm working on a B2B lead gen product" or "I prefer TypeScript", that context should be available when they switch to a Code session. This would make the Desktop app feel like one unified AI assistant rather than two separate ones sharing a window.

---
Submitted via Claude Code by an active Desktop user 🚀

View original on GitHub ↗

This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗