Cross-machine session sync: connect Claude Code instances across multiple machines
Feature Request: Cross-Machine Session Synchronization
The Problem
Currently each Claude Code session is fully independent. There's no way to coordinate work across multiple machines from a single conversation or shared context.
The Use Case
Real-world development often spans multiple machines:
- Mac — development, planning, code editing
- Windows — compilation (e.g. MetaEditor/MQL5), platform-specific testing
- Linux — CI, environment checks, server-side validation
- Web (claude.ai) — monitoring progress, answering questions on the go
Today, the user must manually relay information between independent sessions. This is slow and error-prone.
The Opportunity
Claude Code already has the agent↔server connection infrastructure. Extending this to allow session synchronization across machines and clients would be transformative:
- Unified conversation — one conversation context, multiple execution environments
- Cross-machine tool execution — "compile on Windows, edit on Mac" from a single prompt
- Web as observer/commander — see what agents are doing, issue commands from claude.ai
- Shared context — all machines see the same plan, progress, and state
Proposed Behavior
- Register multiple Claude Code instances (different machines) under one session/workspace
- Route commands to specific machines ("run this on Windows", "check this on Linux")
- Web client can observe all connected agents and send commands to any of them
- Shared conversation history across all connected instances
Why This Matters
This would turn Claude Code from a single-machine tool into a distributed development orchestrator. Many developers work across OS boundaries (cross-compilation, platform-specific testing, mobile + backend). The infrastructure for agent↔server communication already exists — this is a natural extension.
🤖 Filed via Claude Code on behalf of a user with a Mac+Windows+Linux cross-platform MQL5 development workflow.
This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗