Teleporting a session from Web → Local silently saves 0 messages, then resume fails with "No conversation found"
Environment:
- Extension:
anthropic.claude-codev2.1.211 (linux-x64) - Client: VS Code Remote (devcontainer)
Steps to reproduce:
- Open the Claude Code sidebar, switch to the Web tab, pick an existing remote session.
- Trigger "teleport" to bring that session into the local window (e.g. via the branch-switch dialog / teleport action).
- Watch the extension log (
Claude VSCode.log).
Actual behavior:
- The extension logs the teleport as successful but with zero content:
````
Saved teleported session <local-session-id> with 0 messages
Teleport succeeded for session <remote-session-id> in 842ms (0 messages)
- It then immediately tries to auto-resume that new local session (
launch_claude ... resume: <local-session-id>), which fails because no transcript was ever written:
````
[ERROR] No conversation found with session ID: <local-session-id>
Error spawning Claude (on channel ...): Error: Claude Code returned an error result: No conversation found with session ID: <local-session-id>
- The failed session lingers in the sidebar carrying over a stale/unrelated title from a previous session, and requires deleting it (sometimes more than once —
delete_sessionhad to be sent twice before it cleared) to clean up.
Expected behavior:
- Teleport should either actually copy the remote conversation's messages, or fail loudly/visibly at teleport time (not two steps later as an opaque "Error spawning Claude" toast) so no orphaned empty session is created.
Suspected downstream effect: the Web session list (list_remote_sessions) and local session list appear to accumulate generic-titled duplicate entries (e.g. repeated VsCodeRemote | TroubleShoot) over multiple days, consistent with this same failure recurring on other teleport/resume attempts — and there's no bulk-delete for these orphaned entries, so cleanup means deleting them one at a time.
Request: please also consider a bulk "clear empty/orphaned sessions" action, since manual one-by-one deletion doesn't scale once this has happened repeatedly.
This issue has 1 comment on GitHub. Read the full discussion on GitHub ↗