Menu
Restores a previous AI CLI session — either hot-loads into current context or prepares for /resume.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Removes synced session data from the local staging directory with double confirmation. All cleaned sessions are tombstoned to prevent re-sync.
Deletes a specific session from the sync repository. Requires double confirmation before deletion.
Lists all synced AI CLI sessions. Pulls latest from remote first, then displays.
Prune old synced sessions that exceed a retention period.
Shows the current status of Synctx — sync directory, audit log, prerequisites, and CLI paths.
Full bidirectional sync — pulls from remote, stages local sessions, scans for secrets, and pushes.
Basé sur la classification professionnelle SOC
| name | restore-session |
| description | Restores a previous AI CLI session — either hot-loads into current context or prepares for /resume. |
To restore a session, follow these steps strictly:
Pull the latest data:
node ./scripts/sync-engine.js restore
List Available Sessions:
node ./scripts/sync-engine.js list
To see tagged sessions:
node ./scripts/sync-engine.js tags
Prompt the User: Ask which session they want to restore. They can provide a session UUID, partial ID, a tag name (e.g., "auth-refactor"), or "current" to target the most recent active session.
Ask the user how they want to restore:
Option A: Hot-load into this session — I'll read the session history into our current conversation so you can continue seamlessly. Your current session ID stays the same.
Option B: Prepare for /resume — I'll restore the session files so you can switch to it with
/resume <id>. You'll get the original session with its full history.
If Option A (Hot-Load):
Explain to the user:
ℹ️ Hot-load reads the old session's context into this conversation. Your current session ID stays the same — this is context injection, not a session switch. The original session remains untouched.
Resolve tags if needed. Tags are stored in ~/.synctx/.tags.json.
Auto-tag the current session: If the restored session had a tag (e.g., "auth-refactor"), create a continuation tag for the current session so the user can find it later:
node ./scripts/sync-engine.js tag <current-session-id> <original-tag>-continued
For example, if restoring "auth-refactor", tag the current session as "auth-refactor-continued".
Tell the user: 📌 Tagged this session as "<tag>-continued" for easy reference.
Read the session files from the staging directory:
For Copilot sessions:
cat ~/.synctx/github-copilot/session-state/<session-id>/events.jsonl
For Claude sessions:
find ~/.synctx/claude/projects/<project-name> -type f | head -20
cat ~/.synctx/claude/projects/<project-name>/<session-file>
Parse the conversation history and adopt it into your current context. Summarize the key points. Print: "✅ Session hot-loaded into active memory! Here is where we left off: [brief summary]."
If Option B (Prepare for /resume):
Run restore with the tag or session ID to copy it back to the CLI session directory:
node ./scripts/sync-engine.js restore <tag-or-session-id>
Tell the user:
✅ Session restored. To switch to it, run:
/resume <session-id>
IMPORTANT: Never delete or modify the user's original session directories. Restore only copies FROM staging TO the CLI directory.