cc-sync

Sync /resume history between Codex CLI and Claude Code. Conversations imported into either tool show up in its /resume picker with full user + assistant message history.

Install

# pipx (recommended, isolated env)
pipx install git+https://github.com/vipul-sharma20/cc-sync.git

# or from a local clone
pip install -e .

# or skip installing entirely
python3 -m cc_sync status

No dependencies outside the Python 3.10+ standard library.

Usage

# see what you have
cc-sync status

# preview before writing anything
cc-sync codex-to-claude --dry-run
cc-sync claude-to-codex --dry-run

# import
cc-sync codex-to-claude
cc-sync claude-to-codex

Both directions can be run independently or together. Order doesn't matter.

What gets synced

User messages and assistant text responses, session metadata (cwd, timestamps), and — for the Codex direction — sqlite thread registration so sessions appear in /resume.

Tool calls, file diffs, reasoning tokens, and usage stats are not transferred. They're tool-specific and not actionable on the other side.

Idempotency

Re-running is safe. Each import embeds the source session ID (sourceId in Claude Code entries, source_session_id in Codex session_meta). Subsequent runs detect these and skip already-imported sessions.

$ cc-sync codex-to-claude
  0 converted, 20 skipped (already imported)

Backups

Backups are created automatically before every write operation. Codex's history.jsonl and state_5.sqlite are copied with a .bak.<timestamp> suffix. Backups are never cleaned up automatically.

Custom data directories

cc-sync --codex-home /path/to/.codex --claude-home /path/to/.claude status

Or via environment variables: CODEX_HOME, CLAUDE_HOME.

Project structure

cc_sync/
    cli.py          entry point, argument parsing
    commands.py     codex_to_claude, claude_to_codex, show_status
    parsers.py      read Codex/Claude Code session JSONL into a common format
    builders.py     produce session files in the target tool's format
    dedup.py        scan for already-imported source session IDs
    codex_db.py     register imported threads in Codex's sqlite index
    paths.py        resolve data directories
    constants.py    shared import tags