openclaw/docs/install/migrating.md

summary, read_when, title

summary	read_when	title
Migration hub: cross-system imports, machine-to-machine moves, and plugin upgrades	You are moving OpenClaw to a new laptop or server You are coming from another agent system and want to keep state You are upgrading an in-place plugin	Migration guide

OpenClaw supports three migration paths: importing from another agent system, moving an existing install to a new machine, and upgrading a plugin in place.

Import from another agent system

Bundled migration providers bring instructions, MCP servers, skills, model config, and (opt-in) API keys into OpenClaw. Plans are previewed before any change and secrets are redacted in reports. Standalone openclaw migrate is backed by a verified backup; fresh onboarding imports instead stage and verify local artifacts before publishing them with configuration committed before any irreversible external activation.

Import Claude Code and Claude Desktop state, including `CLAUDE.md`, MCP servers, skills, and project commands. Import Hermes config, providers, MCP servers, memory, skills, and supported `.env` keys.

The CLI entry point is openclaw migrate. Onboarding can also offer migration when it detects a known source (openclaw onboard --flow import).

Move OpenClaw to a new machine

Copy the state directory (~/.openclaw/ by default) and your workspace to preserve:

• Config — openclaw.json and all gateway settings.
• Auth — per-agent auth-profiles.json (API keys plus OAuth), plus any channel or provider state under credentials/.
• Sessions — conversation history and agent state.
• Channel state — WhatsApp login, Telegram session, and similar.
• Workspace files — MEMORY.md, USER.md, skills, and prompts.

Run `openclaw status` on the old machine to confirm your state directory path. Custom profiles use `~/.openclaw-/` or a path set via `OPENCLAW_STATE_DIR`.

Migration steps

On the **old** machine, stop the Gateway, then create and verify a portable archive:

```bash
openclaw gateway stop
mkdir -p ~/Backups/openclaw
openclaw backup create --output ~/Backups/openclaw --verify
```

Stop the Gateway before taking a machine-move snapshot. A raw copy of a
changing SQLite database can capture mismatched database and WAL files;
quiescing the Gateway also keeps the rest of the state tree stable. If you
use multiple profiles, run the command once with each profile selected.

[Install](/install) the CLI (and Node if needed) on the new machine. It is fine if onboarding creates a fresh `~/.openclaw/` — you overwrite it next. Transfer the generated `.tar.gz` archive via `scp`, an external drive, or another protected channel. On the new machine, restore it to a fresh staging directory:

```bash
openclaw backup restore <archive.tar.gz> --target ~/openclaw-restored
```

Restore never activates in place. With the Gateway stopped, use the
restored `manifest.json` mapping to move the state and workspace assets to
their recorded destinations, or point `OPENCLAW_STATE_DIR` at the restored
state asset. Confirm ownership matches the user that will run the Gateway.

<Warning>
Restoring older channel state can desynchronize ratcheting credentials such
as WhatsApp. Approvals and delivery/dedupe state also roll back, and plugin
`node_modules` trees must be reinstalled. See [Restore a full archive](/install/backups#restore-a-full-archive).
</Warning>

On the new machine, run [Doctor](/gateway/doctor) to apply config migrations and repair services:

```bash
openclaw doctor
openclaw gateway restart
openclaw status
```

If Telegram or Discord uses the default env fallback (TELEGRAM_BOT_TOKEN or DISCORD_BOT_TOKEN), verify the migrated state-dir .env contains those keys without printing the secret values:

awk -F= '/^(TELEGRAM_BOT_TOKEN|DISCORD_BOT_TOKEN)=/ { print $1 "=present" }' ~/.openclaw/.env

openclaw doctor also warns when an enabled default Telegram or Discord account has no configured token and the matching env variable is unavailable to the doctor process.

Common pitfalls

If the old gateway used `--profile` or `OPENCLAW_STATE_DIR` and the new one does not, channels will appear logged out and sessions will be empty. Launch the gateway with the **same** profile or state-dir you migrated, then rerun `openclaw doctor`. The config file alone is not enough. Model auth profiles live under `agents//agent/auth-profiles.json`, and channel and provider state lives under `credentials/`. Always migrate the **entire** state directory. If you copied as root or switched users, the gateway may fail to read credentials. Ensure the state directory and workspace are owned by the user running the gateway. If your UI points at a **remote** gateway, the remote host owns sessions and workspace. Migrate the gateway host itself, not your local laptop. See [FAQ](/help/faq#where-things-live-on-disk). The state directory contains auth profiles, channel credentials, and other provider state. Store backups encrypted, avoid insecure transfer channels, and rotate keys if you suspect exposure.

Verification checklist

On the new machine, confirm:

• openclaw status shows the gateway running.
• Channels are still connected (no re-pairing needed).
• The dashboard opens and shows existing sessions.
• Workspace files (memory, configs) are present.

Upgrade a plugin in place

In-place plugin upgrades preserve the same plugin id and config keys but may move on-disk state into the current layout. Plugin-specific upgrade guides live alongside their channels:

• Matrix migration: encrypted-state recovery limits, automatic snapshot behavior, and manual recovery commands.

Related

• openclaw migrate: CLI reference for cross-system imports.
• Install overview: all installation methods.
• Doctor: post-migration health check.
• Uninstall: removing OpenClaw cleanly.