Use OpenClaw’s native openclaw backup command group instead of copying its live files. Create and verify a full archive before upgrades, resets, or a machine move; restore it into a new empty staging directory; then point OpenClaw at the restored state and run openclaw doctor before restarting the Gateway. For database-heavy installations, use verified SQLite snapshots, private Git history, or continuous replication rather than manual file copies.
Antihost offers VPS plans with one-click OpenClaw, NVMe storage and a 14-day free trial.
What an OpenClaw backup contains
A full archive is the simplest way to preserve an OpenClaw installation because it captures the state that makes the installation usable, not just one database file. A normal full archive can contain configuration, authentication profiles, provider and channel credentials, sessions, databases, and—unless you disable it—workspace data.
As an Amazon Associate I earn from qualifying purchases.
| Data | Included in a full archive? | Important qualification |
|---|---|---|
| OpenClaw configuration | Yes | The active JSON configuration is captured. A configuration-only backup does not parse or include files referenced through $include. |
| Authentication profiles | Yes | Treat the archive as a secret-bearing file. |
| Provider and channel credentials | Yes | These records can allow access to external services if the archive leaks. |
| Sessions | Yes | Session history may contain sensitive conversation or operational context. |
| Databases | Yes | OpenClaw creates a consistent archive or snapshot; do not substitute a live-file copy. |
| Workspaces | Optional | Use --no-include-workspace when workspace files are stored or protected separately. |
Local full archives are plaintext .tar.gz files. Encryption applied by an offsite storage destination does not encrypt the local file that was created first, so both copies need access controls and, where appropriate, encryption.
Choose the right backup method
The best method depends on whether you need a complete machine move, compact database recovery, version history, or protection against very small recovery-point gaps.
#1 Best Overall
- Easily store and access 2TB to content on the go with the Seagate Portable Drive, a USB external hard drive
- Designed to work with Windows or Mac computers, this external hard drive makes backup a snap just drag and drop
- To get set up, connect the portable hard drive to a computer for automatic recognition no software required
- This USB drive provides plug and play simplicity with the included 18 inch USB 3.0 cable
- The available storage capacity may vary.
| Method | Best use | Main trade-off |
|---|---|---|
| Full archive | One-off protection before an update, reset, uninstall, or machine move | Complete copies use more storage and re-upload the full archive. |
| SQLite snapshot | Compact, verified recovery of OpenClaw database state | Protects database state rather than every workspace file. |
| Git backup | Versioned, incremental database history | Requires a private repository and an explicit policy for secrets. |
Litestream or sqlite3_rsync |
Frequent or near-continuous database protection | Adds operational components and requires a tested restore plan. |
Use a full archive when you want the least complicated rollback or are moving OpenClaw to another server. Add SQLite or Git protection when databases change more frequently than you can reasonably archive. Continuous tools are appropriate only when you are prepared to monitor another service and test its recovery procedure.
Prepare before creating an archive
- Decide whether workspaces belong in this backup or are covered by a separate file-backup system.
- Choose at least one destination that is independent of the OpenClaw host. An external SSD that can be disconnected between runs provides a separate local copy; protected offsite object storage provides geographic separation.
- Restrict the destination to the administrators who need to restore OpenClaw. Do not put an archive in a broadly readable shared folder.
- Plan encryption for the archive at rest and during transfer. The local
.tar.gzremains plaintext unless you encrypt it yourself. - Record custom agent roots and their configured
agentDirpaths. A restore cannot preserve files that are outside the archived state unless those paths are also protected. - Keep enough free space for both the archive and a separate extracted restore test. Extraction is deliberately performed into a new target.
Create a full OpenClaw archive
Preview what will be captured
Start with a dry run so you can inspect the planned contents without committing to an archive:
Free tools Windows power users keep installed
One-click scans. No signup required.
openclaw backup create --dry-run --jsonThe JSON output is useful for automation and for checking whether workspace data and the expected state locations are included.
Create and verify a local archive
Write the archive to a dedicated backup directory and ask OpenClaw to verify it as part of creation:
openclaw backup create --output ~/Backups
openclaw backup create --verifyIf you want a single run to both choose the destination and verify the result, combine the applicable options in one command. Keep the resulting .tar.gz outside the live state directory so a cleanup or reinstall cannot remove both the source and its backup.
Exclude workspaces when they are handled elsewhere
openclaw backup create --no-include-workspaceThis produces a state-focused archive. Use it only when you have a separate, tested backup for the workspace files that your agents need.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Create a configuration-only safety copy
When the immediate risk is a malformed configuration, preserve the active JSON file before editing it:
openclaw backup create --only-config --output ~/Backups/openclaw --verifyThis operation saves only the active configuration. It does not resolve, parse, or include files referenced by $include. After repairing the configuration, make a new full archive so the repaired state and the rest of the installation are protected.
Apply retention and an offsite destination
Managed destinations can keep daily, weekly, and monthly generations. For example:
openclaw backup create --to offsite --keep-daily 7 --keep-weekly 4 --keep-monthly 12The names of configured destinations are installation-specific; offsite in this example must already exist in your OpenClaw backup configuration. Retention limits reduce storage growth but do not replace an occasional restore test.
Rank #2
- Easily store and access 5TB of content on the go with the Seagate portable drive, a USB external hard Drive
- Designed to work with Windows or Mac computers, this external hard drive makes backup a snap just drag and drop
- To get set up, connect the portable hard drive to a computer for automatic recognition software required
- This USB drive provides plug and play simplicity with the included 18 inch USB 3.0 cable
- The available storage capacity may vary.
Schedule backups and understand secret handling
OpenClaw’s managed backup jobs support archive and Git modes, with a documented default interval of 24 hours. Pushed schedules are designed to reduce the impact of a remote repository or storage compromise: credential-bearing tables and machine-state rows whose names begin with the secret prefix are redacted by default.
Use --include-secrets only when the remote destination is private, trusted, and governed as carefully as the OpenClaw host. This option enables full-fidelity remote history. Local non-push schedules retain full fidelity by default, but the local destination still needs filesystem permissions and encryption appropriate for credentials and session data.
Decide explicitly what a remote copy is for. If it is only a disaster-recovery reference, redaction may be preferable. If it must recreate the installation without access to the original host, full-fidelity storage may be necessary, but the repository or object store then becomes a high-value secret store.
Protect databases without copying live files
SQLite snapshots
For a compact database-focused backup, use OpenClaw’s snapshot mechanism. A global snapshot can be created with:
openclaw backup sqlite create --global --repository ~/Backups/openclaw-sqliteFor one agent, specify its agent name:
openclaw backup sqlite create --agent main --repository ~/Backups/openclaw-sqliteOpenClaw verifies database integrity, foreign keys, schema information, and hashes as part of its database backup workflow. Restore snapshots into a fresh target and retain the verification results; never merge an extracted snapshot into a live database directory.
Git-backed history
A private Git repository is useful when you need to inspect changes over time or retain incremental database history. Initialize and secure a repository that is not publicly readable, then use OpenClaw’s Git backup commands:
openclaw backup git create
openclaw backup git log
openclaw backup git verify
openclaw backup git restoreKeep secrets out of any pushed history unless the repository is private, access-controlled, and covered by your incident-response plan. The managed push behavior can redact credential-bearing records; review what is omitted before relying on a remote history for a complete rebuild.
Continuous or incremental protection
Litestream and sqlite3_rsync are documented alternatives when a 24-hour schedule leaves too much recoverable work at risk. They are not substitutes for planning: define where replicas live, how credentials are supplied, how a replica is promoted, and how you will validate a recovered database before reconnecting OpenClaw.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →The live-file rule
Do not copy a running database by copying files that end in .sqlite, -wal, -shm, or -journal. Those files can represent an in-progress transaction and an inconsistent set when copied independently. Use OpenClaw’s archive or snapshot commands, which coordinate the database state and verify the result.
Restore a full archive safely
1. Start with a trusted archive
Use an archive you created or one whose provenance you can verify. Because a full archive can contain credentials and sessions, treat an untrusted archive as executable-sensitive data: do not restore it over a production state directory.
Rank #3
- Easily store and access 1TB to content on the go with the Seagate Portable Drive, a USB external hard drive.Specific uses: Personal
- Designed to work with Windows or Mac computers, this external hard drive makes backup a snap just drag and drop. Reformatting may be required for Mac
- To get set up, connect the portable hard drive to a computer for automatic recognition no software required
- This USB drive provides plug and play simplicity with the included 18 inch USB 3.0 cable
- The available storage capacity may vary.
2. Extract into a new, empty staging directory
openclaw backup restore ARCHIVE.tar.gz --target ./restored-openclawThe target must be new and empty. Keep it outside the live OpenClaw state directory and outside configured agent directories. OpenClaw checks archive structure, manifests, links, and database snapshots before writing. If extraction fails, incomplete output is cleaned up rather than left as a misleading partial restore.
3. Preserve custom agent roots
If agents use custom roots, compare the restored paths with each configured agentDir. Either restore those directories as part of your protected data set or update the configuration to point to the restored locations. A state archive alone cannot recreate files that were deliberately stored elsewhere.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →4. Put the restored state in service
After the staging restore has passed verification, move it into the intended state location, or point OpenClaw at it with the OPENCLAW_STATE_DIR environment variable. Do not delete the old state until the restored installation has been checked and you have a fallback copy.
5. Run the doctor check before restarting
openclaw doctorRun the check while the Gateway is stopped or before you bring the restored state online. Resolve reported paths, permissions, configuration, or database issues first; only then restart the Gateway and test an agent, a configured channel, and any provider connection that the restored credentials are supposed to support.
Restore database backups and Git history
Database restore commands also require a fresh target. OpenClaw verifies integrity, foreign keys, schema, and hashes before accepting the restored database. Follow the restore subcommand corresponding to the snapshot or Git repository you created, and keep the restored database separate from the live one until validation is complete.
For Git history, use openclaw backup git restore against the selected revision or repository state, then run the same integrity checks and openclaw doctor workflow before reconnecting the Gateway. For SQLite snapshots, use the snapshot restore operation documented for your OpenClaw version and repository layout rather than manually replacing database files.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteA database-only recovery does not restore workspace documents, custom scripts, or files kept outside the OpenClaw state directory. If those files matter to an agent, combine the database recovery with the separate workspace backup.
Move OpenClaw to a new server
- Protect the source installation with a verified full archive. Include workspaces if the destination will not receive them from another backup.
- Securely transfer the archive to the new host. Do not leave a credential-bearing plaintext archive in a world-readable transfer directory.
- Install OpenClaw on the destination and stop its Gateway before changing state.
- Restore the archive into a new staging directory with
openclaw backup restore. - Check custom
agentDirpaths and copy any separately protected workspace or external files. - Move the verified state into the destination location or set
OPENCLAW_STATE_DIRto the restored directory. - Run
openclaw doctor, correct permissions and paths, then restart the Gateway. - Exercise a representative workflow: load an agent, confirm its session history, send a test through each required channel, and verify provider authentication.
Keep the source host or its archive untouched until the destination has passed these tests. That gives you a rollback path if a custom path, permission, or external workspace was omitted.
Verify backups instead of trusting their filenames
- Use
--dry-run --jsonto inspect a planned full archive. - Use
--verifywhen creating full or configuration-only archives. - Perform a test restore into a clean directory on a schedule, not only after a failure.
- Check that the archive contains the expected workspaces, agent roots, and database snapshots.
- For database methods, retain the verification output for schema, foreign-key, integrity, and hash checks.
- Measure whether the restore target has enough space before an emergency migration.
- After recovery, run
openclaw doctorand a real end-to-end agent test.
A successful archive command proves that OpenClaw could create a file; a successful clean restore proves that the file can actually recover the installation.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Security, storage, and retention
Encrypt and restrict every copy
Archives may contain authentication profiles, provider and channel credentials, session history, and other sensitive records. Encrypt local archives where practical, restrict filesystem permissions, and use private, access-controlled offsite storage. An object store that encrypts uploaded bytes does not remove the need to protect the plaintext local staging file.
Rank #4
- Easily store and access 4TB of content on the go with the Seagate Portable Drive, a USB external hard drive.Specific uses: Personal
- Designed to work with Windows or Mac computers, this external hard drive makes backup a snap just drag and drop
- To get set up, connect the portable hard drive to a computer for automatic recognition no software required
- This USB drive provides plug and play simplicity with the included 18 inch USB 3.0 cable
- The available storage capacity may vary.
Use separation for disaster recovery
Keep at least one copy independent of the host. An external SSD used as an archive destination can be disconnected or stored offline between runs, limiting exposure to a host compromise or ransomware event. An offsite destination protects against loss of the machine or its premises. These copies should not share the same administrator credentials if you are trying to contain a single-account compromise.
Rotate credentials after exposure
If an archive is lost, publicly exposed, or accessed by an unauthorized person, assume its credentials and session material may be compromised. Revoke or rotate provider, channel, and authentication credentials, invalidate sessions where supported, replace the affected archive, and review access logs. Redacted remote schedules reduce exposure but do not protect secrets that were already present in older full-fidelity copies.
Set retention around recovery needs
Keep short-interval copies for recent mistakes and longer weekly or monthly generations for delayed discovery of corruption. The example retention policy of seven daily, four weekly, and twelve monthly copies is a starting point, not a universal requirement. Adjust it to the amount of state you can recreate, available storage, and the longest period in which you may need to discover an error.
Troubleshoot common backup and restore problems
“Target is not empty” during restore
OpenClaw intentionally refuses a non-empty restore target to prevent files from two states being mixed. Create a new empty directory, verify its path, and run the restore again. Do not work around the check by deleting files from a live state directory.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Scan for outdated or missing drivers - takes under a minute3Repair Windows errors before they cause bigger problemsThe restored agent cannot find its files
Check the agent’s configured agentDir. Custom roots may be outside the state archive or may have been restored under a different path. Restore those files separately or update the configuration before running openclaw doctor.
The configuration repair did not restore included settings
A --only-config backup saves the active JSON configuration but does not include $include dependencies. Locate and protect those dependency files separately, then create a full archive after the configuration is repaired.
A remote backup is missing credentials
Managed pushed schedules redact credential-bearing tables and secret-prefixed machine-state rows by default. This is expected. Use a trusted private destination and --include-secrets only when a full-fidelity remote copy is genuinely required and its access controls are strong enough for credentials.
A database copy opens with errors
Discard the manually copied live files and create a new SQLite snapshot or full archive while OpenClaw is managed by its backup commands. The -wal, -shm, and -journal companions are part of why direct copying is unsafe.
Recommended Free Tools
The archive exists but the Gateway will not start
Keep the restored staging directory intact, run openclaw doctor, and address its path, permission, configuration, or database findings. Confirm that OPENCLAW_STATE_DIR points to the intended restored state and that external agent roots and workspace files are readable by the service account.
A practical OpenClaw recovery plan
For most installations, use a verified full archive before every upgrade, reset, or machine move; retain generations on an external or offsite destination; and test a clean restore periodically. Add SQLite snapshots or Git history when database recovery needs to be more granular than full archives. Use Litestream or sqlite3_rsync only when the operational overhead is justified by a smaller recovery-point objective.
The decisive safety rules are simple: never copy live SQLite files, never restore over a directory containing an active installation, verify before switching state directories, and run openclaw doctor before restarting the Gateway. Protect every archive as if it contained the credentials and conversations it can contain, because it does.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
