For a normal supported installation, run openclaw update. OpenClaw identifies whether it was installed with npm, pnpm, Bun, or from a Git checkout, checks for an update while the existing Gateway keeps serving, activates the candidate, and verifies the result. Use openclaw update --dry-run first when you want a rehearsal, and create a verified backup before a significant upgrade.
Antihost offers VPS plans with one-click OpenClaw, NVMe storage and a 14-day free trial.
Fastest safe update: the supported installation path
Open a shell as the same account that owns your OpenClaw installation, then run:
As an Amazon Associate I earn from qualifying purchases.
openclaw update
The updater detects the installation owner and uses the matching update mechanism. It is designed to keep the old Gateway serving while it checks the candidate, then activate and verify the new version. Do not change users, installation prefixes, profiles, or state directories between the check and the update; doing so can create a second installation that does not use your existing configuration.
For a rehearsal that does not activate anything, use:
#1 Best Overall
openclaw update --dry-run
A dry run is useful before a production maintenance window or when you are unsure which owner or channel the installation uses. It lets you inspect the proposed action before making a change.
Prepare the machine before updating
1. Make and verify a backup
Before a significant update, create a verified backup of the OpenClaw state and configuration. Verification matters: a backup that was never checked cannot reliably restore credentials, channels, plugins, tasks, flows, or other data. Keep the backup available until the new Gateway has passed its checks.
2. Confirm the supported Node runtime
Supported Node runtimes are Node 24.16 or newer on the 24.x line, or Node 26.1 or newer. Check the runtime used by the same account that runs OpenClaw:
node --version
If the command reports an older runtime, upgrade Node through your normal system or version-manager process before updating OpenClaw. A different shell, user, or service environment can resolve a different Node executable, so check the runtime from the account that owns the Gateway.
3. Keep the installation identity unchanged
- Use the same owning account.
- Keep the same global installation prefix.
- Keep the same OpenClaw profile.
- Keep the existing state and configuration paths.
- Do not mix a new package-manager installation with the old Gateway’s service definition unless you intentionally migrate ownership.
These precautions ensure the updater sees the installation you actually use rather than an empty profile or a second copy elsewhere on the machine.
Choose an update channel deliberately
OpenClaw supports stable, extended-stable, beta, and development channels. Select a channel explicitly when you need something other than the installation’s current default:
Recommended Free Tools
Rank #2
openclaw update --channel stable
openclaw update --channel extended-stable
openclaw update --channel beta
openclaw update --channel dev
The stable channel is the conservative choice for a Gateway that must remain predictable. Extended-stable is another slower-moving option when it is available to your installation. Beta and dev expose newer changes sooner and therefore deserve a backup, a dry run, and a maintenance window. Switching a source checkout to a package-oriented channel or back to development is documented through the same channel option; use --channel dev for the development path and --channel stable for the stable path.
Do not select a channel merely to force an update past a failed migration. Resolve the migration or configuration error first, then choose the channel you intend to run.
What openclaw update does
- It identifies whether the installation is npm-managed, pnpm-managed, Bun-managed, or a Git/source checkout.
- It checks for a candidate while the old Gateway continues serving.
- It activates the selected candidate.
- It verifies the resulting installation.
This behavior makes the command the preferred first choice for a supported installation. Manual package-manager commands are useful when you need to control lifecycle-script policy, when the CLI itself is unavailable, or when the official migration procedure specifically calls for them.
Manual commands by installation method
Use the command that matches the owner of your existing installation. Running a command from a different package manager can create a separate global copy rather than update the Gateway you already operate.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitches| Installation owner | Update command | Important condition |
|---|---|---|
| npm 12 or npm 11.16+ | npm install -g openclaw@latest --allow-scripts=openclaw |
Allows the OpenClaw lifecycle script under the newer npm policy. |
| npm 11.15 or earlier | npm install -g openclaw@latest |
Do not add the newer lifecycle-script flag. |
| pnpm | pnpm add -g --allow-build=openclaw openclaw@latest |
Allows the OpenClaw build step. |
| Bun | bun add -g --trust openclaw@latest |
Trusts the package during installation. |
| Git/source checkout | openclaw update |
Use the channel option when changing between development and stable paths. |
After a manual package-manager update, run the verification steps in this guide. Installing a package successfully is not the same as proving that the Gateway, migrations, and configuration are healthy.
Updating a very old OpenClaw installation
Installations older than June or July 2026 require an intermediate bridge release according to the current official procedure. The guide uses version 2026.9.5 to import or repair retired task, flow, plugin, channel, credential, and configuration data before installing the latest release. Because the guide describes the cutoff as “June or July 2026,” treat an installation from that period as old if you cannot establish that it already completed the newer migration path.
Rank #3
Use the bridge sequence for an npm-managed Gateway
Stop the Gateway first, then run the commands in this order:
openclaw gateway stop
npm install -g [email protected] --allow-scripts=openclaw
openclaw --version
openclaw doctor --fix
npm install -g openclaw@latest --allow-scripts=openclaw
openclaw doctor --fix
openclaw gateway start
openclaw gateway status --deep
Confirm each stage before continuing
- Stop the Gateway. This prevents the old process from writing state while the bridge changes its data.
- Install exactly
2026.9.5. Do not skip this bridge when the installation falls into the old-version category. - Confirm the bridge version. Run
openclaw --versionand make sure the command reports2026.9.5before running Doctor. - Run the first migration.
openclaw doctor --fiximports or repairs data that newer releases no longer handle directly. - Resolve failed imports before proceeding. If Doctor reports conflicts, preserve the original state and configuration and fix the reported issue rather than deleting data or starting with a blank profile.
- Install the latest release. Use the npm command shown above, including the lifecycle-script flag for the npm versions covered by that sequence.
- Run Doctor again. The second pass applies any fixes required by the current release.
- Start and inspect the Gateway. Start it through OpenClaw, then use the deep status check to verify readiness.
The bridge is a migration step, not a permanently preferred version. Once the latest release has been installed and verified, keep the current version and backup until your tasks, channels, plugins, credentials, and configuration behave as expected.
macOS app-owned installations
If the installation owner is macos-app, update OpenClaw.app through its app updater. The app owns the bundled Gateway and runtime, so the app updater moves those components together. A CLI update request in this arrangement returns an owner hint instead of changing the package.
Do not work around that hint by installing a second global npm, pnpm, or Bun copy unless you are intentionally moving away from the app-owned setup. A second copy can use a different runtime, profile, state path, and Gateway service, making it appear that the update did not work.
Docker, Podman, and Kubernetes deployments
Container deployments are updated by replacing the image through the deployment tool that owns the workload. Pull or select the intended image in that tool, roll out the replacement, and then follow the image entrypoint’s Doctor and readiness behavior.
The host’s global package manager is not the owner of a containerized Gateway. Updating npm on the host therefore does not update the running container. Preserve the mounted state and configuration paths during the replacement, and do not treat a newly created empty container as a successful migration.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Rank #4
Verify the update and inspect the next action
Check recorded update status
When you need to know what target was recorded or what the updater expects next, inspect its machine-readable status:
openclaw update status --json
Use this before launching another update attempt. It can distinguish a completed action from one that stopped with a pending next step, reducing the risk of repeatedly running an incomplete migration.
Run Doctor for migration or configuration errors
If the update reports migration or configuration problems, run:
openclaw doctor --fix
Read the changes Doctor proposes and keep your verified backup. On an old installation, preserve the original state and configuration whenever an import conflict appears; the bridge procedure depends on that data being available for repair.
Check the Gateway deeply
For a Gateway managed through OpenClaw, use:
openclaw gateway status --deep
A deep status check is the final check in the documented old-installation sequence and is useful after a normal update as well. It verifies more than whether a process exists, helping reveal readiness or configuration problems.
Respect the actual service owner
If a supervisor, operating-system service, container orchestrator, or another process owns the Gateway, start or restart it through that owner. Do not launch an untracked second Gateway from an interactive shell. After the service owner reports the workload is running, use the deepest OpenClaw status check supported by that deployment.
Best Value
- 110 Sheets Log Book: This vehicle maintenance log book includes 110 record sheets to track repairs, services, fuel costs, and other vehicle-related expenses. It also comes with 36 convenient “Service Due” stickers to help remind you of upcoming maintenance schedules. Everything you need to keep your car records organized in one place.
- Detailed Information: Each page is carefully structured with dedicated sections for date, mileage, service description, parts replaced, vehicle information, cost, and additional notes. The clear and organized format allows you to maintain accurate records, monitor repair history, and better manage long-term vehicle performance and expenses.
- Enough Writing Space: The record sheets provide ample space for clear and thorough entries. Whether you’re documenting routine oil changes or major repairs, there’s plenty of room to write detailed information without feeling cramped.
- Easy to Use: The spiral binding allows the notebook to lay flat for comfortable writing, making entries quick and convenient. The included 36 “Service Due” stickers can be placed inside the book or on visible surfaces like your windshield or dashboard, helping you keep important maintenance reminders in sight.
- Proper Size: This compact logbook measures 5.5” x 7.1”, which fits easily in your glove compartment, center console, backpack, or desk drawer. Its portable size ensures you can access and update your vehicle records anytime, whether at home, in the garage, or at the service center.
Troubleshooting by symptom
The CLI says the installation is app-owned
Use the updater in OpenClaw.app. The message is an ownership instruction, not a package-download failure. A global package-manager command would target a different installation.
The update fails during a migration
Stop repeating the update. Run openclaw doctor --fix, read the reported migration or configuration issue, and preserve the original state and configuration. If the installation is old enough to require the bridge, return to the 2026.9.5 sequence and confirm the bridge version before Doctor runs.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
The package installs but the Gateway is still old
Check that the package manager, account, prefix, profile, and state path match the running Gateway. Then inspect openclaw update status --json and restart through the actual service owner. A successful package command can update an unused global copy while the service continues running an older one.
Lifecycle or build permission errors appear
Use the flag required by your package manager and version: npm 12 or npm 11.16+ needs --allow-scripts=openclaw; npm 11.15 or earlier uses no lifecycle-script flag; pnpm needs --allow-build=openclaw; Bun needs --trust. Do not combine flags from different managers.
The update command proposes an unexpected channel
Run a dry run, then select the intended channel explicitly with --channel stable, --channel extended-stable, --channel beta, or --channel dev. Choose one channel rather than repeatedly retrying an unspecified update.
The Gateway will not start after the update
Run Doctor, inspect the recorded update status, and start the Gateway through its service owner. Check openclaw gateway status --deep where supported. Keep the backup intact while you resolve configuration or migration errors; do not erase the existing profile as a first response.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallCrashes, 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 practical maintenance checklist
- Confirm the Node runtime is Node 24.16+ on 24.x or Node 26.1+.
- Create and verify a backup before a significant update.
- Identify the owner: npm, pnpm, Bun, Git, macOS app, or container deployment.
- Keep the same account, prefix, profile, state path, and configuration path.
- Run
openclaw update --dry-runwhen you need a rehearsal. - Select the channel explicitly if stable, extended-stable, beta, or dev matters.
- Use the package-manager-specific lifecycle or build permission only when updating manually.
- Use the
2026.9.5bridge and two Doctor passes for installations covered by the old-version procedure. - Inspect
openclaw update status --jsonbefore another attempt after an interruption. - Run
openclaw doctor --fixfor migration or configuration issues. - Restart through the real service owner and verify with
openclaw gateway status --deepwhere supported.
Which method should you use?
| Situation | Preferred method | Reason |
|---|---|---|
| Current supported npm, pnpm, Bun, or Git installation | openclaw update |
Detects the owner, checks the candidate while the Gateway serves, activates, and verifies. |
| You want to preview the operation | openclaw update --dry-run |
Rehearses the update without activation. |
| You need package-manager control | Use the matching npm, pnpm, or Bun command | Applies the correct lifecycle-script or build-trust policy. |
| Installation older than the documented June/July 2026 cutoff | Bridge with 2026.9.5, run Doctor, then install latest |
Imports or repairs retired data before the current release. |
macos-app owner |
OpenClaw.app updater | Keeps the bundled Gateway and runtime together. |
| Docker, Podman, or Kubernetes | Replace the image through deployment tooling | The deployment tool, not the host package manager, owns the running Gateway. |
Bottom line for a safe OpenClaw upgrade
Use openclaw update for a supported installation, preserve its ownership and paths, and verify the Gateway afterward. Add a dry run and verified backup for important systems. If the installation predates the documented migration boundary, use the 2026.9.5 bridge and Doctor sequence instead of jumping directly to @latest. App-owned and containerized deployments must be updated through their actual owners.
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.




