Updating
WednesdayAI is moving fast before 1.0. Treat updates like shipping infrastructure: update, run checks, restart, then verify. Theopenclaw
command remains a compatibility alias for the wednesdayai binary.
Recommended: re-run the website installer (upgrade in place)
The preferred update path is to re-run the installer from the website. It detects existing installs, upgrades in place, and runswednesdayai doctor when
needed.
-
Add
--no-onboardif you don’t want the onboarding wizard to run again. -
For source installs, use:
The installer will
git pull --rebaseonly if the repo is clean. -
For global installs, the script installs the
wednesdayainpm package from the stablelatestdist-tag under the hood.openclawremains a binary alias, not a separate package rail. -
Legacy note:
clawdbotremains available as a compatibility shim.
Before you update
- Know how you installed: global (npm/pnpm) vs from source (git clone).
- Know how your Gateway is running: foreground terminal vs supervised service (launchd/systemd).
- Snapshot your tailoring:
- Config:
~/.openclaw/openclaw.json - Credentials:
~/.openclaw/credentials/ - Workspace:
~/.openclaw/workspace
- Config:
Update (global install)
Global install (pick one):wednesdayai. The openclaw command remains available as a compatibility alias after install. Release channels map to npm dist-tags for stable package installs: stable uses latest, and beta uses beta. The dev channel switches to a source checkout so it can rebase and reinstall from main.
To switch update channels (git + npm installs):
--tag <dist-tag|version> for a one-off install tag/version.
See Release channels for channel semantics,
promotion checks, and support expectations. Chat users who only need to know
what an update should look like can start at
How WednesdayAI updates reach you.
Note: on npm installs, the gateway logs an update hint on startup (checks the current channel tag). Disable via update.checkOnStart: false.
Core auto-updater (optional)
Auto-updater is off by default and is a core Gateway feature (not a plugin).stable: when a new version is seen, WednesdayAI waitsstableDelayHoursand then applies a deterministic per-install jitter instableJitterHours(spread rollout).beta: checks onbetaCheckIntervalHourscadence (default: hourly) and applies when an update is available.dev: no automatic apply; use manualwednesdayai update.
wednesdayai update --dry-run to preview update actions before enabling automation.
Then:
- If your Gateway runs as a service,
wednesdayai gateway restartis preferred over killing PIDs. - If you’re pinned to a specific version, see “Rollback / pinning” below.
Update (wednesdayai update)
For source installs (git checkout), prefer:
- Requires a clean worktree.
- Switches to the selected channel (tag or branch).
- Fetches + rebases against the configured upstream (dev channel).
- Installs deps, builds, builds the Control UI, and runs
wednesdayai doctor. - Restarts the gateway by default (use
--no-restartto skip).
wednesdayai update will try to update via your package manager. If it can’t detect the install, use “Update (global install)” instead.
Update (Control UI / RPC)
The Control UI has an Update button in Settings (RPC:update.run). It:
- Runs the same source-update flow as
wednesdayai update(git checkout only). - Writes a restart sentinel with a structured report (stdout/stderr tail).
- Restarts the gateway and pings the last active session with the report.
Update (from source)
From the repo checkout: Preferred:pnpm buildmatters when you run the packagedwednesdayaibinary or use Node to rundist/.- If you run from a repo checkout without a global install, use
pnpm openclaw ...orpnpm wednesdayai ...for CLI commands. - If you run directly from TypeScript (
pnpm openclaw ...orpnpm wednesdayai ...), a rebuild is usually unnecessary, but config migrations still apply → run doctor. - Switching between global and git installs is easy: install the other flavor, then run
wednesdayai doctorso the gateway service entrypoint is rewritten to the current install.
Always run: wednesdayai doctor
Doctor is the “safe update” command. It’s intentionally boring: repair + migrate + warn.
Note: if you’re on a source install (git checkout), wednesdayai doctor will offer to run wednesdayai update first.
Typical things it does:
- Migrate deprecated config keys / legacy config file locations.
- Audit DM policies and warn on risky “open” settings.
- Check Gateway health and can offer to restart.
- Detect and migrate older gateway services (launchd/systemd; legacy schtasks) to current WednesdayAI services.
- On Linux, ensure systemd user lingering (so the Gateway survives logout).
One-time session actor backfill
On the first gateway start after upgrading to this build, the session store backfills a stored actor (channel, account, peer kind/id, canonical identity, DM scope) onto existing session rows from identity columns the gateway already persisted at write time. There is no operator step and no CLI command: the backfill runs at storage open (Postgres: inside the startup initialization lease; SQLite: at worker init), pages large stores in batches of 500 rows, skips tombstoned rows, never overwrites an existing actor, and records completion as theactor-meta-v1 marker row in
ctx_schema_backfills — later starts skip it entirely.
Rows that cannot yield a complete actor (invalid DM scope, or a group row with no group peer)
stay actorless, as do internal buckets that have no external counterpart (cron, subagent, ACP
sessions). An actorless session gets no identity guessed from its session key: no group tool
policy, no per-DM history limit, no session.sendPolicy rule that requires channel or
chatType (prefix rules on the key still match). wednesdayai sessions classifies the row’s
kind from the stored actor (then chatType); the Control UI labels the row from the actor,
falling back to the raw key when the actor has no channel. The next inbound message in that
conversation re-establishes identity from the live message and the row is written normally
again.
Source-install operators can verify the result against the live database:
verify-context-schema-consumers: PASS. It fails unless the
actor-meta-v1 marker exists, no eligible row still lacks an actor, and the sessionKey
identity-parse guard prints check-no-sessionkey-identity-parse: PASS
(pnpm lint:sessions:no-sessionkey-identity-parse runs the same guard in pnpm check).
Start / stop / restart the Gateway
CLI (works regardless of OS):- macOS launchd (app-bundled LaunchAgent):
launchctl kickstart -k gui/$UID/ai.wednesdayai.gateway(useai.wednesdayai.<profile>; legacyai.openclaw.<profile>,com.openclaw.*still work) - Linux systemd user service:
systemctl --user restart wednesdayai-gateway[-<profile>].service - Windows (WSL2):
systemctl --user restart wednesdayai-gateway[-<profile>].servicelaunchctl/systemctlonly work if the service is installed; otherwise runwednesdayai gateway install.
Rollback / pinning (when something breaks)
Pin (global install)
Install a known-good version (replace<version> with the last working one):
npm view wednesdayai version.
Then restart + re-run doctor:
Pin (source) by date
Pick a commit from a date (example: “state of main as of 2026-01-01”):If you’re stuck
- Run
wednesdayai doctoragain and read the output carefully (it often tells you the fix). - Check: Troubleshooting
- Ask in Discord: https://discord.gg/clawd