OpenClaw v2026.9.1: Updates that leave you working, delivery notes

v2026.9.1 is a channel and session story. The official source describes a skills, plugin, marketplace, or SDK change, and the important question is what survives the handoff between a Gateway, a conversation, and a reconnect.

Treat this as a stable maintenance checkpoint and keep the release record beside the backup and rollback evidence. A release tag describes the project change. Your Gateway's provider, channel, platform, and account support should be recorded separately.

Follow the message path

Diagrams in every chat

What changed. Mermaid blocks now render as diagrams in the Control UI and in the native macOS, iOS, and Android apps, with enlarge previews and a retry when a diagram fails to render on mobile.

Check it with. A successful click does not establish Diagrams in every chat. Compare the state before and after a reload, then verify that a user can tell what happened without reading internal logs.

From install to chat in one prompt

What changed. fresh installs (including npx openclaw@latest) get a quick-start lane that detects existing Claude Code or Codex logins and API keys, verifies them live, and opens the web dashboard from a foreground Gateway; the full wizard stays available as Custom setup.

Check it with. Check From install to chat in one prompt with a known prompt and a disposable account. Save the selected route, reasoning setting, latency, and refusal or fallback result so another operator can reproduce it.

Personal skill libraries on shared Gateways

What changed. keep your own skills beside the workspace set with openclaw skills library, import them from ZIP archives, and share or publish them per identity on team Gateways.

Check it with. Keep discovery separate from permission for Personal skill libraries on shared Gateways. Review the manifest, source, dependencies, and removal path in a disposable workspace before enabling it.

Updates that leave you working

What changed. openclaw update now rolls back the npm candidate when the post-update Doctor fails, preserves your configuration and secret references across a failed upgrade, hands failures to a built-in triage agent, waits for plugin readiness before restarting, accepts npm 12 local archives, lets agent-launched updates finish outside the Gateway process tree, and proceeds without a Gateway service instead of refusing when no service manager exists (users on 2026.8.2 without a service manager should run openclaw update --no-restart once). #135655.

Check it with. Read Updates that leave you working as an authorization checkpoint: prove who may act, what the action can reach, and how the rejected attempt is surfaced to the operator.

The less visible details

A Gateway that stays up

startup recovers under load and with large agent rosters, malformed legacy cron rows are quarantined instead of blocking boot, migration warnings degrade the Gateway instead of refusing to start, local model servers become the preferred OOM victims, and Windows Gateways stay online after an agent restart. #134458, #135150, #136275, #120134, #134851.

Check A Gateway that stays up with a known prompt and a disposable account. Save the selected route, reasoning setting, latency, and refusal or fallback result so another operator can reproduce it.

Personal GitHub accounts

connect "My GitHub" beside the system account in your profile, publish pull requests under an explicit personal identity, and switch sessions without re-verifying credentials every time.

Read Personal GitHub accounts as an authorization checkpoint: prove who may act, what the action can reach, and how the rejected attempt is surfaced to the operator.

Model setup that shows how you are signed

Model Setup lists account versus API-key access with the runtime-reported email for Codex and Claude candidates, the catalog "+" opens the native Codex or Claude CLI in a terminal, and the provider defaults page autosaves with a simpler model picker footer.

The safe reading of Model setup that shows how you are signed is operational: prove the model can answer the intended task, then force a bounded failure and verify that no side effect is repeated.

Operator checklist before rollout

1. Before touching persistent state, capture the version, host, provider, and workspace for Diagrams in every chat.
2. Read the linked release record and verify a backup whenever the change can alter configuration or session data.
3. Exercise the named capability with a non-sensitive prompt, including the fallback or refusal result if it occurs.
4. Force a timeout or provider fallback in a disposable session and verify that no side effect is repeated.
5. Close the test with the evidence attached, and do not widen the rollout while a behavior remains unexplained.

openclaw --version
openclaw gateway status
openclaw security audit

Keep the source trail attached

Close the loop with an explicit next action: keep the candidate isolated, proceed with the staged upgrade, or stop and investigate. Do not treat a green install as proof that the runtime path is correct.

Regression cases from the source

Ingress and token safety. bound concurrent pre-auth reads on SMS webhooks, stop delivering commands to a Watch node after its device is revoked, require the per-process nonce on the Copilot Azure BYOK proxy, isolate webhook rate limits by client, let authorized scoped node tokens be managed by their owner, and reject oversized A2A JSON-RPC batches and responses. Treat the fix as a boundary condition and retain the source wording beside your local result.

Secrets stay redacted. config.get no longer returns unredacted pre-migration snapshots, blank sensitive fields stay editable without being treated as secrets, systemd unit backups no longer leak Gateway tokens, and iOS omits deep-link URLs from logs. #131781. Keep the reproduction and the post-fix result together in the acceptance record.

This page is a practical reading of the v2026.9.1 source record; the release index and skills documentation remain the references to use for migrations and support boundaries.