OpenClaw v2026.5.28-beta.1: Agent and Codex runtime recovery is steadier, operator field notes

The source record for v2026.5.28-beta.1 names a provider, model, voice, or context change. This is an operator's reading of that record, with the exact source facts kept next to a bounded experiment rather than a generic feature list.

Pre-release status matters here: stage the tag beside the stable package and do not infer support for every account or platform. A release tag describes the project change. Your Gateway's provider, channel, platform, and account support should be recorded separately.

A small field test

01 / Agent and Codex runtime recovery is steadier

Source signal. subagents keep cwd/workspace separation, hook context stays prompt-local, session locks release on timeout abort, stale restart continuations are avoided, and Codex app-server/helper failures no longer tear down shared runtime state.

The safe reading of Agent and Codex runtime recovery is steadier is operational: prove the model can answer the intended task, then force a bounded failure and verify that no side effect is repeated.

02 / Channel delivery and session identity got safer

Source signal. Channel delivery and session identity got safer across outbound plugin hooks, Matrix room ids, iMessage reactions/approvals, Slack final replies, Discord recovered tool warnings, and Microsoft Teams service URL trust checks.

Read Channel delivery and session identity got safer as an authorization checkpoint: prove who may act, what the action can reach, and how the rejected attempt is surfaced to the operator.

03 / Mobile and chat surfaces got a broader refresh

Source signal. the iOS Pro UI, Gateway chat transport, onboarding, Talk permissions, WebChat reconnect delivery, and session picker behavior now preserve more state across reconnects and empty searches.

Before widening access for Mobile and chat surfaces got a broader refresh, replay a narrow allow and deny case. A readable denial and a revoked test credential are part of the acceptance record.

04 / CLI, auth, doctor, and provider paths fail faster and recover more

Source signal. malformed numeric/version options are rejected, OAuth and local service startup requests are bounded, legacy api_key auth profiles migrate to canonical form, and restart guidance is actionable.

Read CLI, auth, doctor, and provider paths fail faster and recover as an authorization checkpoint: prove who may act, what the action can reach, and how the rejected attempt is surfaced to the operator.

The details that affect a runbook

Plugin and Gateway hot paths do less repeated work while

Plugin and Gateway hot paths do less repeated work while preserving cache correctness for install records, config JSON parsing, tool search catalogs, session stores, manifest model rows, auto-enabled plugin config, browser tokens, and viewer assets.

The safe reading of Plugin and Gateway hot paths do less repeated work while is operational: prove the model can answer the intended task, then force a bounded failure and verify that no side effect is repeated.

Status

show active subagent details in status output.

Keep Status measurable: name the input, expected output, failure signal, and operator who approved the test before calling the change ready.

Diffs

split the default language pack and expand default Diffs language coverage while keeping the host floor aligned.

Use Diffs for a small reproducible task: save the input, visible result, useful failure, and version context another operator would need.

Operator checklist before rollout

1. Before touching persistent state, capture the version, host, provider, and workspace for Agent and Codex runtime recovery is steadier.
2. Read the linked release record and verify a backup whenever the change can alter configuration or session data.
3. Interrupt one delivery after admission, reconnect the transport, and compare the final reply with the original conversation.
4. Force a short transport interruption and verify that retry or recovery preserves the original thread and sender.
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

Agents. fall back to local config pruning when the optional agents delete Gateway probe cannot authenticate, so offline installs can still delete agents without removing shared workspaces. Treat the fix as a boundary condition and retain the source wording beside your local result.

Tighten phone-control mutation authorization [AI]. The release records a focused fix in tighten phone-control mutation authorization [ai]. Keep the reproduction and the post-fix result together in the acceptance record.

This page is a practical reading of the v2026.5.28-beta.1 source record; the release index and model provider reference remain the references to use for migrations and support boundaries.