OpenClaw 2026.5.7 is an operator-safety maintenance release
OpenClaw 2026.5.7 is not a flashy feature drop; it is a broad maintenance release aimed at the paths that decide whether a self-hosted agent stays governable. It tightens native command owner enforcement, requires admin scope for global Active Memory toggles, routes inline skill dispatch through before-tool-call authorization, fixes misleading empty delivery success, repairs cron model overrides and last-channel failures, improves Telegram access-group and polling watchdog behavior, and makes Codex approval prompts less noisy while keeping plugin approval choices accurate.
The release folds several earlier risk reports into a tagged build and moves OpenClaw in the right direction on permission boundaries, recovery visibility, and channel reliability. For operators, that matters more than a new demo feature.
- Official GitHub release v2026.5.7 published 2026-05-07 20:57 UTC
- Release notes include owner enforcement for native commands via PR #78864 and admin scope for global Active Memory toggles via PR #78863
- Inline skill tool dispatch now goes through before-tool-call authorization hooks via PR #78517
- Delivery now reports deliverySucceeded=false when no adapter result is returned, fixing issue #78532
- Cron and Gateway fixes cover computed cron JSON status, bad persisted cron model overrides, last-channel delivery failures before model execution, daily session transcript rollover, stale CLI tasks, and hot-reload deferrals
- Telegram fixes cover accessGroup sender allowlists, polling watchdog liveness, same-chat message tool delivery, and model callback parsing for provider ids with dots
- Codex approval handling, model-provider compatibility, ClawHub publishing recovery, plugin install shell behavior, Discord/WhatsApp/voice paths, and context-cache invalidation are all included in the same tagged release
- The release is wide, so regression testing should follow the channels and runtimes you actually use rather than only checking startup
- Some May 8 OpenClaw issues and PRs are already post-release, including managed image media auth, pairing-scope preservation, malformed node catalog handling, and cross-tool loop circuit breaking
- Teams exposed to untrusted users should still audit Gateway/network exposure; a tagged maintenance release is not a substitute for least-privilege deployment