OpenClaw 2026.7.2 Pre-Release: State Safety, Session Rewind, MCP Dashboards, Wear OS, and Native Local Inference
OpenClaw’s latest pre-release, v2026.7.2, is a significant reliability and capability upgrade that touches nearly every layer of the stack. Here’s what’s new.
State Safety and Recovery
OpenClaw now protects persisted data through a quarantine store that survives primary-database damage. SQLite snapshots are crash-recoverable, filesystem publications are crash-durable, schema upgrades reject data loss, and a rollback-writer snapshot recovery mechanism handles corruption scenarios that would previously have required manual intervention.
Durable Channel Delivery
Messages across Telegram, Signal, Slack, QQBot, Twitch, Synology Chat, Tlon, IRC, and Zalo User are now kept recoverable across gateway restarts and local crashes via a shared ingress drain and dead-letter recovery system.
Session Rewind and Branching
Users can now rewind or fork conversations from individual messages, switch transcript branches across web and native apps, fork upstream Codex sessions, and restore prompt images after a fork. Branch-safe queued sends and stale-pane write rejection keep multi-branch sessions clean.
Interactive MCP Apps and Dashboards
The MCP Apps system gains ticketed hostable apps with bound tools, resources, and bounded context updates. Apps can be opened from channel replies, pinned to durable dashboards, and declared directly by native plugins. The sandbox is hardened.
Questions and Approvals Everywhere
Agents can now ask structured questions with option cards across web, channels, macOS, and native apps. Approvals gain push notifications, history, fair queuing, headless resolution, Claude tool-request relay, reviewer detail, and clearer formatted prompts.
Meetings and Realtime Talk
OpenClaw can now join Teams, Zoom, and Google Meet calls via default-enabled meeting plugins, collecting durable transcripts. Realtime Talk adds OpenAI and Gemini video support plus GPT Live through Codex OAuth.
Wear OS Companion
A new Wear OS companion app pairs via phone proxy and provides home-screen agent/session/model selection, realtime Talk controls, audio-reactive playback, and an instant-talk tile.
Guided Local Inference Setup
Onboarding now detects local inference providers, guides users through in-process llama.cpp/Gemma setup, offers downloadable models from web and macOS, and includes an in-process RAM-gated llama.cpp path.
New Models
The release also adds Claude Opus 5, Kimi K3, and GPT Live through Codex OAuth across the catalog and runtime.