OpenClaw v2026.6.11 Release Notes | Fixes for Misplaced Replies, Stuck Sends, model setup failures, and more!

We heard the feedback. v2026.6.11 focuses on the rough edges that make OpenClaw feel less dependable, with fixes for misplaced replies, stuck sends, reconnects, model setup failures, and safer admin defaults. Replies, sends, and reconnects Across Telegram, WhatsApp, Matrix, Google Chat, iMessage, Feishu, and Mattermost, replies, commands, queued messages, and attachments are less likely to be dropped, duplicated, misrouted, or attached to the wrong conversation. WebChat and the Control UI keep the active conversation visible more consistently after reconnects. The terminal UI now clears completed or rejected sends instead of leaving them looking stuck. Models and fallback recovery Model selection and setup recover more clearly when catalogs, credentials, streams, timeouts, compaction, or fallbacks go wrong. Affected OpenAI, OpenRouter, and OpenCode Go setups are less likely to leave users with a stale model choice or a stalled request. Follow-up fixes improve fast mode in affected provider and fallback paths. Automatic fast mode itself is not new in this release. Sessions, memory, and safer recovery Sessions, compaction, memory, and QMD-backed memory preserve the intended conversation and useful context more consistently through long-running work, reconnects, upgrades, and transcript repair. Tool search also recovers the right context or capability more reliably. Encrypted Matrix recovery now stops safely when required key state cannot be verified. Tool policies, approvals, and secret handling stay attached to the intended runtime state, while higher-risk actions remain disabled unless explicitly enabled. Plugins and installation Plugin management now handles more official integrations through normal external package installation and repair flows. The plugin inventory and setup checks give clearer guidance when a package is missing, incompatible, or needs to be reinstalled. Admin and deployment controls Slack router relay mode gives managed or multi-gateway deployments a supported way to centralize incoming Slack traffic while the correct gateway still handles mentions, threads, and replies. The Raft channel and Raft plugin add a local CLI wake path for External Agents, including setup and status checks. Gateway health and troubleshooting signals now line up more consistently with whether OpenClaw is ready, restarting, or unable to continue. Agent runs started through the CLI and the broader gateway recover more cleanly from disconnects, shutdowns, routing changes, and failed startup conditions. Setup, commands, and scheduled work Common CLI commands now handle configuration, paths, output, and failure cases more consistently. Shell completion, doctor, config commands, and gateway configuration provide clearer guidance when an installation or setting needs attention. Scheduled jobs and built-in tools now finish, retry, report failures, and preserve their intended inputs more consistently. The plugin SDK runtime also improves reliability for tool-backed extensions that load, return results, or run scheduled work. Full Release Notes This release includes 302 PR-backed units and 704 direct commits. Full notes: https://docs.openclaw.ai/releases/2026.6.11