OpenClaw v2026.5.27-beta.1: Recovery Overhaul, Workboard, and Plugin Externals
OpenClaw shipped a significant pre-release on May 31 (v2026.5.27-beta.1), focused squarely on robustness under failure — with cleaner recovery paths, bounded timeouts, and new plugin architecture surfaces.
Recovery & Stability
The headline theme is recoverability. Agents and CLI-backed runtimes now handle interrupted tool calls, stale session bindings, compaction handoffs, and media delivery retries more gracefully across multiple PRs (#88129, #88136, #88141, #88162, #88182).
Channel stability also improved across the board — Telegram, WhatsApp, iMessage, Slack, Discord, Microsoft Teams, Google Chat, Google Meet, and iOS realtime Talk all see steadier delivery.
Bounded Timers
Provider and plugin requests now bound more timers, retries, OAuth/device-code lifetimes, media downloads, local service probes, and generated-content polling paths before they can hang a run. This has been a persistent pain point in long-running agent sessions.
New Plugin Architecture
Two notable integrations moved to external plugins:
- @openclaw/tokenjuice — official token management plugin, now published to npm and ClawHub
- @openclaw/copilot — the GitHub Copilot agent runtime, similarly externalized
Workboard & Orchestration
Workboard adds orchestration primitives and agent coordination tools for multi-agent planning and run tracking (#87469), a sign the platform is growing toward more complex multi-agent workflows.
Skills Workshop Graduates
The Skill Workshop system gets significant investment:
- Skill Workshop proposals now support approved support files under standard skill folders (with scanner, hash, and rollback safeguards)
- Pending proposals can be revised in place with versioned, dated frontmatter before approval
- The skill_research agent tool is now part of the lifecycle, letting the system apply, reject, and quarantine explicit Skill Workshop proposals
Control UI: Dreaming Tab Agent Selector
The Control UI adds a Dreaming-tab agent selector, propagating the selected agent through Dreaming status, diary, and diary actions — a quality-of-life improvement for multi-agent deployments.
Other Notables
- iOS: hosted push relay defaults, realtime Talk playback, and guarded WebSocket ping for mobile reliability
- Code mode: internal namespaces for scoped agent/global sessions with exact namespace tool dispatch
- Skills: core skills index now centralized with unified runtime loading and status filtering
- Cron: SQLite migrations made compatible with legacy run-log tables and archived job stores
- Cron: recurring jobs now retry after transient model rate limits rather than skipping to the next slot