Agent DailyAgent Daily
releaseintermediate

[Release] openclaw/openclaw v2026.6.1-beta.2: openclaw 2026.6.1-beta.2

By steipetegithub
View original on github

OpenClaw v2026.6.1-beta.2 introduces significant stability improvements for agent runtimes, channel delivery, and skill management. Key enhancements include better recovery from interrupted tool calls, steadier multi-channel support (Telegram, WhatsApp, iMessage, Slack, Discord, Teams, etc.), and a new Skill Workshop UI for governed skill creation and review. The release also externalizes plugins (Tokenjuice, GitHub Copilot), improves provider coverage with new models like MiniMax M3, and optimizes hot paths to reduce repeated work while maintaining stability.

Key Points

  • Agent and CLI runtimes now recover more cleanly from interrupted tool calls, stale session bindings, and media delivery retries through improved error handling
  • Multi-channel delivery (Telegram, WhatsApp, iMessage, Slack, Discord, Teams, Google Chat/Meet) is more stable with better realtime Talk support and iOS push relay
  • New Skill Workshop feature provides governed skill creation with proposal lists, revision handoff, searchable file previews, and review states via Control UI and agent tools
  • Provider and plugin requests now have bounded timers, retries, OAuth/device-code lifetimes, and media downloads to prevent runs from hanging
  • Skills and plugin loading handle stale disabled snapshots and loader failures more clearly, preventing disabled SecretRefs from aborting channel turns
  • SQLite-backed state management for iMessage monitor state, plugin install ledgers, and inbound queues improves restart recovery with less filesystem scanning
  • Chat and Control UI startup paths keep sends alive through history loading, stream deltas incrementally, and expose calmer composer controls with local draft typing
  • Provider coverage expanded with MiniMax M3, Google/Vertex catalog fixes, OpenRouter SQLite caching, and Copilot Claude 1M capabilities
  • Workboard adds orchestration primitives and agent coordination tools for multi-agent planning and run tracking
  • Code mode introduces internal namespaces for scoped agent/global sessions and exact namespace tool dispatch with MCP API support

Found this useful? Add it to a playbook for a step-by-step implementation guide.

Workflow Diagram

Start Process
Step A
Step B
Step C
Complete
Quality

Concepts

Artifacts (3)

Skill Workshop Agent Toolworkflow
The skill_workshop agent tool enables:
- Apply, reject, and quarantine explicit proposals through guarded review flow
- Pending proposals carry approved support files under standard skill folders
- Proposals can be revised in place with versioned, dated proposal frontmatter
- Scanner, hash, and rollback safeguards protect skill integrity
- Integration with Control UI for proposal today view, revision dialog, and file preview modal
Code Mode Namespace Configurationconfig
Code mode enhancements:
- Internal namespaces for scoped agent/global sessions
- Exact namespace tool dispatch routing
- MCP API files and docs for code-mode integrations
- Enables precise agent and session isolation within code execution contexts
Plugin Externalizationtemplate
Official plugins now externalized:
1. @openclaw/tokenjuice - npm and ClawHub publish metadata
2. @openclaw/copilot - GitHub Copilot agent runtime with npm and ClawHub metadata

SecretRef provider integration manifest contract for plugin development
Shared LLM core packages extracted for provider/plugin reuse
[Release] openclaw/openclaw v2026.6.1-beta.2: openclaw 2026.6.1-beta.2 | Agent Daily