OpenClaw v2026.10.5: Refreshed Sidebar, Skill Workshop 2.0, and Public Conversation Links

OpenClaw v2026.10.5 is a large release: 2,250 pull requests from 637 contributors. The headline changes are a refreshed sidebar, Skill Workshop 2.0, and public conversation links. Full details are in the release notes, but here are the parts that affect day-to-day use.

Sidebar

The sidebar now handles larger agent lists better. You get search and pinning so frequently used agents stay within reach. Two new panels — Subagents and Processes — let you check delegated work, inspect recent command output, or stop a task beside the parent conversation without leaving the chat.

Skill Workshop 2.0

Skill Workshop 2.0 consolidates learned skills, their files, editing, and version comparisons into a single library.

• Inspect a change and use Undo to restore an earlier version.
• Up to ten saved versions per skill.
• Archive and restore controls for skills you no longer use.

One caveat from the release notes: background learning sends conversation and tool content to your configured model provider and can add usage charges. Set it to Off to stop background learning and unused-skill cleanup. The /learn command still lets you request learning manually.

Public conversation links

Anyone with a published chat's normal address can now read the transcript without signing in. Signed-in readers use the same link under their existing permissions.

Two things to keep in mind before you publish:

• Review the conversation first. The link exposes existing and future user messages and final replies to anyone who has it.
• Redaction is best effort, and turning sharing off cannot recall saved copies.

If your deployment sits behind a login proxy, it needs the public-reader route setup for these links to work.

Who it's for

Teams running multiple agents who need to audit delegated work, and anyone sharing OpenClaw conversations outside their org — the publishing caveats matter most there.