Back to Blog
openclaw clawhub self-hosted agent-workflows mcp

OpenClaw’s /btw + ClawHub Workflow: A Practical Session Hygiene Playbook

OpenClaw’s March 2026 cycle introduced two changes that are easy to miss but high-impact in day-to-day operations: /btw side questions and ClawHub-first install flows.

If you run OpenClaw heavily (especially with long threads and lots of tools), these two together solve a real pain point: context sprawl + plugin sprawl.


What changed (and why it matters)

1) /btw gives you tool-less side answers without polluting main context

In v2026.3.22, OpenClaw added /btw for quick side questions that do not alter your future session context.

Operationally, this means you can ask one-off questions (syntax checks, reminders, tiny clarifications) without contaminating a long-running debugging/research thread.

2) ClawHub became first-class in install workflows

Same release: plugin install behavior now prefers ClawHub before npm for bare npm-safe package names, and OpenClaw added native skills search|install|update flows plus ClawHub-backed plugin metadata for updates.

Operationally, this gives you a cleaner upgrade path and better provenance than ad-hoc copy/paste installs from random repos.


The practical playbook (15 minutes to implement)

Step A — Keep working threads clean with /btw

Use /btw for:

  • one-off factual checks
  • “quick what’s-this-command” questions
  • transient questions unrelated to the current task

Stay in the main thread for:

  • decisions that should affect future outputs
  • constraints and project requirements
  • anything you want retained as durable context

Step B — Standardize skill/plugin intake

Use:

openclaw skills search "calendar"
openclaw skills install <skill-slug>
openclaw skills update --all
openclaw plugins install clawhub:<package>

Then define a simple team rule:

  • No direct install from unknown repos without review
  • Prefer ClawHub-native install paths
  • Track updates on a schedule (skills update --all / plugins update --all)

Step C — Add a lightweight approval gate

Before enabling a new capability in production, verify:

  1. Required permissions/tools are expected
  2. Skill behavior matches your real use case
  3. Rollback path is known (plugins uninstall ... / disable)

This keeps your agent flexible without turning it into a drifting bundle of unvetted extensions.


Why this matters now

The ecosystem is rapidly expanding, and the operational bottleneck is no longer “can the agent do X?” — it’s whether you can keep sessions and capabilities maintainable over time.

/btw helps control context entropy. ClawHub-first workflows help control extension entropy. Together, they’re a concrete reliability upgrade for self-hosted OpenClaw users.


Sources

  • OpenClaw release notes (v2026.3.22): /btw side questions; ClawHub-first install behavior; native install/update workflows
  • OpenClaw ClawHub docs: native commands, install/update behavior, and workspace integration
  • Independent ecosystem write-up (March 2026) confirming practical /btw and ClawHub usage patterns in real deployments

Protect your AI agent with Clawly

Deploy your OpenClaw agent in an isolated, hardened container with encrypted credentials and managed updates. No DevOps required.

Deploy Your Agent