Upgrade OpenClaw
Stay on the edge. Update, audit, discover, propose. Latest features, always.
Trigger
This skill activates when the user says:
- "upgrade openclaw"
- "update openclaw"
- "check for openclaw updates"
- "what's new in openclaw"
Settings
On first run, prompt user for preferred sub-agent model and save to settings.json:
{
"subagentModel": "anthropic/claude-sonnet-4-6"
}
⚠️ Privacy note: If you choose an external provider (e.g.
gpt-4o,claude-*), your local OpenClaw config and environment details will be sent to that provider during the upgrade run. Prefer a locally-hosted model or your primary trusted provider.
Procedure
1. Check Settings
Look for settings.json in this skill's directory. If subagentModel not set, ask:
"Which model for upgrade sub-agents? (e.g.,
claude-sonnet-4-6,deepseek-chat). Note: external providers will receive your config data."
Save choice to settings.json.
2. Check Prerequisites
Verify required binaries are present before proceeding:
which openclaw && openclaw --version
which curl
which clawhub && clawhub -V
Abort and instruct the user to install missing binaries if any are not found.
3. Run Update
openclaw update
4. Discover What's New
openclaw --version
curl -s https://docs.openclaw.ai/llms.txt | head -100
Check:
- Config templates:
github.com/openclaw/openclaw/tree/main/docs/reference/templates - Changelog: GitHub releases
5. Audit Current Setup
openclaw hooks list
openclaw doctor --non-interactive
Look for:
- New hooks not enabled
- Doctor recommendations not applied
- Unused config options
- Relevant new ClawHub skills
6. Present Findings
## 🔍 Post-Upgrade Report
### New Features
- [Feature]: Description
### Recommended Config Changes
| Setting | Current | Recommended | Why |
|---------|---------|-------------|-----|
### New Hooks Available
- hook-name: Description
### New Skills Worth Installing
- skill-name: Description
### Doctor Recommendations
- [Items from openclaw doctor]
---
**Apply these improvements?** (yes/no/select)
7. Apply with Approval
Never apply without explicit user approval.
Wait for the user to reply "yes", "apply", or select specific items. Do not proceed on ambiguous input.
Spawn sub-agent (using model from settings.json) to:
- Config changes via
gateway config.patch - Hook enablement via
openclaw hooks enable <hook> - Skill install via
clawhub install <skill>