Introduce Policy plugin for workspace conformance checks
TL;DR
Added a bundled Policy plugin to perform policy-backed channel conformance checks, doctor lint findings, and opt-in workspace repair via the CLI.
## What changed On 21 May 2026 OpenClaw released a bundled Policy plugin. The plugin runs policy-backed checks on channel conformance, surfaces doctor lint findings, and applies opt-in workspace repairs directly from the CLI.
The feature ships with the core package. No separate install step is needed for existing users who manage OpenClaw through the YAML config and terminal commands.
## Why it matters Self-hosted agents often drift into broken states after updates or custom skill additions. The new plugin gives Vibe Builders an automated way to catch those issues before they surface in daily heartbeat messages or browser tasks.
It also raises the baseline reliability of the free tier. Teams that previously avoided OpenClaw because of silent configuration errors now have a concrete reason to test the CLI path again.
## How to use it Update to the current release through the standard CLI command, then run openclaw policy check to scan the workspace. Review the lint output in the terminal before deciding on fixes.
To apply repairs, use openclaw policy repair with the opt-in flag. The command is available immediately on the free self-hosted version from the GitHub repository.
## Watch for Monitor community reports over the next four weeks to see whether reported setup failures decline. The bet breaks if the plugin rejects valid custom MCP bridges or self-written skills too often. The logical next step is a similar plugin for security and token-usage guardrails.
Harsh’s take
The Policy plugin papers over OpenClaw's core weakness: a 45-minute self-hosted install that still leaves most solo operators with fragile workspaces. Automated linting helps, yet it adds another CLI surface that must be learned and monitored.
The honest trade-off is fewer silent breaks versus extra commands to run before every major change. Pure non-technical users gain little; they should stay with managed agents.
Update your instance today and run the first policy check before your next batch of scheduled tasks.
by Harsh Desai
About OpenClaw
View the full OpenClaw page →All OpenClaw updatesMore from OpenClaw
- IntegrationAdd native DM draft previews for Telegram tool progress
Telegram integration now supports allowlisted native DM draft previews for transient tool progress while keeping final answers on the normal persistent delivery path.
- FeatureAdd CLI commands for building and validating tool plugins
Introduced defineToolPlugin alongside new CLI commands to build, validate, and initialize typed simple tool plugins with generated manifest metadata.
- FeatureEnhance browser automation with modal dialog handling
The browser tool now surfaces pending and recently handled modal dialogs in snapshots, returns blockedByDialog status, and allows answering dialogs via CLI.