Openclaw Safety Coach
PassAudited by ClawScan on May 1, 2026.
Overview
This is a benign instruction-only safety guide, though it may refuse or redirect risky requests and recommends user-run OpenClaw audit/fix commands.
This appears safe to install if you want a moderation and security-coaching skill. Expect it to refuse some unsafe requests and to suggest OpenClaw security commands; review any command, especially automatic fix commands, before running it.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
When this skill is invoked, the agent may say no to requests it considers unsafe and steer the user toward safer alternatives.
The skill intentionally directs the agent to refuse or redirect certain requests. This changes response behavior, but it is disclosed and matches the safety-coach purpose.
Mission: enforce OpenClaw's 2026-era security posture, block risky actions, and coach users toward safer workflows.
Install it if you want safety-coaching behavior; disable or avoid invoking it if you need an unrestricted assistant for benign administrative work.
If the user runs the suggested fix command, OpenClaw settings may be changed automatically.
The README recommends a user-run command that may automatically modify OpenClaw security settings. It is disclosed and purpose-aligned, but it can affect the user's environment if followed.
Run `openclaw security audit --fix` to automatically address most issues
Run the read-only audit first, review the proposed changes, and back up important configuration before using any automatic fix option.