Skill Onboarder

Auto-wires new skills into core system. On skill install detected, reads SKILL.md/AGENT.md/SOUL.md/hooks and injects into soul/memory/agent files.

This skill does what it says — it auto-wires other skills — but it gains high-impact, persistent ability to change core agent/soul configuration. Before installing: 1) Prefer not to enable always:true; require manual confirmation for onboarding actions. 2) Restrict this skill to administrators and run it in a sandbox first. 3) Audit its templates and the exact write paths it will modify; ensure the platform logs and file-change audit are enabled. 4) Require that incoming skills be validated (only trusted sources) before the onboarder processes them, because a malicious skill could craft SKILL.md/AGENT.md/hooks to create persistent triggers, hard rules, or pre-response hooks that alter behavior or exfiltrate data. 5) If you must use it, remove or disable 'Always Fires' / HARD-rule auto-enabling by default and require explicit user consent per injected rule.

Type: OpenClaw Skill Name: skill-onboarder Version: 1.1.0 The skill-onboarder is designed to automate the integration of other skills by reading their configuration files and injecting instructions directly into core system files such as 'soul/master.md' and 'agent/skills-active.md'. While the stated purpose is administrative automation, the ability to modify the agent's core logic and rules based on external, potentially untrusted skill files represents a significant security risk for prompt injection and persistence. It lacks explicit validation or sanitization of the content being injected, effectively acting as a mechanism for privilege escalation within the OpenClaw environment.