OC Full Ops Audit Recipe

End-to-end OpenClaw audit and remediation recipe for gateway, channels, nodes, security, and memory sync.

This skill is a coherent, high-level audit recipe and appears benign, but it is intentionally vague about two important things: making changes and where results are stored. Before installing or running it, confirm: (1) where 'daily memory' and 'shared memory' live, who can read them, how long entries are retained, and whether outputs will be redacted to avoid leaking secrets; (2) that the agent will require and request explicit, itemized approval before performing any disruptive fixes, and that any credentials needed for remediation are managed and consented to separately; (3) that rollback points and verification steps are implemented so changes can be safely reverted. If you want a safer baseline, restrict this skill to read-only checks until remediation procedures, credential handling, and memory/retention policies are explicitly documented.

Type: OpenClaw Skill Name: oc-full-ops-audit-recipe Version: 0.1.0 The skill bundle is classified as suspicious due to the broad and high-privilege instructions given to the AI agent in `SKILL.md`. While lacking explicit malicious intent, the instructions to 'Apply fixes' and perform 'remediation' across 'gateway, channels, nodes, security, and memory sync' grant the agent significant modification capabilities. This represents a high-risk operation, as an agent executing such commands could potentially cause disruptive changes or unintended side effects if not perfectly constrained, even with the stated safeguards like 'explicit approval'.