config-hardener

The skill presents a coherent security-oriented configuration audit workflow for OpenClaw with a strong emphasis on least privilege, sandboxing, and auditable permission usage. The most critical risk is the potential absence of AGENTS.md, which the policy marks as a highest-priority finding. Other risks are moderate and addressable through the hardened templates and explicit per-skill permissions. Overall, the footprint is proportionate to the stated purpose, with benign network behavior and no evident unverifiable binaries in the supplied content. The primary actionable step is to generate and review AGENTS.md and the hardened settings, then apply changes via the reviewer-approved workflow.