security-hardener

To vet OpenClaw skills, run an automated multi-layer pipeline covering static analysis, metadata checks, dependency audits, and runtime sandboxing to identify risks and enforce safety.

Runtime sandboxing in Docker isolates skill execution environments by enforcing strict permission boundaries and continuous monitoring to meet production security standards.

Enforce least-privilege boundaries by running an automated vetting pipeline that applies strict input sanitization, permission restrictions, and token management before onboarding new skills.

Docker sandboxing integrates with token management by isolating runtime execution while enforcing strict token boundaries and continuous monitoring to meet production security standards.

Static analysis evaluates code structure and metadata while dependency auditing identifies malicious risks in external packages, ensuring compromised skills are caught before deployment.

Input sanitization is needed when vetting skills for safe deployments, acting as a critical layer to enforce permission boundaries and prevent malicious execution in production.