Security audit
Security checks for vulnerabilities and agentic risk
This skill is a normal image-editing guide that uses external image models on user-provided images, with only a minor documentation ambiguity about restoration tasks.
Before installing, be aware that images or masks provided to this skill may be sent to Runware image models for editing. Use a different restoration-focused skill for pure upscale, denoise, deblur, or old-photo repair tasks until the publisher clarifies the restore wording.
The documentation contradicts itself about whether restoration belongs in this skill, which creates ambiguity for the agent selecting tools. In security-sensitive orchestration, contradictory capability definitions can lead to unintended tool use, weakened guardrails, and unreliable enforcement of task separation.
The skill boundary is inconsistent: the manifest says restoration/quality-only work should be handled by a different skill, but the body text explicitly includes restore behavior in this skill. That can cause incorrect routing, policy bypass, or invocation of the wrong tooling path in agent pipelines that rely on the manifest for capability scoping and safety decisions.
