Skill Auditor

This tool analyzes a provided skillJson string representing a tool or skill definition to identify potential security risks, misconfigurations, or unsafe behaviors and produces structured findings such as an overall riskScore and threatLevel. It accepts the full skillJson as a required string payload and an optional depth parameter that controls analysis depth, where depth can be set to fast or deep and defaults to deep when omitted. The worker must rely only on these inputs and produce an output JSON object that strictly validates against the system-provided structured output schema (auto-produced elsewhere), with no extra keys, type mismatches, or formatting outside of JSON.