automatic-stateful-prompt-improver

[Skill Scanner] Installation of third-party script detected The skill’s stated purpose, capabilities, data flow, and execution model are coherent and proportionate. It describes a legitimate optimization loop using a centralized MCP server with explicit triggers and a feedback mechanism. No credential requirements, anomalous data flows, or misleading claims are evident in the fragment. Overall, the footprint aligns with the described goal of automated prompt optimization rather than any harmful activity. LLM verification: Functionally, the skill matches its purpose (automatic prompt optimization with a learning loop). However, the specification lacks essential security controls and provenance around the external optimization service and persistent history. That design gap creates a realistic risk of sensitive data exfiltration or improper retention if a malicious or compromised MCP endpoint is used. Before deployment, require endpoint vetting, authentication, redaction/PII controls, explicit consent/opt-in, and d