customer-persona

[Skill Scanner] Pipe-to-shell or eval pattern detected The skill description is coherent with its stated goal of data-driven persona generation, but the installation approach (curl ... | sh) is a significant red flag for supply-chain security. Dependency on multiple external AI services is expected for this capability, but it increases data flow to third-party domains and enlarges trust surface. Overall, the footprint is suspicious rather than benign due to the remote installer pattern and broad external service calls; it should only be used with strict provenance, verified installers, and minimal data exposure. LLM verification: The SKILL.md content itself is benign and aligns with its stated purpose (persona research and avatar generation). However, its operational instructions rely on a high-risk installer pattern (curl | sh) and centralize user queries and credentials through a third-party CLI and agent backends without describing provenance, verification, or data handling policies. This creates a meaningful supply-chain and privacy risk: running the installer executes remote code and authenticating with infsh hands