Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed China-focused marketing planner with an optional benchmark API script, with no evidence of hidden execution, persistence, credential access, or destructive behavior.
Before installing, be aware that benchmark lookups use an external Tencent SCF endpoint and may reveal your IP address, timing, and requested platform. Avoid putting sensitive campaign plans, customer data, or confidential copy into any future API-backed compliance or suggestion workflow unless the publisher documents data handling clearly.
The skill advertises an executable shell script and real API-backed behavior, but it does not declare any tool scope or permissions boundaries. That creates a capability/expectation mismatch where an agent or user may invoke shell access without explicit consent controls, increasing the risk of unintended command execution and outbound access.
The trigger phrases are extremely broad for generic marketing topics, making the skill likely to activate in many unrelated or only loosely related conversations. Over-broad invocation increases the chance that users are routed into a skill that can run scripts or call external APIs without their informed intent.
The skill explicitly describes real external API endpoints and a CLI script, but gives no warning that using them transmits data off-platform. In a strategy-planning context, prompts may include sensitive business plans, budgets, campaign copy, or customer information, so silent network transmission creates privacy, compliance, and data-governance risk.
The script silently makes outbound requests to a third-party remote API endpoint whenever it is run, without disclosing that behavior or obtaining user consent. In an agent skill context, this can leak usage metadata such as invocation timing, selected platform values, IP address, and environment/network context to an external service, which is more sensitive because users may assume a local planning utility rather than a networked integration.
The script formats all benchmark output in Chinese string literals, which imposes a specific language on users without any opt-in or locale selection. This is a natural-language policy concern because the file does not offer an alternative language or document why Chinese-only output is required.
No suspicious patterns detected.
