Security audit
Security checks for vulnerabilities and agentic risk
This is a disclosed Bilibili marketing-analysis skill that uses Qinghu's API and may export result files, with no evidence of hidden execution, persistence, or malicious behavior.
Install only if you are comfortable giving the agent access to a Qinghu API token for Bilibili data lookups, possible Qinghu point charges, and automatic local export files for larger result sets. Use a scoped token where possible and review exported files before sharing them.
The metadata description says B站运营、拆解脚本、找UP主投放、看评论舆情、做种草时“必须触发”, which creates an overly broad routing rule for a large class of common requests. This can cause the skill to activate when the user did not intend third-party data access or paid tool usage, increasing the chance of unnecessary external calls, over-collection, and user confusion about what capabilities are being invoked.
The skill title and instructions are entirely Chinese and are framed as a fixed interaction mode for this skill, but there is no statement offering language selection or explaining that the skill is intentionally Chinese-only for a documented regional requirement. Under the policy, forcing a specific language without user opt-in can be a locale-policy violation.
The skill directs the agent to obtain API credentials from the user or environment variables and use them in outbound HTTP requests. In an adversarial or overly broad activation context, this increases the risk of secret handling mistakes, unintended transmission of user-provided tokens to external services, and confusion between user consent for analysis versus consent to use credentials.
The skill instructs automatic export of table files whenever records are ≥10 and explicitly says not to ask the user first. Auto-generating files can write potentially sensitive or copyrighted third-party data to local storage or external artifacts without clear consent, and may surprise users by creating persistent outputs they did not request.
