Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed content workflow, but it can automatically publish generated posts to public platforms without a clearly enforced final approval step.
Install only if you intend to connect WeChat/Xiaohongshu publishing tools. Use it in draft or preview mode where possible, require explicit approval before every live post, verify connected accounts and credentials, and avoid enabling scheduled runs unless you have review and rollback procedures.
The README explicitly describes automatic publishing to public platforms and says only a light human review is needed before release. Actions like publishing content to WeChat and Xiaohongshu can affect user accounts, brand reputation, and external systems, yet the documentation does not present any clear warning, confirmation gate, or requirement for explicit user consent before posting.
The skill prominently advertises one-click multi-platform publishing without an upfront warning that content may be transmitted to external services and posted publicly. Because the workflow includes automated publication to WeChat and Xiaohongshu, users may unknowingly trigger irreversible public actions, creating privacy, reputational, and account-misuse risks.
The trigger phrases are generic enough to overlap with ordinary content-assistance requests, which raises the chance the skill activates when a user did not clearly intend to invoke an automated publishing workflow. In this skill, unintended activation is more concerning because the workflow can progress from drafting into external publication actions across multiple platforms.
The workflow is described as a high-automation, end-to-end content creation and publishing pipeline, but it does not define clear invocation boundaries, approval gates, or trigger constraints. In practice, broad scopes like this can enable misuse for mass content generation or unintended publishing behavior, especially when combined with downstream publication steps.
The workflow includes automated publishing to Xiaohongshu and WeChat, but the manifest does not provide an explicit user-facing warning that generated content may be posted to external platforms. This creates a real risk of accidental publication, reputational damage, spam/policy violations, or disclosure of sensitive/generated content if the workflow is run without clear informed consent and a final confirmation step.
The skill describes automated cross-platform searching and aggregation of engagement data from multiple sources without any privacy, terms-of-service, or data-handling guidance. While this is less severe than auto-publishing, it can still lead to collection or processing of third-party data in ways users may not expect, especially if platform restrictions or sensitive sources are involved.
The description states that content will be written in the style of “刘润等头部公众号风格,” which imposes a specific stylistic constraint by default. The file does not indicate that users can choose a different language/style setting or opt in to that preset persona.
The workflow instruction explicitly says to write the article “按刘润风格,” which enforces a particular stylistic policy in the operational steps. No alternative or user-selectable style behavior is documented alongside this instruction.
The example output includes “刘润风格” as the expected result, reinforcing that the skill defaults to a specific style/persona. The file does not present this as an optional setting chosen by the user.
The manifest description and workflow are written entirely for a Chinese-language publishing context centered on WeChat and Xiaohongshu, but there is no indication that the user can opt into this locale/language behavior. This may violate language-choice expectations if the skill is used in a broader environment.
No suspicious patterns detected.
