Security audit
Security checks for vulnerabilities and agentic risk
This is a coherent content-production prompt skill with minor scope and audience-framing issues, but no evidence of hidden, destructive, or deceptive behavior.
Install only if you want a content repurposing prompt pack. Review outputs for audience fit, especially the Christian parent and mom-oriented assumptions, and direct the agent where drafts should be saved before allowing file writes or optional sub-agent runs.
The README presents activation phrases like "Run the content factory on this article" and multiple "Act as the ..." commands as the way to invoke the skill, but it does not define a bounded set of accepted triggers or any exclusion conditions. These phrases are broad natural-language requests that could plausibly appear in normal conversation, making unintended invocation more likely.
The prompt hard-codes a specific religious and demographic identity for the protagonist/audience ('Christian mom, 28-42, kids 3-6') without any user opt-in or contextual check. This can cause the agent to inject unwanted identity assumptions into outputs, leading to exclusionary, manipulative, or mismatched content when the actual user does not share that identity.
The skill metadata states it reads source content and writes output drafts to the working directory, but the markdown body does not clearly warn users about file creation or describe where files may be written. This can lead to unexpected filesystem side effects, especially in automated or shared workspaces, even though the skill does not appear to perform destructive actions.
The instruction to test whether 'a tired mom at 10pm' would understand the output imposes a specific demographic and English-language cultural framing rather than offering a user-selected audience or locale. This can conflict with language/locale neutrality expectations because the prompt hard-codes one audience persona instead of making it optional.
No suspicious patterns detected.
