Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed content-marketing workflow that uses web research, workspace draft files, optional publishing integrations, and memory for brand preferences in ways that fit its stated purpose.
Install this if you want a broad content-marketing assistant. Before using it with live CMS or social integrations, confirm each publish or schedule action, and periodically review stored brand preferences, avoided topics, competitor exclusions, and content calendars so stale memory does not affect future drafts.
The trigger list and usage examples include everyday phrases such as requests to write, research, or decide what to write next, without meaningful constraints. In agent ecosystems, overly broad triggers can cause unintentional routing to a powerful skill that performs multi-phase actions, which may expose user data or take actions the user did not specifically request.
The manifest description is extremely broad and designed to activate on generic writing, research, strategy, and publishing requests. That increases the chance this skill is invoked outside the user's intended scope, causing unexpected web research, memory access, content generation, or downstream publishing actions that exceed least-privilege expectations.
Persisting brand preferences and reusing them for every future content generation 'without asking again' enables autonomous decision-making based on stored memory rather than current user intent. In context, this is more concerning because the skill also spans research, formatting, publishing, and promotion, so stale or overly broad memory could silently influence outputs or actions across sessions.
- **Things to avoid**: jargon level, competitors not to mention, topics to skip
- **Existing content URL**: for internal linking and avoiding duplication
Once stored, use these preferences for every future content generation without asking again.
## Edge Cases
No suspicious patterns detected.