Security audit
Security checks for vulnerabilities and agentic risk
This skill is a static community-analysis prompt, but it explicitly asks agents to map social weaknesses and identify what to exploit in real communities.
Install only if you intentionally want adversarial discourse-risk analysis and can keep it limited to authorized, appropriately redacted community data. Treat its outputs as sensitive, and avoid using them to manipulate, pressure, or target real people or communities.
The required output explicitly includes 'What Smith Exploits,' which operationalizes the analysis into actionable exploitation guidance after modeling resistance and skepticism vectors. That turns the skill from passive observation into a decision aid for influence operations against real communities, making misuse straightforward and increasing the likelihood of targeted manipulation at scale.
The skill explicitly ingests 'chat_log', 'thread', 'discussion', 'community_context', and 'social_post', which are likely to contain personal, behavioral, or sensitive community data, but it provides no operational guidance for minimization, consent, retention, redaction, or safe handling. In a skill focused on mapping trust, fear, resistance, and skepticism in real communities, the absence of privacy safeguards increases the risk of profiling users and processing sensitive social data inappropriately.
This section moves beyond neutral sentiment or risk analysis into identifying social weaknesses, distrust triggers, and where messaging 'will fail first,' which creates a playbook for targeting persuasion pressure points in real communities. In context, the skill is branded as 'adversarial-intelligence' and oriented around infiltration, so the language materially enables manipulative social engineering rather than merely defensive analysis.
The trigger phrases are broad, generic, and overlap with ordinary requests about community analysis, sentiment, and discourse, which can cause the skill to activate outside a narrowly intended use case. In the context of a skill explicitly framed as 'infiltration' and resistance mapping, unintended activation increases the chance the agent is steered into manipulative or socially targeted analysis without clear user intent or adequate safeguards.
No suspicious patterns detected.
