Security audit
Security checks for vulnerabilities and agentic risk
This is mostly a markdown marketing-playbook bundle, but some modules claim direct ad-account and social-publishing access without clear approval controls.
Use this skill as draft/read-only marketing guidance unless you explicitly review each action. Do not allow it to change ad budgets, launch campaigns, upload audience lists, scrape social platforms, schedule or publish posts, or use customer behavioral data without separate approval, privacy/legal review, and clear account-level controls.
Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
- `references/marketing-ideas/SKILL.md`
- `references/marketing-psychology/SKILL.md`
## Output rules
- Prefer 80/20: biggest levers first.
- Never invent metrics or keyword volumes. If missing, label assumptions.
Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
- Cool-down period after dismiss (hours vs. days)
- Escalating urgency over time vs. consistent messaging
- Once per feature vs. consolidated prompts
- Re-show rules after major engagement
**Dismiss Behavior**
- "Maybe later" vs. "No thanks" vs. "Remind me tomorrow"
Model output is used without validation or sanitization. Unvalidated output injected into downstream contexts (SQL, shell, HTML) enables injection attacks and arbitrary code execution.
<Head>
<script
type="application/ld+json"
dangerouslySetInnerHTML={{ __html: JSON.stringify(schema) }}
/>
</Head>
{/* Page content */}
The skill description says to use the skill when the user wants to "plan, design, or implement an A/B test or experiment" and also when they mention terms like "test this change" or "hypothesis." Phrases such as "test this change" and "hypothesis" are broad and could appear in ordinary discussion without the user actually seeking A/B test setup, increasing the chance of unintended invocation.
The trigger description is very broad and includes many common analytics terms, which can cause the skill to activate in situations where the user only loosely references tracking. Over-broad activation increases the chance of inappropriate guidance being injected into unrelated workflows and can lead to unintended handling of sensitive implementation or compliance questions without explicit user intent.
Tool defaults are unsafe or overly permissive (e.g. disabled TLS verification, no authentication, world-writable permissions). Unsafe defaults widen the attack surface.
**GA4 DebugView**
- Real-time event monitoring
- Enable with ?debug_mode=true
- Or via Chrome extension
**GTM Preview Mode**
The trigger description is overly broad and matches generic competitor-related phrases such as 'competitor comparison' and 'vs page,' which can cause the skill to activate outside its intended narrow workflow. This can lead to inappropriate routing, where a specialized SEO page-generation skill takes over ordinary research or discussion requests and produces biased or mis-scoped output.
The manifest description says to use the skill when the user wants to "edit, review, or improve" copy and includes phrases like "polish this" and "make this better." These are broad, common utterances that could match many non-marketing contexts, increasing the risk of unintended invocation despite the marketing-copy mention.
The manifest description says to use the skill whenever the user wants to create or optimize an email sequence, drip campaign, automated email flow, or lifecycle email program, which is a wide natural-language activation condition rather than a tightly bounded trigger. It provides examples, but no exclusion conditions or negative examples, increasing the chance of unintended invocation for general email-marketing requests.
The skill encourages behavior-driven outreach based on user activity signals such as usage drops, failed actions, and help-doc viewing, but provides no guidance on consent, data minimization, or privacy/legal constraints. In practice, this can lead agents to generate surveillance-like messaging or operationalize sensitive behavioral profiling without verifying that the underlying data use is permitted.
The recurring usage reports and personalized summaries section promotes inclusion of activity metrics, comparisons, and achievements without warning about secure handling, audience appropriateness, or sensitivity of the data. This can result in overexposure of personal or business-sensitive usage information in email, especially if reports are sent to the wrong recipient, shared broadly, or generated from data that should not be surfaced.
The skill description contains very broad trigger phrases such as 'contact form,' 'form fields,' and 'form optimization,' which could cause the agent to invoke this skill for general conversations rather than only when the user explicitly wants CRO guidance. Over-broad routing is a real security and reliability issue because it increases unintended activation surface and may override more appropriate, narrower skills.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
## Field-by-Field Optimization
### Email Field
- Single field, no confirmation
- Inline validation
- Typo detection (did you mean gmail.com?)
- Proper mobile keyboard
The trigger description is overly broad and includes common phrases like "calculator," "generator," and "interactive tool," which can cause the skill to activate for many ordinary user requests outside its intended scope. Unintended invocation can misroute conversations, override more appropriate skills, and lead the agent to produce marketing-oriented guidance when the user did not ask for it.
The skill recommends email gating, lead capture, follow-up sequences, and sharing anonymized results, but does not instruct users to obtain consent, disclose data usage, or handle personal data in compliance with privacy requirements. This can encourage collection or secondary use of user data without adequate notice or safeguards, creating privacy, legal, and trust risks.
The trigger description is broad enough to activate on common terms like 'launch' and 'announcement' without sufficient contextual constraints. This can cause the skill to be invoked for unrelated user requests, leading to misrouting, irrelevant guidance, and reduced trust in agent behavior, though it does not directly enable code execution or data exfiltration.
The activation description is very broad and includes common phrases like 'marketing ideas' and 'how to market,' which can cause the skill to trigger for many general requests without sufficient specificity. Over-broad routing is dangerous because it can override more specialized skills or inject this skill's canned guidance in contexts where narrower, safer, or more context-appropriate behavior should apply.
The skill recommends tactics involving tracking, email capture, retargeting, and audience/data sharing, but it does not include any guardrails around consent, lawful basis, transparency, or platform/privacy compliance. In practice, this can lead users to implement invasive or noncompliant data collection and ad-targeting workflows, creating privacy, legal, and reputational risk.
The skill description is activated by a very broad set of common terms such as 'psychology,' 'persuasion,' 'decision-making,' and 'consumer behavior,' without meaningful constraints on scope or context. This can cause the skill to trigger in many loosely related conversations and steer the agent toward persuasion-oriented guidance when the user did not specifically request this specialized behavior, increasing the risk of inappropriate invocation and manipulative recommendations.
The manifest description activates the skill whenever a user wants to optimize onboarding, activation, first-run experience, or time-to-value, which are broad product phrases that may appear in many general strategy conversations. Although examples are provided, the trigger scope is still expansive and lacks explicit exclusion conditions beyond two related skills, increasing the chance of unintended invocation.
The skill description contains broad trigger phrases such as "optimize," "improve," and "why isn't this page working," which can match routine marketing conversations beyond explicit CRO requests. This can cause the agent to invoke the skill in unintended contexts, leading to inappropriate guidance, context hijacking, or suppression of a more suitable skill for the user's actual task.
The trigger description is broad enough to match generic marketing or advertising conversations, increasing the chance the skill is invoked outside its intended scope. In an agentic environment, unintended invocation can lead to irrelevant or premature campaign recommendations and, when paired with account-connected capabilities, could influence spend-affecting actions without clear user intent.
The skill states it has direct access to ad platform accounts, but it provides no safety boundary around making account changes, budget edits, audience uploads, or launching campaigns. Because these actions can immediately spend money, alter targeting, or impact live business operations, the absence of explicit confirmation and change-control guidance creates a real operational security risk.
The manifest says to use the skill 'when the user wants help with pricing decisions, packaging, or monetization strategy,' which is a wide natural-language trigger that could match many general business conversations. Although examples are provided, the description does not include negative examples or clear limits on when this skill should not activate.
The description says to use the skill when the user wants to create SEO-driven pages at scale and includes broad phrases like "comparison pages" and "integration pages." Those terms are common across many content and product-marketing tasks, and the file does not provide exclusion criteria beyond a single note for SEO audits, which could cause unintended invocation.
No suspicious patterns detected.