Security audit

Ai Starter

Security checks for vulnerabilities and agentic risk

Overview

This is a plain brainstorming prompt skill with broad activation language but no code, tools, persistence, credential access, or hidden behavior.

Install this skill if you want a lightweight brainstorming assistant. Be aware it may activate for broad idea-exploration phrases, so users who prefer only explicit opt-in brainstorming may want tighter trigger wording.

A.I.G

Vulnerability Patterns

Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Vague Triggers

Medium

Confidence: 93% confidence
Finding: The 'When to Use' guidance is broad and loosely scoped, encouraging activation for many everyday requests involving thinking, ideas, or exploration. In an agent ecosystem, this increases the chance of over-selection and accidental use, which can degrade safety, relevance, and predictability by hijacking normal conversations.

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The trigger table uses very generic phrases such as 'What if...' and 'Go deeper on...' that are common in ordinary conversation. In systems that auto-activate skills based on prompt matching, this can cause unintended invocation, routing user input into this skill when the user did not explicitly request it, which may override more appropriate skills or alter assistant behavior unexpectedly.

Vague Triggers

Low

Confidence: 88% confidence
Finding: The quick guide's 'when' conditions are vague statements like 'when you need to think' or 'when you want options,' which are nearly universal and not meaningfully distinguishing. This ambiguity can contribute to accidental activation, though the impact is lower because it is guidance text rather than a primary trigger list.

Natural-Language Policy Violations

Low

Confidence: 84% confidence
Finding: The trigger list mixes English phrases with Chinese phrases such as "换个角度" and "换个思路" but does not state that multilingual triggering is optional or explain the locale scope. This creates an undocumented language/locale behavior in the natural-language activation policy.

Static analysis

No suspicious patterns detected.