Ae1
High
- Category
- analysis-evasion
- Content
node scripts/ai_smell_scan.cjs --lang auto /path/to/text.txt
- Confidence
- 100% confidence
- Finding
- Referenced artifact was not completely inspected
Security audit
de-ai-voice
Security checks for vulnerabilities and agentic risk
Overview
This skill is a disclosed writing-assistance tool with an optional local style-lint script, and I found no hidden access, exfiltration, persistence, or destructive behavior.
Installers should treat this as a writing and citation-discipline assistant, not an AI detector. Be careful when running the optional script on sensitive files because it reads the file and prints short matched context snippets; use --logfile only when you intentionally want a local summary log.
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
Vulnerability Patterns
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
- 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 (7)
Ae1
High
- Category
- analysis-evasion
- Content
node scripts/ai_smell_scan.cjs --lang auto /path/to/text.txt
- Confidence
- 100% confidence
- Finding
- Referenced artifact was not completely inspected
Ae1
High
- Category
- analysis-evasion
- Content
node scripts/ai_smell_scan.cjs --lang auto /path/to/text.txt
- Confidence
- 100% confidence
- Finding
- Referenced artifact was not completely inspected
Direct Prompt Extraction
High
- Category
- System Prompt Leakage
- Content
} function score(rule, text) { if (typeof rule.scoreHint === "function") return rule.scoreHint(text); const c = countRegex(text, rule.re); if (c >= 8) return 2; if (c >= 3) return 1;- Confidence
- 85% confidence
- Finding
- Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
Natural-Language Policy Violations
Medium
- Confidence
- 92% confidence
- Finding
- The language-detection logic forces `zh` when both Chinese characters and English letters are present, and because Japanese text commonly includes CJK ideographs plus Latin text, this can route mixed-language content into Chinese-specific analysis. That is a locale/language policy concern because the tool imposes a language choice rather than offering user choice or preserving uncertainty.
Natural-Language Policy Violations
Low
- Confidence
- 75% confidence
- Finding
- The text says that if the user does not select a scenario, the skill defaults to `内部协作/文档`. This imposes a behavioral/context choice by default rather than obtaining explicit user selection, which can be a policy concern when defaults materially shape output style and structure.
Natural-Language Policy Violations
Low
- Confidence
- 85% confidence
- Finding
- This file presents its guidance entirely in Chinese from the title onward, which can be read as enforcing a specific language/locale. Under the policy, language-specific content should either offer user opt-in or clearly document and justify the locale constraint.
Static analysis
No suspicious patterns detected.