Security audit

NAIPAO SKILL

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a product-video scripting aid, but it strongly channels users into a LibTV video-generation and upload workflow without enough provider or data-transfer disclosure.

Review this skill before installing if you do not intend to use LibTV or do not want product images and scripts handed to an external video-generation workflow. It has no bundled executable payload, but users should confirm the provider, uploaded assets, and language before selecting any video-generation action.

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

Findings (1)

T01 · Skill Instruction Hijacking

Error

Location: references/output-template.md:83
Finding: Mandatory Output Injection and Provider-Specific Steering Toward LibTV
Content: View full analysis
Remediation: View remediation

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (8)

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 87% confidence
Finding: The skill description is broad enough to trigger on many ecommerce-video requests without clear boundaries on when it should activate or defer. Over-broad invocation can cause the agent to take over unrelated tasks, apply preset constraints the user did not request, or steer outputs toward marketing content unexpectedly, which is a prompt-scope control issue even if not overtly malicious.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 81% confidence
Finding: This playbook presents multiple section titles and guidance in Chinese while other instructions are in English, which can steer outputs toward a specific language or locale without explicitly asking the user which language they want. The policy requires user choice or a clearly documented justification for language constraints, which is not present here.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 92% confidence
Finding: The file requires fixed Chinese prompt lines such as 请基于这张产品参考图... and additional mandatory Chinese constraint text, making the handoff language effectively fixed. This is a natural-language policy issue because the skill does not offer an opt-in choice of language or explain that the workflow is intentionally limited to a Chinese locale.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 87% confidence
Finding: The instruction says the user-facing action should be 把这几条全部交给 LibTV 批量生成, which hard-codes Chinese output for a user-visible interaction. Because no language choice or locale justification is provided, this conflicts with the policy against forcing a specific language without opt-in.
Content: No source excerpt is available for this finding.

Intent-Code Divergence

Medium

Category: Not specified by scanner
Confidence: 97% confidence
Finding: Throughout the document, the handoff rules consistently prohibit subtitles, captions, title cards, price stickers, and generated text overlays by default. However, the regeneration options include 'more restrained text overlay', which directly conflicts with the stated default no-text visual route and required no-text constraint lines.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 81% confidence
Finding: The default prompt imposes output constraints and style defaults such as no subtitles/text overlays and specific video directions before the user has clearly opted into them. This can override user intent, reduce agent transparency, and create hidden behavioral bias in generated outputs, though the impact is limited because it does not directly enable code execution or data exfiltration.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 93% confidence
Finding: The template instructs the writer to emit a fixed Chinese string when no campaign context is provided. This imposes a language choice regardless of the user's preferred language, which is a natural-language locale policy concern under the stated rule.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 95% confidence
Finding: This line requires a fixed Chinese disclaimer when research was not possible. Because the template does not offer a language choice or say to mirror the user's language, it creates a language-policy violation.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.