Ae1
High
- Category
- analysis-evasion
- Confidence
- 100% confidence
- Finding
Referenced artifact was not completely inspected
- Content
md [English](./SKILL.md) · [中文](./SKILL-cn.md)
Security audit
声音克隆 ElevenLabs Voice Clone
Security checks for vulnerabilities and agentic risk
Overview
The skill is a disclosed voice-cloning wrapper, but it asks users to upload sensitive voice data while its documentation has material copy-paste inconsistencies and weak consent/privacy guidance.
Review this carefully before installing. Only upload voice samples you own or have explicit permission to clone, avoid confidential recordings, check dLazy and ElevenLabs retention/deletion terms, and prefer the per-run npx form until the command/output documentation is corrected.
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
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (8)
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 90% confidence
- Finding
This skill handles voice cloning, which involves uploading human voice samples that may be biometric and sensitive personal data, yet the document does not clearly warn users about consent, privacy, retention, and impersonation risks. In a voice-cloning context, missing safety and privacy disclosures increases the chance of unauthorized cloning or accidental exposure of sensitive audio.
- Content
Intent-Code Divergence
Medium
- Category
- Not specified by scanner
- Confidence
- 96% confidence
- Finding
The documented output format says this voice-cloning skill returns an image (
type: image,result.png), which materially contradicts the stated function of cloning/uploading a voice. This can mislead users or downstream agents into mishandling outputs, validating the wrong artifact type, or trusting documentation that appears copy-pasted from another skill, which raises reliability and security concerns around what data is actually uploaded and returned.- Content
Intent-Code Divergence
Medium
- Category
- Not specified by scanner
- Confidence
- 97% confidence
- Finding
The command example and error messages reference
--promptand image-file failures instead of the documented--audio_urlworkflow for voice cloning. This inconsistency can cause users or automation to supply the wrong input type, potentially uploading unintended files or misunderstanding required parameters for a biometric-processing operation.- Content
Whitespace Padding
Medium
- Category
- Prompt Injection
- Confidence
- 70% confidence
- Finding
Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.
- Content
md ## 错误处理 | Code | 错误类型 | 示例信息 | | ---- | -------------------------------- | ------------------------------------------------------------------------------------------------------------------------ | | 401 | 未授权 (API Key缺失或无效) | `ok: false, code: "unauthorized", message: "API key is missing or invalid"` | | 501 | 缺少必填参数 | `error: required option '--prompt <prompt>' not specified` |
Intent-Code Divergence
Medium
- Category
- Not specified by scanner
- Confidence
- 90% confidence
- Finding
The documented options list only
--audio_url,--name, and--description, yet the example invocation and error table claim the command requires--prompt. This creates an intent-code/documentation divergence because the documentation presents contradictory interfaces for what the command actually accepts.- Content
Intent-Code Divergence
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The skill is described throughout the manifest as an ElevenLabs voice-cloning tool, but the documented output example shows
type: "image",result.png, andimage/png. This is not merely incomplete documentation; it actively describes a different modality than the skill's declared audio/voice behavior.- Content
Whitespace Padding
Medium
- Category
- Prompt Injection
- Confidence
- 70% confidence
- Finding
Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.
- Content
md ## Error Handling | Code | Error Type | Example Message | | ---- | ---------------------------------- | ------------------------------------------------------------------------------------------------------------------------ | | 401 | Unauthorized (No API Key) | `ok: false, code: "unauthorized", message: "API key is missing or invalid"` | | 501 | Missing required parameter | `error: required option '--prompt <prompt>' not specified` |
Static analysis
No suspicious patterns detected.