Flyworks Avatar Video

• Outcome: Review
• Latest audit: Jul 10, 2026 · 3:49 AM UTC
• Version: 1.0.0

Overview

The skill appears to do what it promises, but it can upload face and voice media to a third-party service and automatically uses a shared demo token, so users should review it before installing.

Review before installing. Use your own Flyworks/HiFly token, assume any image or audio path you provide may be uploaded to Flyworks/HiFly for processing, and only use portraits or voice samples that you own or have explicit permission to use. Check the provider's retention, deletion, and acceptable-use terms before processing sensitive or biometric media.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (11)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 92% confidence
• Finding: The skill declares no explicit permissions even though its documented behavior requires environment variable access, local file read/write, and outbound network access. This creates a transparency and governance gap: users and platforms may not realize the skill can transmit user media and tokens to an external service or persist aliases locally.

Context-Inappropriate Capability

High

• Confidence: 98% confidence
• Finding: The script embeds a hardcoded default API token and automatically uses it when no environment variable is set, giving anyone who runs the skill immediate credentialed access to a third-party account. Hardcoded shared credentials are dangerous because they bypass normal user consent and account separation, and they may expose the vendor account to abuse, quota theft, or attribution of user actions to the wrong party.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The README explicitly promotes voice cloning from user audio but provides no guidance about consent, lawful use, retention, or protection of biometric voice data. In an agent skill context, this omission increases the chance that users or automated agents will process sensitive audio without appropriate authorization or privacy safeguards, enabling impersonation and privacy harms.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The skill encourages uploading images, audio, text, and other user-provided content to the Flyworks/HiFly API but does not clearly warn users that this data leaves the local environment and is sent to a third party. This can lead to inadvertent disclosure of sensitive personal, biometric, or proprietary content under the false impression that processing is local.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The skill advertises voice cloning without a specific warning about consent, impersonation risk, and the sensitive biometric nature of voice samples. In context, this is more dangerous because the core feature directly enables cloning from uploaded audio, which can facilitate non-consensual impersonation, fraud, or privacy violations if users submit someone else's voice.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The documentation instructs users to upload portrait images from local files or URLs to create talking-photo avatars, but it omits any warning about consent, biometric privacy, copyright, or the sensitivity of facial images. In a digital-human/video-generation skill, that omission increases the likelihood of misuse such as cloning a person’s likeness without permission or exposing personal images to a third-party service.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The skill provides step-by-step voice cloning instructions, including cloning from local files or remote URLs, without any warning about consent, biometric privacy, impersonation risk, or the sensitivity of uploaded audio. Because voice prints are biometric identifiers and cloned voices can enable fraud, social engineering, or unauthorized impersonation, the omission of safety guidance materially increases misuse risk.

Missing User Warnings

Medium

• Confidence: 88% confidence
• Finding: The skill accepts local image/audio paths and transparently uploads those files to the Flyworks service without an explicit privacy notice at the point of use. This is risky because users may assume local processing and inadvertently send private media off-device, especially in an agent/skill context where file handling can be abstracted away.

Missing User Warnings

High

• Confidence: 96% confidence
• Finding: Voice cloning involves transmitting biometric voice data to an external API, yet the code provides no clear warning, consent flow, or retention guidance. This is especially sensitive because cloned voice samples can enable impersonation, fraud, and irreversible privacy harm if the audio belongs to the user or another person.

Unpinned Dependencies

Low

• Category: Supply Chain

Content

requests

• Confidence: 97% confidence
• Finding: requests

Known Vulnerable Dependency: requests — 10 advisory(ies): CVE-2014-1830 (Exposure of Sensitive Information to an Unauthorized Actor in Requests); CVE-2024-47081 (Requests vulnerable to .netrc credentials leak via malicious URLs); CVE-2024-35195 (Requests `Session` object does not verify requests after making first request wi) +7 more

High

• Category: Supply Chain
• Confidence: 94% confidence
• Finding: requests

VirusTotal

66/66 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.