ClawHub
Back to skill

Security audit

Seedance 2 Video Generator

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed WeryAI video-generation integration; it sends prompts and chosen media to WeryAI, including optional local-file uploads, which fits its stated purpose.

Install only if you are comfortable sending prompts, media URLs, and any explicitly selected local media files to WeryAI. Prefer public HTTPS media URLs, use dry-run to inspect uploadPreview before paid runs, and do not provide sensitive private files or store WERYAI_API_KEY in the repository.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Context-Inappropriate Capability

Low
Confidence
89% confidence
Finding
resolvePublicUrlFromSource accepts arbitrary http:// URLs and passes them through as valid remote sources, despite the surrounding skill context emphasizing public HTTPS references. Allowing plaintext HTTP enables man-in-the-middle tampering, content substitution, and metadata leakage when media is fetched over untrusted networks, which is especially risky for a media-generation workflow that may consume attacker-controlled external resources.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
This CLI automatically uploads any non-URL local image/video/audio path to the WeryAI remote API during a normal run, which can exfiltrate local user files outside the host boundary. Although the skill’s purpose is media generation and upload is functionally expected, the lack of an explicit just-in-time warning/consent step makes accidental disclosure of sensitive local files more likely, especially when the tool is invoked by an agent on the user’s behalf.

Missing User Warnings

Low
Confidence
86% confidence
Finding
The script sends prompts and media references to external WeryAI endpoints, but the CLI usage/help text does not clearly disclose the network destinations or that submitted content is transmitted off-system. In an agent skill context, this weak disclosure can cause users to unknowingly share sensitive prompts, URLs, or media metadata with third-party services.

Static analysis

Detected: suspicious.env_credential_access

Environment variable access combined with network send.

Critical
Code
suspicious.env_credential_access
Location
scripts/vendor/weryai-core/upload.js:147

Environment variable access combined with network send.

Critical
Code
suspicious.env_credential_access
Location
scripts/video_gen.js:24