ClawHub

IMA Studio All-in-One — Image, Video, Music, SeeDream, Veo, Suno. Banana

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed IMA Studio content-generation integration that sends prompts, uploads, and an API key to expected provider endpoints, with only minor privacy notes around local preferences and logs.

Install only if you trust IMA Studio with your prompts, uploaded source media, and IMA API key. Use a dedicated or test API key where possible, monitor usage, and delete ~/.openclaw/memory/ima_prefs.json or ~/.openclaw/logs/ima_skills/ if you do not want local preference or generation history retained.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Intent-Code Divergence

Medium
Confidence
87% confidence
Finding
The document repeatedly claims that --user-id is never transmitted to IMA, yet all command examples pass it into the bundled script and this file provides no verifiable guarantee that the script strips it before outbound requests. If the script forwards or logs this value, users may unknowingly disclose identifiers to third parties or persist them in logs contrary to the privacy claims.

Intent-Code Divergence

Low
Confidence
81% confidence
Finding
The skill promises not to log sensitive prompts and IDs, but also declares persistent generation logs without specifying their contents or redaction behavior. That inconsistency can mislead users about local exposure, especially because prompts, model choices, errors, and identifiers are commonly written during generation workflows.

Context-Inappropriate Capability

Medium
Confidence
85% confidence
Finding
The script stores per-user model preferences in ~/.openclaw/memory/ima_prefs.json without any user consent, minimization, or protection controls. While not highly sensitive on its own, persistent per-user tracking exceeds the core content-generation function and can expose behavioral metadata or create privacy issues on shared systems.

VirusTotal

61/61 vendors flagged this skill as clean.

View on VirusTotal