Oomol Fusion API

• Outcome: Pass
• Latest audit: Jun 3, 2026, 6:14 PM
• Version:
  1.0.0

Overview

This is a documented OOMOL Fusion API wrapper with some confusing conservative warnings, but no evidence of hidden execution, data theft, or destructive behavior beyond disclosed user-approved API actions.

Install only if you intend to let Codex operate your OOMOL Fusion API account through the oo CLI. Review schemas and payloads before approving uploads, media generation, edits, or deletes, and be aware that the documented first-time setup may run a remote oo CLI installer if the CLI is not already installed.

SkillSpector

By NVIDIA

Vulnerability Patterns

• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Intent-Code Divergence

Medium

• Confidence: 93% confidence
• Finding: The documentation is internally inconsistent: it describes the action as returning an image with background removed, but also labels the action as destructive and capable of removing or overwriting Oomol Fusion API data. This can mislead an agent or operator into treating a read/result-fetching action as safe while the warning suggests side effects, increasing the chance of accidental destructive execution or incorrect approval handling.

Intent-Code Divergence

Medium

• Confidence: 97% confidence
• Finding: The file documents a read-only state retrieval action, but the warning says the action is destructive and removes or overwrites data. This kind of contradictory guidance can mislead an agent or operator into treating a safe read as a risky write, causing unnecessary confirmation flows, incorrect automation behavior, or confusion about the action’s real safety properties; in a broader skill set, such mislabeling can also hide truly dangerous actions by making warnings unreliable.

Intent-Code Divergence

Medium

• Confidence: 94% confidence
• Finding: The documentation describes an image background-removal submission action as a destructive operation that removes or overwrites Oomol Fusion API data, which does not match the stated purpose of the action. Mislabeling an action’s risk and behavior can cause operators or downstream agents to make incorrect trust and approval decisions, potentially leading to unsafe execution patterns or confusion about what the action actually does.

Intent-Code Divergence

Medium

• Confidence: 90% confidence
• Finding: The action name and description indicate a result-retrieval operation, but the documentation warns that it is a write action that changes API state. Mislabeling a read-like action as state-changing can cause agents or users to apply the wrong safety and approval logic, leading to unnecessary confirmations or incorrect execution policies; if the opposite mismatch exists upstream, it could also hide real side effects.

VirusTotal

64/64 vendors flagged this skill as clean.