Back to skill

Security audit

OOMOL Fusion API

Security checks for vulnerabilities and agentic risk

Overview

This skill needs review because some media operations bypass its approval rules and setup executes a remote installer without verification.

Before installing, have the publisher correct the action classifications and make CLI setup explicitly user-directed. Confirm costs and external processing before generation or editing requests. Use a reviewed CLI release and consider disabling startup skill synchronization with OO_SKILLS_SYNC_DISABLED=1 before invoking it. The remote installers were unavailable for inspection, so their exact local changes remain uncertain.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (5)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

The supplied chunk performs an installation action and contains no Fusion API operations. Although CLI installation could support an API workflow, no such workflow is shown here; installation is the chunk's sole behavior. The downloaded script is not supplied, so its further effects cannot be determined.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
97% confidence
Finding

The setup instructions pipe a remotely fetched, unpinned script directly into Bash; the PowerShell example uses the equivalent fetch-and-execute pattern. If the distribution endpoint or release infrastructure is compromised, attacker-controlled code could execute with the agent's local permissions and access local files or credentials; the missing-CLI condition limits exposure but does not provide integrity verification.

Content

Scanner excerpt · SKILL.md (reported line 138)May include surrounding context.

L127 incorrectly classifies untagged generation and editing actions as reads, bypassing the stated confirmation policy.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The blanket rule that untagged actions are reads incorrectly covers generation and editing operations such as qwen_image_generate and qwen_image_edit_plus_edit. This lets the agent bypass the skill's confirmation policy for operations that may consume paid credits, process supplied content externally, or create outputs, although it does not establish that those operations overwrite existing user data.

Content

No source excerpt is available for this finding.

The safety instructions classify untagged image-editing and generation actions as reads, allowing them to run without disclosing their actual effects.

Medium
Category
Output Handling
Confidence
98% confidence
Finding

The safety instructions permit untagged generation and editing actions to run directly while describing them as reads. This can omit meaningful user confirmation of processing effects and charges; it is the same underlying classification defect as SDI-4, not an independent vulnerability.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 153)May include surrounding context.

md
- **`OOMOL_INSUFFICIENT_CREDIT`** — OOMOL billing stop. Recharge at `https://console.oomol.com/billing/token-recharge` before retrying.

- **HTTP 402 without `OOMOL_INSUFFICIENT_CREDIT`** — the third-party account itself has no active plan or credit. Report the provider error message as is; do not ask the user to recharge OOMOL.

## Resources