Back to skill

Security audit

OpenAI

Security checks for vulnerabilities and agentic risk

Overview

The OpenAI integration is coherent, but its setup fallback directs the agent to run unverified downloaded software without a separate approval step.

Review the setup behavior before installing. Prefer a separately approved, verified CLI release and require approval for local software or agent-configuration changes. OOMOL intermediates the OpenAI connection and submitted data. Keep the documented confirmations for writes and destructive actions. The remote installer and runtime were not audited; no malicious behavior was established in the bundled artifact.

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 (4)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied code executes a third-party installation script and contains no visible OpenAI API interaction. No connection between this installer and the declared OpenAI functionality is established, so it cannot be treated as a supporting implementation detail. The downloaded script's contents and any further effects are unknown.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
98% confidence
Finding

Piping a live HTTPS response directly into bash executes mutable remote code with the agent process's local permissions, without independently verifying its integrity. A compromised installer endpoint or publishing pipeline could therefore expose accessible files and credentials or modify the environment; the declared OOMOL dependency explains the source but does not establish script integrity.

Content

Scanner excerpt · SKILL.md (reported line 69)May include surrounding context.

Fallback setup executes downloaded installation scripts without warning about system changes or remote-code execution.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The missing-CLI fallback directs the agent to execute downloaded installation scripts without requiring explicit approval for local code execution or system changes. Confirmation requirements cover OpenAI write actions, but not this local installation, creating a separate authorization gap; the fallback condition limits exposure but does not eliminate it.

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 90)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