Back to skill

Security audit

OpenAI-Compatible LLM Gateway

Security checks for vulnerabilities and agentic risk

Overview

This is a documentation-only skill for using a third-party OpenAI-compatible LLM gateway, with external prompt transmission clearly part of its stated purpose.

Only use this gateway for data you are allowed to send to api.gocreativeai.com and any upstream model providers it may route to. Treat funded keys as credentials, keep them revocable, and review the provider's privacy, retention, billing, and compliance terms before sending confidential or regulated content.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (6)

External Transmission

Medium
Category
Data Exfiltration
Confidence
84% confidence
Finding

The base URL directs all compatible client traffic to a third-party endpoint, meaning requests intended for an OpenAI-style interface are transmitted outside the user's trusted boundary. In practice this can expose prompts, conversation history, and usage metadata to the gateway and any upstream providers it routes to.

Content

Scanner excerpt · SKILL.md (reported line 21)May include surrounding context.

Base URL

text
https://api.gocreativeai.com/v1

Endpoints

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill instructs users to send prompts to an external gateway but does not prominently warn that prompts, model inputs, and potentially sensitive user data will leave the local environment and be processed by a third party. This omission can cause accidental disclosure of confidential data because users may assume a standard local/OpenAI-compatible interface without appreciating the extra intermediary.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
89% confidence
Finding

The SDK example configures an OpenAI client to send chat messages to api.gocreativeai.com rather than a first-party provider, so all prompt contents and responses transit a third-party service. Because the interface is drop-in compatible, users may not realize they are introducing an additional processor and trust boundary.

Content

Scanner excerpt · SKILL.md (reported line 40)May include surrounding context.

Drop-in example (OpenAI SDK)

python
from openai import OpenAI
client = OpenAI(base_url="https://api.gocreativeai.com/v1", api_key="gck_<your-key>")
resp = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Summarize this in one line: ..."}],

External Transmission

Medium
Category
Data Exfiltration
Confidence
90% confidence
Finding

The curl example explicitly sends user-supplied prompts and an authorization credential to an external service. This is dangerous in agent environments because it creates a direct data exfiltration path for prompts, outputs, and possibly sensitive context to a third-party gateway, even if the documentation presents it as normal usage.

Content

Scanner excerpt · SKILL.md (reported line 48)May include surrounding context.

print(resp.choices[0].message.content)

text

## curl

curl -X POST https://api.gocreativeai.com/v1/chat/completions
-H "Authorization: Bearer gck_" -H "Content-Type: application/json" \

External Transmission

Medium
Category
Data Exfiltration
Confidence
90% confidence
Finding

This example transmits both an authorization bearer token and prompt content to an external API over the network. While expected for a gateway service, it is still a real security concern because the skill does not pair the example with sufficient disclosure, consent, or guidance about handling sensitive prompts and credentials.

Content

Scanner excerpt · SKILL.md (reported line 50)May include surrounding context.

curl

text
curl -X POST https://api.gocreativeai.com/v1/chat/completions \
  -H "Authorization: Bearer gck_<your-key>" -H "Content-Type: application/json" \
  -d '{"model":"gpt-4o-mini","messages":[{"role":"user","content":"hi"}]}'

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill is presented as an OpenAI-compatible LLM gateway, but it also advertises access to 350+ data and compliance tools via the same base URL. This scope expansion is security-relevant because users may trust the skill as a narrow LLM proxy while it actually exposes broader third-party capabilities and data flows not clearly declared in the manifest.

Content

No source excerpt is available for this finding.