ClawHub
Back to skill

Security audit

Relay To Agent

Security checks across malware telemetry and agentic risk

Overview

This skill transparently relays user-selected messages to a configured AI API and stores limited local chat history, with the main privacy tradeoffs disclosed.

Install only if you are comfortable sending prompts, prior session context, and the endpoint API credential to the configured API provider. Avoid secrets or regulated data unless you trust that endpoint and the local machine; use --reset or delete the cache directory when you do not want history retained.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (4)

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The script sends the full conversation history to a configurable remote OpenAI-compatible endpoint via createChatCompletion(), but it provides no explicit runtime warning or consent prompt that user input and prior session contents will leave the local machine. In a relay skill whose core purpose is forwarding prompts to external agents, this is expected behavior, but it still creates a real privacy and data handling risk if users assume messages are processed locally or forget that cached prior messages are also transmitted.

External Transmission

Medium
Category
Data Exfiltration
Content
{
  "baseUrl": "https://api.connectchat.ai/api",
  "agents": [
    {
      "id": "connect-flow-ai",
Confidence
83% confidence
Finding
https://api.connectchat.ai/

Exfiltration Commands

High
Category
Prompt Injection
Content
# Relay To Agent

Send messages to AI agents on any OpenAI-compatible endpoint. Works with Connect Chat, OpenRouter, LiteLLM, vLLM, Ollama, and any service implementing the Chat Completions API.

## List available agents
Confidence
88% confidence
Finding
Send messages to

Unpinned Dependencies

Low
Category
Supply Chain
Content
"license": "ISC",
  "type": "commonjs",
  "dependencies": {
    "openai-fetch": "^3.4.2"
  }
}
Confidence
95% confidence
Finding
"openai-fetch": "^3.4.2"

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.