uncloak

fevziegeyurtsevenler/uncloak
★ 0 stars Python AI/LLM Updated today
Reveal hidden prompt injection & supply-chain risks in AI agent extensions — scans Claude/agent Skills, MCP servers & rules files for invisible Unicode instruction smuggling, tool poisoning & the lethal trifecta. Terminal · JSON · SARIF · zero deps.
View on GitHub → 🔍 Audit Wallet Slippage →

Quick Install

Copy the config for your editor. Some servers may need additional setup — check the README.

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "uncloak": {
      "command": "uvx",
      "args": [
        "uncloak"
      ]
    }
  }
}

Or install with pip: pip install uncloak

README Excerpt

<p align="center"><img src="assets/banner.svg" alt="uncloak" width="100%"></p> <h1 align="center">uncloak</h1> <p align="center"> <b>Reveal the prompt injection you can't see.</b><br> A zero-dependency, <b>multilingual</b> scanner that finds <b>hidden instructions</b> and <b>supply-chain risks</b><br>

Topics

agent-securityai-securityappsecclaudellm-securityllmopsmcpmodel-context-protocolprompt-injectionred-teamingsarifsecurity-toolssupply-chain-securityunicode