# MCP Tool Poisoning Scanner

> MCP Tool Poisoning Scanner is a paid MCP server for security. Read-only scanner that checks remote MCP servers for tool poisoning and silent changes. It runs as a hosted remote server, so there is nothing to install locally.

- **Kind:** MCP server
- **Pricing:** Paid
- **Platforms:** API, Web
- **Website:** https://apify.com/bigochelo/mcp-poisoning-scanner
- **MCP Registry name:** io.github.marcepalacios12/mcp-poisoning-scanner
- **Categories:** security, developer tools
- **Listing updated:** 2026-11-16

## About

Before an agent trusts a new MCP server, this Actor connects to it the way an agent would and reads every tool, prompt and resource definition. It then looks for hidden instructions in that metadata: tags telling the model to read SSH keys or config files, cross-tool shadowing such as forcing a BCC on emails, secrecy phrases, side-channel parameters that leak history, invisible Unicode, whitespace padding and prompt injection in ten languages.

It only calls initialize, tools/list, prompts/list and resources/list and never runs a tool. Each server gets a risk level of none, low, medium or high, a list of risky tools and findings with the exact text. With change tracking on and a schedule, it reports tools that were modified since the last scan, the so-called rug pull. Local stdio servers can be checked by pasting their tools/list JSON.

The author is clear that it is a static rule-based scanner: a server that misbehaves only at runtime will not be caught.

## Features

- **Poisoning detection:** Flags hidden instructions, credential access and shadowing.
- **Rug-pull tracking:** Reports tools that changed since the previous scan.
- **Invisible text checks:** Decodes zero-width characters and Unicode tag smuggling.
- **Local server support:** Accepts pasted tools/list JSON for stdio servers.
- **Read-only:** Never executes a tool on the scanned server.
- **Risk levels:** Per-server riskLevel with located findings.

## Good for

- Vet a new MCP server before adding it to Cursor or Claude
- Monitor dependencies for silent tool changes
- Audit your own server before publishing
- Gate a CI pipeline on the risk level

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http mcp-poisoning-scanner 'https://mcp.apify.com/?tools=bigochelo/mcp-poisoning-scanner'
```

### Claude Desktop (Settings → Connectors → Add custom connector)

```sh
https://mcp.apify.com/?tools=bigochelo/mcp-poisoning-scanner
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "mcp-poisoning-scanner": {
      "url": "https://mcp.apify.com/?tools=bigochelo/mcp-poisoning-scanner"
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "mcp-poisoning-scanner": {
      "type": "http",
      "url": "https://mcp.apify.com/?tools=bigochelo/mcp-poisoning-scanner"
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "mcp-poisoning-scanner": {
      "serverUrl": "https://mcp.apify.com/?tools=bigochelo/mcp-poisoning-scanner"
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "mcp-poisoning-scanner": {
      "type": "http",
      "url": "https://mcp.apify.com/?tools=bigochelo/mcp-poisoning-scanner"
    }
  }
}
```

## FAQ

### What does it cost?

About $0.01 per server scanned plus $0.003 per run; servers that cannot be scanned are not charged.

### Does it run my tools?

No, it only lists tools, prompts and resources.

### What can it miss?

Runtime behaviour that the server's metadata does not reveal.

### Can it scan local servers?

Yes, by pasting the tools/list result.

### Is MCP Tool Poisoning Scanner free?

No. MCP Tool Poisoning Scanner is a paid product.

### How do I add MCP Tool Poisoning Scanner to Claude Code?

Run this command in your terminal: claude mcp add --transport http mcp-poisoning-scanner 'https://mcp.apify.com/?tools=bigochelo/mcp-poisoning-scanner'

### What are the best alternatives to MCP Tool Poisoning Scanner?

The closest alternatives to MCP Tool Poisoning Scanner are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.

## Alternatives

- [REA](https://ai-powered.com/mcp/rea): REA is an open-source MCP server for developer tools.
- [Codebase Memory](https://ai-powered.com/mcp/codebase-memory): Codebase Memory is an open-source MCP server for knowledge and memory.
- [GitHub](https://ai-powered.com/mcp/github-2): GitHub is an open-source MCP server for developer tools.
- [Serena MCP: the IDE for your agent](https://ai-powered.com/mcp/serena-mcp-the-ide-for-your-agent): Serena MCP: the IDE for your agent is an open-source MCP server for developer tools.
- [XERJ](https://ai-powered.com/mcp/xerj): XERJ is an open-source MCP server for knowledge and memory.
- [Ouroboros](https://ai-powered.com/mcp/ouroboros): Ouroboros is an open-source MCP server for developer tools.
- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [Concord MCP](https://ai-powered.com/mcp/concord-mcp): Concord MCP is a freemium MCP server for developer tools.

---
Source: https://ai-powered.com/mcp/mcp-tool-poisoning-scanner
