mcp-warden

Zavelinski/mcp-warden

★ 0 stars Unknown language AI/LLM Updated 14d ago

Vet an MCP server before enabling it. Produces a security & compatibility sheet, statically scans the server as hostile data, and returns an ALLOW / REVIEW / BLOCK verdict. The MCP sibling of skill-security-scan.

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": {
    "mcp-warden": {
      "command": "npx",
      "args": [
        "-y",
        "Zavelinski/mcp-warden"
      ]
    }
  }
}

README Excerpt

Vet a third-party MCP server **before** you enable it in Claude Code. Produces a security & compatibility sheet, statically scans the server as hostile data, and returns an ALLOW / REVIEW / BLOCK verdict with file:line evidence. The MCP sibling of `skill-security-scan`. An MCP server is untrusted code with network access and tool privileges: it can read your data, hit the network, and mutate state. Tool misuse / malformed tool calls are ~31% of agent production failures, and the community has be

mcp-warden

Quick Install

README Excerpt

Related AI/LLM Servers