Claude Code Skills·Claude Skills·The open SKILL.md registry for Claude
ClaudSkillsArcadeMCP Servers › mcp-safeguard

mcp-safeguard

Security scanner for MCP servers — detect prompt injection, credential leaks, exposed endpoints, and tool poisoning

MCP · uvx MIT

Install

MCP server — add to your Claude Code config:

{
  "mcpServers": {
    "mcp-safeguard": {
      "command": "uvx",
      "args": [
        "mcp-safeguard"
      ]
    }
  }
}

Or run directly:

uvx mcp-safeguard

Security & trust

Local (stdio) MIT

Runs on your machine via uvx — it executes with your shell environment and credentials. Review the source before installing.

This is a structural risk profile — how the server runs and where it's published — not a static code scan (the catalog doesn't hold the server's source). How we think about agent security ›

Details

Websitehttps://pypi.org/project/mcp-safeguard/
LicenseMIT
Pricingfree
Tagsmcp security ai-security prompt-injection model-context-protocol llm-security ai-agents claude
Sourcespypi
Discovered2026-06-05

More in MCP Servers

See all MCP Servers →

ClaudSkills Arcade · All categories · Catalog JSON · CC BY 4.0 · independent of Anthropic PBC