
Npmguard
@AyoubTadlaoui
About Npmguard
Pre-install risk gate for npm packages. Stops AI coding agents from running malicious or typosquatted packages before lifecycle scripts run.
Config
Add this server to your MCP-compatible client using the configuration below.
{
"mcpServers": {
"npmguard": {
"command": "/usr/local/bin/npmguard-mcp"
}
}
}Tools
No tools detected
We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.
Overview
What is Npmguard?
Npmguard is a native pre-install risk gate for npm packages, packaged as an MCP tool for AI coding agents. It pulls data from the npm registry, OSV.dev (including the MAL-* malware namespace), and GitHub repo signals in parallel, computes a composite risk score, and returns an ok / warn / block verdict before lifecycle scripts can execute. The server is a single static Rust binary distributed via GitHub Releases (not npm) so the gate itself cannot be compromised by the npm supply chain it protects against.
How to use Npmguard?
Add Npmguard to your MCP host (Claude Code, Cursor, Windsurf) by adding a JSON entry with the path to the npmguard-mcp binary. It exposes one tool: install_package(name, version?) which returns { level, score, signals, recommendation }. AI agents can use the structured tool response to decide whether to proceed with an installation. The binary is installed by downloading a prebuilt release from GitHub Releases for your platform.
Key features of Npmguard
- Pre-install risk scoring with composite score and verdict
- Eight risk signals: LifecycleScripts, PackageAge, MaintainerChurn, SoleMaintainer, RepoHealth, Typosquat, KnownCve, Deprecated
- Integrates OSV.dev malware namespace (MAL-*) for known malicious packages
- Single static Rust binary distributed outside npm for supply chain security
- MCP tool returns structured signals for AI agent decision-making
- Prebuilt binaries for macOS, Linux, and Windows
Use cases of Npmguard
- AI coding agents evaluating npm packages before permitting installation
- Blocking typosquat attacks and known malware before lifecycle scripts execute
- Manual risk assessment via
npmguard install <package>(supports CLI)
FAQ from Npmguard
What does Npmguard do that npm audit or Snyk don't?
Npmguard is a pre-install risk gate that computes a composite score and verdict before lifecycle scripts execute, pulling from OSV.dev including the malware namespace. It is distributed as a Rust binary outside npm to avoid supply chain compromise, serving as an additional layer rather than a replacement for existing tools.
How do I install Npmguard?
Download the prebuilt binary for your platform (macOS x86_64/arm64, Linux x86_64/arm64, Windows x86_64) from the GitHub Releases page. The server is not distributed via npm.
What are the current limitations of Npmguard?
Npmguard is not yet a real npm install wrapper, does not sandbox lifecycle scripts, and does not verify npm provenance. It does not replace npm audit, Snyk, Socket, or Dependabot.
How does the MCP tool work in practice?
The tool install_package returns a structured response containing risk level, score, signals, and a recommendation. When an AI coding agent calls it, the recommendation is fed back as a tool response, allowing the model to stop and ask even if the user said to install without checking.
Frequently asked questions
What does Npmguard do that npm audit or Snyk don't?
Npmguard is a pre-install risk gate that computes a composite score and verdict *before* lifecycle scripts execute, pulling from OSV.dev including the malware namespace. It is distributed as a Rust binary outside npm to avoid supply chain compromise, serving as an additional layer rather than a replacement for existing tools.
How do I install Npmguard?
Download the prebuilt binary for your platform (macOS x86_64/arm64, Linux x86_64/arm64, Windows x86_64) from the GitHub Releases page. The server is not distributed via npm.
What are the current limitations of Npmguard?
Npmguard is not yet a real npm install wrapper, does not sandbox lifecycle scripts, and does not verify npm provenance. It does not replace npm audit, Snyk, Socket, or Dependabot.
How does the MCP tool work in practice?
The tool `install_package` returns a structured response containing risk level, score, signals, and a recommendation. When an AI coding agent calls it, the recommendation is fed back as a tool response, allowing the model to stop and ask even if the user said to install without checking.
Basic information
More Other MCP servers
ACI: Open-Source Infra to Power Unified MCP Servers
aipotheosis-labsACI.dev is the open source tool-calling platform that hooks up 600+ tools into any agentic IDE or custom AI agent through direct function calling or a unified MCP server. The birthplace of VibeOps.
Blender
ahujasidOpen-source MCP to use Blender with any LLM
Codelf
unbugA search tool helps dev to solve the naming things problem.

Peekaboo MCP – lightning-fast macOS screenshots for AI agents
steipetePeekaboo is a macOS CLI & optional MCP server that enables AI agents to capture screenshots of applications, or the entire system, with optional visual question answering through local or remote AI models.
Awesome Mlops
visengerA curated list of references for MLOps
Comments