SafeDep Vet MCP
平台与服务by safedep
为 AI agents 和 IDEs 提供防护,识别恶意开源软件包,降低依赖引入与开发流程中的安全风险。
给 AI agents 和 IDEs 加一道依赖安全闸,能提前识别恶意开源包,把风险拦在引入和开发流程之前。
什么是 SafeDep Vet MCP?
为 AI agents 和 IDEs 提供防护,识别恶意开源软件包,降低依赖引入与开发流程中的安全风险。
README
[!NOTE]
vetsupports special mode for Agent Skills. Runvet scan --agent-skill <owner/repo>to scan an Agent Skill hosted in a GitHub repository.
Why vet?
70-90% of modern software is open source code — how do you know it's safe?
Traditional SCA tools drown you in CVE noise. vet takes a different approach:
- Shadow AI discovery — Discover AI tool usage signals across various tools and configurations
- Catch malware before it ships — Zero-day detection through static and dynamic behavioral analysis (requires SafeDep Cloud access)
- Cut through vulnerability noise — Analyzes actual code usage to surface only the risks that matter
- Enforce policy as code — Express security, license, and quality requirements as CEL expressions
- CI/CD integration — Zero-config security guardrails in CI/CD
Free for open source. Hosted SaaS available at SafeDep.
Quick Start
Install in seconds:
# macOS & Linux
brew install safedep/tap/vet
# Using npm
npm install -g @safedep/vet
or download a pre-built binary
Get started immediately:
# Scan for malware in your dependencies
vet scan -D . --malware-query
# Fail CI on critical vulnerabilities
vet scan -D . --filter 'vulns.critical.exists(p, true)' --filter-fail
# Get API key for advanced malware detection
vet cloud quickstart
Architecture
vet follows a pipeline architecture: readers ingest package manifests from diverse sources (directories, repositories, container images, SBOMs), enrichers augment each package with vulnerability, malware, and scorecard data from SafeDep Cloud, the CEL policy engine evaluates security policies against enriched data, and reporters produce actionable output in formats like SARIF, JSON, and Markdown.
graph TB
subgraph "OSS Ecosystem"
R1[npm Registry]
R2[PyPI Registry]
R3[Maven Central]
R4[Other Registries]
end
subgraph "SafeDep Cloud"
M[Continuous Monitoring]
A[Real-time Code Analysis<br/>Malware Detection]
T[Threat Intelligence DB<br/>Vulnerabilities • Malware • Scorecard]
end
subgraph "vet CLI"
S[Source Repository<br/>Scanner]
P[CEL Policy Engine]
O[Reports & Actions<br/>SARIF/JSON/CSV]
end
R1 -->|New Packages| M
R2 -->|New Packages| M
R3 -->|New Packages| M
R4 -->|New Packages| M
M -->|Behavioral Analysis| A
A -->|Malware Signals| T
S -->|Query Package Info| T
T -->|Security Intelligence| S
S -->|Analysis Results| P
P -->|Policy Decisions| O
style M fill:#7CB9E8,stroke:#5A8DB8,color:#1a1a1a
style A fill:#E8A87C,stroke:#B88A5A,color:#1a1a1a
style T fill:#7CB9E8,stroke:#5A8DB8,color:#1a1a1a
style S fill:#90C695,stroke:#6B9870,color:#1a1a1a
style P fill:#E8C47C,stroke:#B89B5A,color:#1a1a1a
style O fill:#B8A3D4,stroke:#9478AA,color:#1a1a1a
Key Features
Malicious Package Detection
Real-time protection against malicious packages powered by SafeDep Cloud. Free for open source projects. Detects zero-day malware through active code analysis.
Vulnerability Analysis
Unlike dependency scanners that flood you with noise, vet analyzes your actual code usage to prioritize real risks.
See dependency usage evidence for details.
Policy as Code
Define security policies using CEL expressions to enforce context specific requirements:
# Block packages with critical CVEs
vet scan --filter 'vulns.critical.exists(p, true)' --filter-fail
# Enforce license compliance
vet scan --filter 'licenses.contains_license("GPL-3.0")' --filter-fail
# Require minimum OpenSSF Scorecard scores
vet scan --filter 'scorecard.scores.Maintained < 5' --filter-fail
Multi-Ecosystem Support
Package managers: npm, PyPI, Maven, Go, Ruby, Rust, PHP Container images: Docker, OCI SBOM formats: CycloneDX, SPDX Source repositories: GitHub, GitLab
Malicious Package Detection
Real-time protection against malicious packages with active scanning and behavioral analysis.
Quick Setup
# One-time setup for advanced scanning
vet cloud quickstart
# Scan for malware with active scanning (requires API key)
vet scan -D . --malware
# Query known malicious packages (no API key needed)
vet scan -D . --malware-query
Example detections:
- MAL-2025-3541: express-cookie-parser
- MAL-2025-4339: eslint-config-airbnb-compat
- MAL-2025-4029: ts-runtime-compat-check
Key security features:
- Real-time analysis against known malware databases
- Behavioral analysis using static and dynamic analysis
- Zero-day protection through active code scanning
- Human-in-the-loop triaging for high-impact findings
- Public analysis log for transparency
Advanced Usage
# Specialized scans
vet scan --vsx --malware # VS Code extensions
vet scan -D .github/workflows --malware # GitHub Actions
vet scan --image nats:2.10 --malware # Container images
# Analyze specific packages
vet inspect malware --purl pkg:npm/nyc-config@10.0.0
Production Ready Integrations
GitHub Actions
Zero-config security guardrails in CI/CD:
- uses: safedep/vet-action@v1
with:
policy: ".github/vet/policy.yml"
See vet-action documentation.
GitLab CI
Enterprise scanning with vet CI Component:
include:
- component: gitlab.com/safedep/ci-components/vet/scan@main
Container Integration
Run vet anywhere using our container image:
docker run --rm -v $(pwd):/app ghcr.io/safedep/vet:latest scan -D /app --malware
Installation
Homebrew (Recommended)
brew install safedep/tap/vet
npm
npm install @safedep/vet
Direct Download
See releases for pre-built binaries.
Go Install
go install github.com/safedep/vet@latest
Container Image
# Quick test
docker run --rm ghcr.io/safedep/vet:latest version
# Scan local directory
docker run --rm -v $(pwd):/workspace ghcr.io/safedep/vet:latest scan -D /workspace
Verify Installation
vet version
# Should display version and build information
Advanced Features
Learn more in our comprehensive documentation:
- AI Usage Discovery - Discover AI tool usage signals across various tools and configurations
- AI Agent Mode - Run vet as an AI agent
- MCP Server - Run vet as an MCP server for AI-assisted code analysis
- Reporting - SARIF, JSON, CSV, HTML, Markdown formats
- SBOM Support - CycloneDX, SPDX import/export
- Query Mode - Scan once, analyze multiple times
- GitHub Integration - Repository and organization scanning
- GitHub Actions Pinning - Pin GitHub Actions to commit SHAs to prevent supply chain attacks
Privacy
vet collects anonymous usage telemetry to improve the product. Your code and package information is never transmitted.
# Disable telemetry (optional)
export VET_DISABLE_TELEMETRY=true
Community & Support
<div align="center">Join the Community
</div>Get Help & Share Ideas
- Interactive Tutorial - Learn vet hands-on
- Complete Documentation - Comprehensive guides
- Discord Community - Real-time support
- Issue Tracker - Bug reports & feature requests
- Contributing Guide - Join the development
<div align="center">
Star History
Built With Open Source
vet stands on the shoulders of giants:
OSV • OpenSSF Scorecard • SLSA • OSV-SCALIBR • Syft
<p><strong>Secure your supply chain today. Star the repo and get started!</strong></p>
Created with love by SafeDep and the open source community
</div> <img referrerpolicy="no-referrer-when-downgrade" src="https://static.scarf.sh/a.png?x-pxid=304d1856-fcb3-4166-bfbf-b3e40d0f1e3b" />常见问题
SafeDep Vet MCP 是什么?
为 AI agents 和 IDEs 提供防护,识别恶意开源软件包,降低依赖引入与开发流程中的安全风险。
相关 Skills
Slack动图
by anthropics
面向Slack的动图制作Skill,内置emoji/消息GIF的尺寸、帧率和色彩约束、校验与优化流程,适合把创意或上传图片快速做成可直接发送的Slack动画。
✎ 帮你快速做出适配 Slack 的动图,内置约束规则和校验工具,少踩上传与播放坑,做表情包和演示都更省心。
MCP构建
by anthropics
聚焦高质量 MCP Server 开发,覆盖协议研究、工具设计、错误处理与传输选型,适合用 FastMCP 或 MCP SDK 对接外部 API、封装服务能力。
✎ 想让 LLM 稳定调用外部 API,就用 MCP构建:从 Python 到 Node 都有成熟指引,帮你更快做出高质量 MCP 服务器。
接口测试套件
by alirezarezvani
扫描 Next.js、Express、FastAPI、Django REST 的 API 路由,自动生成覆盖鉴权、参数校验、错误码、分页、上传与限流场景的 Vitest 或 Pytest 测试套件。
✎ 帮你把API与集成测试自动化跑顺,减少回归漏测;能力全面,尤其适合复杂接口场景的QA团队。
相关 MCP Server
Slack 消息
编辑精选by Anthropic
Slack 是让 AI 助手直接读写你的 Slack 频道和消息的 MCP 服务器。
✎ 这个服务器解决了团队协作中需要 AI 实时获取 Slack 信息的痛点,特别适合开发团队让 Claude 帮忙汇总频道讨论或发送通知。不过,它目前只是参考实现,文档有限,不建议在生产环境直接使用——更适合开发者学习 MCP 如何集成第三方服务。
by netdata
io.github.netdata/mcp-server 是让 AI 助手实时监控服务器指标和日志的 MCP 服务器。
✎ 这个工具解决了运维人员需要手动检查系统状态的痛点,最适合 DevOps 团队让 Claude 自动分析性能数据。不过,它依赖 NetData 的现有部署,如果你没用过这个监控平台,得先花时间配置。
by d4vinci
Scrapling MCP Server 是专为现代网页设计的智能爬虫工具,支持绕过 Cloudflare 等反爬机制。
✎ 这个工具解决了爬取动态网页和反爬网站时的头疼问题,特别适合需要批量采集电商价格或新闻数据的开发者。不过,它依赖外部浏览器引擎,资源消耗较大,不适合轻量级任务。