audit-mcp-cli
A lightweight dependency vulnerability audit tool that works as both a CLI and an MCP Server — so your AI coding assistant can find and fix security issues for you
audit-mcp-cli
English | 中文
A lightweight dependency vulnerability audit tool for Node.js projects. Supports CLI and MCP Server modes, covers npm and pnpm projects, and generates structured Markdown/HTML reports with full dependency chains.
Features
- Full dependency chains — traces the complete path from your package.json to each vulnerable package
- npm + pnpm support — auto-detects package manager by lockfile
- Remote GitHub audit — audit any public or private repo without cloning
- MCP Server — integrates with AI coding assistants (Claude, Cursor, etc.)
- Markdown / HTML reports — clean, structured reports sorted by severity
- CI gate —
--fail-onexit code for CI/CD pipelines - Ignore mechanism — suppress accepted vulnerabilities with expiration dates
- Severity filtering — show only vulnerabilities above a threshold
Install
# Run directly
npx audit-mcp-cli
# Or install globally
npm install -g audit-mcp-cli
Requires Node.js >= 18.
Usage
# Audit current directory
audit-mcp-cli
# Specific project path
audit-mcp-cli --path /path/to/project
# Remote GitHub repo (branch)
audit-mcp-cli --remote github:facebook/react --ref main
# Remote GitHub repo (tag)
audit-mcp-cli --remote github:facebook/react --ref v18.2.0
# Remote GitHub repo (commit SHA)
audit-mcp-cli --remote github:facebook/react --ref abc123def
# HTML report
audit-mcp-cli --format html --output report.html
# CI: fail if high+ severity vulnerabilities found
audit-mcp-cli --fail-on high
# Severity filtering (only show high and critical)
audit-mcp-cli --severity high
CLI Options
| Option | Description | Default |
|---|---|---|
--path <path> | Local project path | process.cwd() |
--remote <repo> | Remote repo: github:owner/repo or https://github.com/owner/repo | — |
--ref <ref> | Git ref (branch name / tag / commit SHA) | main |
--token <token> | GitHub personal access token (for private repos) | GITHUB_TOKEN env |
--format <fmt> | Report format: md or html | md |
--output <path> | Output file path | audit-report.md or .html |
--severity <level> | Minimum severity to display: low / moderate / high / critical | low |
--fail-on <level> | CI fail threshold — exit 1 if vulnerabilities at this level or above exist | — |
--mcp | Start as MCP Server | — |
--lang <lang> | Language: en or zh-CN | Auto-detect from system |
--fail-on exit codes
| Value | Exits 1 when |
|---|---|
critical | Any critical vulnerability found |
high | Any high or critical found |
moderate | Any moderate, high, or critical found |
low | Any vulnerability found |
| (not set) | Always exits 0 |
MCP Server
Run as an MCP stdio server for AI assistants:
audit-mcp-cli --mcp
Claude Desktop
Basic (local projects & public repos):
{
"mcpServers": {
"audit-mcp-cli": {
"command": "npx",
"args": ["-y", "audit-mcp-cli", "--mcp"]
}
}
}
With GitHub token (private repos / avoid rate limits):
{
"mcpServers": {
"audit-mcp-cli": {
"command": "npx",
"args": ["-y", "audit-mcp-cli", "--mcp"],
"env": {
"GITHUB_TOKEN": "ghp_xxxx"
}
}
}
}
Cursor
Add to .cursor/mcp.json:
Basic (local projects & public repos):
{
"mcpServers": {
"audit-mcp-cli": {
"command": "npx",
"args": ["-y", "audit-mcp-cli", "--mcp"]
}
}
}
With GitHub token (private repos / avoid rate limits):
{
"mcpServers": {
"audit-mcp-cli": {
"command": "npx",
"args": ["-y", "audit-mcp-cli", "--mcp"],
"env": {
"GITHUB_TOKEN": "ghp_xxxx"
}
}
}
}
Tool: audit_dependencies
The MCP server exposes one tool that supports both local and remote auditing:
| Parameter | Description |
|---|---|
projectPath | Local project path |
remoteRepo | Remote repo: github:owner/repo |
ref | Git ref (branch / tag / SHA) |
token | GitHub token (for private repos, or use GITHUB_TOKEN env) |
format | md or html |
severity | Minimum severity filter |
outputPath | Custom output file path |
Returns: report file path + structured vulnerability details (CVSS, dependency chains, fix suggestions).
Token is optional. Local project auditing never requires a token. Remote public repos work without a token (60 requests/hour). Only private repos require a GitHub token.
Ignore Mechanism
Create .audit-mcp-cli-ignore.json in your project root to suppress accepted vulnerabilities:
{
"ignore": [
{
"packageName": "minimist",
"advisorySource": 1179,
"reason": "Accepted risk, limited impact in our usage",
"expiresAt": "2025-12-31T00:00:00Z"
}
]
}
packageName— match all advisories for this package, or combine withadvisorySourcefor exact matchexpiresAt— optional, ignore auto-expires after this date- Ignored vulnerabilities are shown in a separate section of the report and excluded from
--fail-onchecks
CI Integration
# GitHub Actions example
- name: Security Audit
run: npx audit-mcp-cli --fail-on high
# Generic CI
npx audit-mcp-cli --fail-on high && echo "pass" || echo "fail"
License
相關伺服器
Tripletex MCP Server
MCP server for Tripletex — Norwegian ERP and accounting platform. Invoices, customers, employees, projects, and ledger entries.
humanMCP
Personal MCP server for humans. Publish poetry, essays, listings with Ed25519 signatures. Store skills and memories. Bootstrap session unlocks expert team. Go stdlib only, zero deps.
Actiongate
Pre-execution safety layer for autonomous agent wallets — risk scoring, simulation, and policy enforcement via MCP, with free tier and pay-per-call pricing over x402
Currency Exchange & Crypto Rates
Real-time forex and crypto conversion with multi-source failover across 5 providers. 60+ fiat currencies, 30+ cryptocurrencies, no API keys needed.
GetHumanDesign
Calculate your human design chart and ask AI how you're designed to make decisions, build relationships, and find your path.
Send247 Human Logistics
Human Logistics in the AI Layer
OpenCRAVAT
Genetic variant annotation
inspirehep-mcp
InspireHEP MCP Server - integrate high-energy physics literature with LLMs.
Satellite MCP Server
Performs satellite orbital mechanics calculations using natural language, with a built-in world cities database for location lookup.
ReadyPermit
AI-powered property zoning, buildability, and ADU eligibility analysis for any U.S. address via 20+ government data sources.