v1.8.2 · Advisory Database · 11 Tools · Zero Dependencies ·

Your AI agent's
security guardrail

An MCP server that guards your codebase from three threats: compromised npm packages, insecure dependencies, and debris left by AI coding agents. Backed by a curated advisory database of known attacks. Every package is verified before install, every coding session is reviewed, every unused dependency is flagged.

Get Started View on GitHub

// Add depguard to any MCP-compatible AI agent
$ claude mcp add depguard -- npx -y depguard-cli --mcp

// 1. AI agent wants to install a package
Agent    "I need RTSP streaming support"
depguard guard — go2rtc (score 72/100, allow)

// 2. AI agent finishes coding
depguard review — Found 3 console.logs, 1 empty catch, 1 TODO
Agent    Fixing 5 issues before committing...

// 3. Cleanup unused packages
depguard sweep — lodash unused (~1.4 MB savings)

The Problem

Your AI agent installs blindly and leaves debris

AI coding agents create two problems: they install packages without checking security, and they leave garbage in your code (console.logs, empty catch blocks, broken imports, orphan files). Without depguard, every npm install is a blind trust decision and every coding session adds invisible debt.

2,200+

Malicious packages detected on npm in 2025

8M+

Weekly downloads compromised by event-stream attack

100x

Faster package installs with AI agents vs manual review

18+

Malware patterns depguard detects in package source code

99%

AI tokens saved per audit vs manual web search and reasoning

How It Works

How MCP integration works

One command connects depguard to your AI agent. From that moment, every install is verified and every coding session is reviewed.

Guard

Before every npm install — whether the user asked or the AI decided on its own — depguard verifies the package exists, checks for typosquatting, and runs a security audit.

depguard_guard("express")

Review

After coding, depguard scans the source files for debris the AI left behind: console.logs, empty catch blocks, broken imports, orphan files. The AI agent fixes them before reporting done.

depguard_review(".", "quick")

Clean Up

At the end of a session, depguard finds unused packages that were installed but never imported. The AI agent removes them, keeping the project lean.

depguard_sweep(".")

Try It

See depguard in action

This is a limited demo with pre-loaded data for 6 popular packages. For live audits on any package, install depguard-cli locally.

AVAILABLE PACKAGES: express, lodash, react, axios, chalk, moment

// Select a package above or type one to see the audit report
$ depguard-cli audit ___

Get Started

One command. Infinite peace of mind.

No API keys. No accounts. No configuration. Your AI agent gets security superpowers in 10 seconds.

$ claude mcp add depguard -- npx -y depguard-cli --mcp

Setup Guide Star on GitHub npm

Getting Started

Add depguard to your AI agent

One command. Zero configuration. Your AI agent starts checking every dependency immediately.

Claude CodeMCP Ready

Claude DesktopMCP Ready

CursorMCP Ready

WindsurfMCP Ready

Continue.devMCP Ready

ClineMCP Ready

Roo CodeMCP Ready

Any MCP ClientJSON-RPC 2.0

// Claude Code — one command
$ claude mcp add depguard -- npx -y depguard-cli --mcp

// Claude Desktop, Cursor, Windsurf, Continue.dev, Cline, Roo Code
// Add to your MCP config file:
{
  "mcpServers": {
    "depguard": {
      "command": "npx",
      "args": ["-y", "depguard-cli", "--mcp"]
    }
  }
}

// That's it. Your agent now has 10 security tools.
// No API keys. No accounts. No configuration.

MCP Tools

11 tools your AI agent calls automatically

Organized by workflow: before installing, after coding, project health, and quick lookups. The AI agent knows when to call each one.

depguard_guard

Called before every npm install, whether the user asked or the AI decided. Verifies the package exists, detects typosquatting, runs security audit, returns allow/warn/block.

Before Install

depguard_should_use

When the AI needs functionality ("RTSP streaming", "date formatting"), it calls this before choosing a package. Checks native Node.js alternatives first, then scores npm candidates.

Before Install

depguard_review

Called after code changes or before commit. Finds console.logs, empty catch blocks, broken imports, orphan files, abandoned TODOs. The AI fixes them before reporting done.

After Coding

depguard_sweep

Finds packages installed but never imported. Config-aware, workspace-aware, peer-dep-aware. Also detects phantom dependencies in node_modules.

After Coding

depguard_audit

Deep security audit of a single package. Downloads tarball, scans source code for 18+ malware patterns, checks dual advisory databases, analyzes install scripts.

Project Health

depguard_audit_project

Audit all dependencies from a package.json in one call. Auto-detects project license. Use after cloning a repo or reviewing project security.

Project Health

depguard_audit_deep

Crawls the full transitive dependency tree. Aggregates vulnerabilities across all nested packages. Shows circular dependencies and total attack surface.

Project Health

depguard_score

Quick 0-100 quality score. Faster than full audit when you only need the number. Critical vulns cap at 30, high at 50.

Quick Lookup

depguard_verify

Quick existence check + typosquatting detection. Faster than guard when you only need to verify a package name exists.

Quick Lookup

depguard_search

Search npm by keywords, sorted by depguard quality score. Find packages without web searches.

Quick Lookup

depguard_audit_bulk

Audit multiple packages in one call. Use audit_project instead if you have a package.json path.

Quick Lookup

Scoring Model

Transparent, reproducible scores

Every package scored 0–100 across five weighted dimensions. CVSS scores used when available. Hard security ceiling: critical vulns cap total at 30, high vulns at 50. Code analysis findings further reduce the score.

Security

30%

Maintenance

25%

Popularity

20%

License

15%

Dependencies

10%

≥ 60

Install

Safe to use. Meets quality standards.

40 – 59

Caution

Review before using. Potential risks.

< 40

Write Your Own

Better to implement the functionality yourself.

Smart Advisor

Skip the package. Use Node.js.

depguard checks 20+ native alternatives before recommending npm packages. Less bloat, smaller attack surface.

npm install axios

globalThis.fetch()

Built-in HTTP client. Full Request/Response API.

Node 18+

npm install uuid

crypto.randomUUID()

RFC 4122 v4 UUID generation. Cryptographically secure.

Node 19+

npm install lodash.clonedeep

structuredClone()

Deep clone any serializable value. Handles circular refs.

Node 17+

npm install dotenv

process.loadEnvFile()

Load .env files natively. No parsing library needed.

Node 21.7+

npm install glob

fs.glob()

Native file globbing. Async iterator interface.

Node 22+

npm install jest

node:test

Built-in test runner with describe, it, and assertions.

Node 18+

npm install yargs

util.parseArgs()

Parse CLI arguments with type validation. Built-in.

Node 18.3+

npm install better-sqlite3

node:sqlite

Embedded SQLite database. Synchronous API.

Node 22.5+

What You Get

depguard vs doing it manually

Without depguard, you need to check multiple sources, analyze scripts by hand, and make gut-feel decisions. Here's what one command replaces.

Task	depguard	Manual Approach
Check known CVEs	✓ Automated	Search npm + GitHub advisories manually
GitHub Advisory Database	✓ Included	Visit github.com/advisories, search, cross-reference
Static code analysis	✓ 18+ patterns	Download tarball, read source code manually
Inspect install scripts	✓ 20+ patterns	Read package source code line by line
Behavioral mismatch	✓ Automated	Compare description vs code behavior yourself
Check maintenance health	✓ Scored	Check npm page, GitHub commits, last publish date
Verify license compatibility	✓ 25+ SPDX	Read LICENSE file, understand copyleft hierarchy
Evaluate package quality	✓ 0–100	Subjective judgement, no consistent scoring
Find native alternatives	✓ 20+ mapped	Know Node.js APIs by heart or search docs
AI agent integration	✓ 10 MCP tools	Not possible — agents install blindly
Time per package	~2 seconds	15–30 minutes of research

New in v1.6.0

Static Code Analysis

depguard now downloads the package tarball and scans the actual source code for malware patterns, obfuscation, data exfiltration, and behavioral mismatches. Every finding includes a detailed explanation suitable for developers.

Malware Detection

Eval of decoded payloads, reverse shells via net.connect, cryptocurrency mining pool URLs (stratum+tcp). Critical severity — auto-blocks.

Critical

Data Exfiltration

JSON.stringify(process.env), Object.keys(process.env), fetch with dynamic URLs from env vars, bracket-notation env access to evade scanners.

Critical

Code Execution

eval(), new Function(), child_process.exec/spawn, shell interpreter spawning. Flags dangerous patterns with context.

High

Obfuscation

Long hex/unicode-encoded strings, base64 payloads, minified source in non-.min.js files. Techniques used to hide malicious code from reviewers.

High

Behavioral Mismatch

A "string formatter" that makes network calls? A "date utility" that reads the filesystem? depguard compares what a package says it does vs what it actually does.

Unique

Rich Explanations

Every finding includes: severity, category, title, a detailed explanation of why it's dangerous, the exact evidence from source code, and a recommendation.

Developer-Friendly

// What your AI agent receives for a suspicious package:

{
  "severity":       "critical",
  "category":       "data-exfiltration",
  "title":          "Serialization of entire environment",
  "explanation":    "This package serializes your ENTIRE
                     environment into a JSON string. Your API keys,
                     database passwords, JWT secrets would be exposed.",
  "evidence":       "const d = JSON.stringify(process.env);",
  "file":           "src/index.js",
  "recommendation": "Do NOT install this package."
}

Install Scripts

Install script analysis

depguard statically analyzes install scripts (preinstall, install, postinstall) for 20+ suspicious patterns. Each finding now includes a rich explanation and recommendation.

Pattern	Severity	What It Detects
Remote code execution	Critical	curl/wget piped to sh, download-and-execute chains
Credential file access	Critical	Reading ~/.ssh, ~/.aws, ~/.npmrc, ~/.gnupg, /etc/passwd
Sensitive env vars	Critical	Accessing $NPM_TOKEN, $AWS_SECRET, $GITHUB_TOKEN
Eval of decoded content	Critical	eval(atob(...)), eval(Buffer.from(..., "base64"))
Reverse shells	Critical	/dev/tcp connections, netcat with -e flag
Shell typosquatting	Critical	/bin/ssh instead of /bin/sh
Process spawning	High	child_process, execSync, spawn
Dynamic code execution	High	eval(), exec() in install scripts
Obfuscated strings	High	Hex-encoded strings, base64 payloads
Network APIs	High	net.connect, dgram, dns.resolve in install scripts

Real Threats

These attacks happened

Major npm supply chain incidents and how depguard detects the patterns that made them possible.

Package	Year	Severity	What Happened	depguard Detection
event-stream	2018	Critical	Malicious dependency stole cryptocurrency wallet keys. 8M weekly downloads compromised.	Code analysis flags obfuscated code + credential access patterns. Script analysis catches eval of decoded content.
ua-parser-js	2021	Critical	Hijacked account published cryptominers and password stealers via postinstall.	Code analysis detects stratum+tcp mining URLs. Install script analysis catches process spawning and credential file access.
node-ipc	2022	Critical	Maintainer added code to wipe files based on geolocation. Dependency of vue-cli.	Behavioral mismatch: IPC library making network calls + filesystem writes. Advisory database flags GHSA.
colors + faker	2022	High	Maintainer sabotaged packages with infinite loops. Broke thousands of projects.	Maintenance score drops. Advisory database flags incident. Score falls below threshold.
eslint-scope	2018	Critical	Compromised token published version stealing npm tokens from ~/.npmrc.	Code analysis detects credential file reads. Script analysis flags env var exfiltration.
coa + rc	2021	Critical	Hijacked accounts. Obfuscated malware in preinstall scripts.	Code analysis detects base64 + eval pattern. Script analysis flags obfuscated code + process spawning.
crossenv (typosquat)	Ongoing	High	Names mimicking popular packages. Steal environment variables on install.	Typosquatting detection via Levenshtein distance. Code analysis flags JSON.stringify(process.env).

v1.6.0 — Static Code Analysis

Beyond CVEs: analyzing the actual code

depguard now downloads package tarballs and scans source code for malware patterns, obfuscation, and behavioral mismatches. Every finding includes a rich, human-readable explanation.

Tarball Analysis

Downloads the npm package tarball, extracts JS files, and scans with 18+ pattern rules across 6 threat categories: malware, data-exfiltration, code-execution, obfuscation, unexpected-behavior, supply-chain.

[CRITICAL] Serialization of entire environment
JSON.stringify(process.env) detected in src/index.js

Behavioral Mismatch

Compares the package description and keywords against detected code behavior. A "string formatter" making network calls? Flagged as unexpected-behavior with explanation.

[HIGH] Network activity inconsistent
Package "date formatter" makes HTTP requests

SecurityFinding Format

Every finding returns: severity, category, title, detailed explanation, evidence (exact code), file path, and recommendation. AI agents can present these directly to developers.

212 tests · 0 failures
28 new tests for code analysis

v1.6.0 Detection Categories

malware
Eval+decode, reverse shells, crypto mining, TCP backdoors

data-exfiltration
Env serialization, credential reads, dynamic fetch URLs

code-execution
eval(), new Function(), child_process, shell spawn

obfuscation
Hex/unicode strings, base64 payloads, minified source

unexpected-behavior
Network/FS access mismatched with package purpose

supply-chain
Typosquatting patterns in install scripts

The Complete Workflow

Guard. Audit. Review. Clean up.

depguard is the only MCP tool that covers the full AI coding lifecycle: verify packages before install, audit for vulnerabilities, review the code the AI generated, and clean up unused dependencies.

1. Guard

depguard_guard — Before every npm install, whether the user asked or the AI decided. Blocks hallucinated packages, detects typosquatting, runs security audit.

[BLOCK] ai-made-up-pkg
Package does NOT exist on npm!

2. Audit

depguard_audit — Deep security analysis. Downloads the tarball, scans source code for malware, checks dual advisory databases, analyzes install scripts.

[WARN] expresss
Possible typosquat of: express

3. Review

depguard_review — After coding, scans source files for debris the AI left behind. Console.logs, empty catch blocks, broken imports, orphan files, abandoned TODOs. The AI fixes them before reporting done.

Found: 3 console.logs, 1 empty catch
Fixing before commit...

4. Clean Up

depguard_sweep — Finds packages that were installed but never imported. Config-aware (eslint, typescript, jest...), workspace-aware, with estimated size savings.

Unused: lodash (~1.4 MB)
Savings: ~2.4 MB from 3 packages

v1.8 Architecture Redesign

Advisory Database
Curated database of known compromised packages. event-stream, colors, faker auto-blocked with full incident history.

Comment Stripping
Code analysis strips comments before scanning. URLs in JSDoc and license headers no longer flagged.

Informational Findings
Code analysis findings are informational. Only CVEs and advisory DB affect the security score.

Lock File Parsing
Reads package-lock.json and pnpm-lock.yaml. Transitive deps no longer flagged as phantom.

CVSS + Security Ceiling
Critical vulns cap at 30. High at 50. Compromised packages score 0. Non-negotiable.

Intl Alternatives
DateTimeFormat, NumberFormat, RelativeTimeFormat, Collator detected as native alternatives.

Real Impact: v1.8 Score Accuracy

Compromised packages now blocked

event-stream allow (75) → block (0)
colors allow (88) → score 0
faker allow → score 0

False positives eliminated

eslint 50 → 94
vue 50 → 94
fastify 30 → 93
zod 30 → 96
morgan 50 → 92

Your AI agent'ssecurity guardrail

Your AI agent installs blindly and leaves debris

How MCP integration works

Guard

Review

Clean Up

See depguard in action

One command. Infinite peace of mind.

Add depguard to your AI agent

11 tools your AI agent calls automatically

depguard_guard

depguard_should_use

depguard_review

depguard_sweep

depguard_audit

depguard_audit_project

depguard_audit_deep

depguard_score

depguard_verify

depguard_search

depguard_audit_bulk

Transparent, reproducible scores

Skip the package. Use Node.js.

depguard vs doing it manually

Static Code Analysis

Malware Detection

Data Exfiltration

Code Execution

Obfuscation

Behavioral Mismatch

Rich Explanations

Install script analysis

These attacks happened

Beyond CVEs: analyzing the actual code

Tarball Analysis

Behavioral Mismatch

SecurityFinding Format

v1.6.0 Detection Categories

Guard. Audit. Review. Clean up.

1. Guard

2. Audit

3. Review

4. Clean Up

v1.8 Architecture Redesign

Real Impact: v1.8 Score Accuracy

Get in touch

Your AI agent's
security guardrail