๐Ÿšจ Nov 2026 Alert 61,764 GitHub repos probed & 771 poisoned npm postinstall hooks hunting Claude, Cursor & Cloud keys. Scan Any Repo Free โ†“ ยท Read Threat Advisory โ†’
๐Ÿ›ก๏ธ Free Repo Scanner Features Pricing Blog Dashboard
PROTECTED Zero-Day Defense for Nov 2026 Supply-Chain & AI Key Attacks โ†’

ClaudeArmor Security

Zero-Trust Workstation Shield for AI Coding Agents

Stop poisoned npm postinstall scripts, rogue MCP tool calls, and silent exfiltration of your Claude, Cursor, OpenAI, and Cloud credentials before malicious code touches your workstation.

20+
AI Agents Guarded
Zero-Day
Supply-Chain Shield
AES-256
Quarantine Vault
claudearmor โ€” live workstation shield
$ npm install && claudearmor-proxy --guard-agents all
โœ“ ClaudeArmor Zero-Trust Shield Active (20 AI Agents)
๐Ÿšจ BLOCKED: package.json[postinstall] โ†’ bash ./.ci-8f4a21c.sh
๐Ÿ›ก๏ธ BLOCKED: Outbound exfil of ANTHROPIC_API_KEY, ~/.aws, ~/.ssh
โš  QUARANTINED: Hidden dropper isolated in AES-256 Vault
โœ“ Workstation, .env secrets & AI keys remain 100% intact
๐Ÿ›ก๏ธ Free Instant Supply-Chain Diagnostic

Is That GitHub Repo Safe to Clone & npm install?

In the November 2026 supply-chain campaign, attackers planted hidden .ci-*.sh secret-stealers inside 771 package.json install hooks and 8,093 repositories to harvest Claude, Cursor, OpenAI, and Cloud keys. Paste any public GitHub repo below to audit it before cloning.

Try an instant test:
โšก How It Works
Three Steps to Full Protection
From install to protection in under a minute.
1

Install

One command via npm, brew, or download the desktop app. Zero config required.

2

Scan

8 engines scan your files, keys, scripts, MCP tools, and repo history simultaneously.

3

Protect

Threats quarantined in AES-256 vault. AI generates plain-English reports with fixes.

๐Ÿ›ก๏ธ Features
8 Security Engines, One App
Each engine specializes in a different attack surface. Together they cover everything AI agents can break.
๐Ÿ” Engine 1

Deep Secret & Credential Guard

Detects exposed API keys, tokens, and secrets across your entire project using 40+ proprietary signature heuristics with deep entropy analysis.

  • AWS, Azure, GCP, Stripe, OpenAI, and 35+ more providers
  • Information-theoretic entropy scoring to catch obfuscated secrets
  • Context-aware filtering skips non-production mock fixtures automatically
keyleak scan
โš  FOUND: AWS_SECRET_ACCESS_KEY File: src/config.js Line: 14 Entropy: 4.82 (high) Signature: CLOUD_IAM_CREDENTIAL_V4 โš  FOUND: STRIPE_SECRET_KEY File: .env.local Line: 3 Entropy: 5.11 (high) โœ“ 847 files scanned ยท 2 secrets found
๐Ÿ‘๏ธ Engine 2

Real-Time Protection

Monitors your workspace I/O stream in real time. Auto-scans new and modified files the millisecond AI agents write them.

  • Monitors Downloads, Desktop, project directories
  • Sub-millisecond stream telemetry prevents duplicate alerts
  • Self-healing watchdog guarantees continuous enforcement
real-time
โ— Watching 3 directories... 15:42:01 โœ“ src/app.js modified โ€” clean 15:42:03 โœ“ package.json modified โ€” clean 15:42:07 โœ— THREAT install.sh created โ†’ Obfuscated base64 payload detected โ†’ Auto-quarantined to vault 15:42:08 โœ“ README.md modified โ€” clean
๐Ÿ›๏ธ Engine 3

Quarantine Vault

Isolate threats with military-grade AES-256 encryption and a hardware-bound enclave key. One-click restore for verified files.

  • AES-256-GCM hardware-backed encryption
  • Original path, cryptographic hash, and forensic metadata preserved
  • Automated vault lifecycle management
quarantine vault
Vault: 3 items ยท 2.4 MB encrypted โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€ 1. install.sh โ€” quarantined 2m ago Reason: Obfuscated dynamic execution SHA-256: a1b2c3...f9e8d7 2. payload.js โ€” quarantined 1h ago Reason: Neural Heuristic match: trojan_gen 3. config.bak โ€” quarantined 3h ago Reason: Exposed API key
๐Ÿง  Engine 4

AI Forensic Security Reports

ClaudeArmor's proprietary Forensic AI Engine analyzes scan telemetry and generates executive reports with risk scores, remediation solutions, and priority rankings.

  • Risk severity scoring: Critical / High / Medium / Low
  • Step-by-step remediation solutions & patches
  • Exportable compliance & executive HTML certificates
ai report
โ•โ•โ• AI Security Report โ•โ•โ• Overall Risk: HIGH (7.2/10) CRITICAL: AWS key in config.js โ†’ Rotate key immediately via IAM โ†’ Move to .env + add to .gitignore MEDIUM: eval() in install.sh โ†’ Replace with explicit import โ†’ Add CSP header to prevent XSS Action items generated: 4
๐Ÿ” Engine 5

Tool & Script Scanner

Static analysis of shell scripts and automation files. Catches obfuscated commands, data exfiltration, and privilege escalation.

  • Detects base64-encoded payloads and eval chains
  • Flags curl piped to shell (curl | sh)
  • Checks npm postinstall scripts for malice
toolscan
Scanning 12 scripts... โœ— postinstall.sh Line 7: eval(atob("bWFsaWNpb3Vz")) Risk: Critical โ€” obfuscated eval โš  setup.sh Line 3: curl -s http://... | sh Risk: High โ€” pipe to shell โœ“ 10 scripts clean
๐Ÿ”Œ Engine 6

MCP Manifest Scanner

Scores MCP tool manifests for risk. Checks for overly broad permissions, suspicious endpoints, and unsafe tool configurations.

  • Parses JSON manifests for dangerous permission grants
  • Flags tools requesting filesystem or network access
  • Risk score with detailed breakdown per tool
mcpscan
Scanning 3 MCP manifests... โœ“ web-search.json โ€” Score: Low risk โš  fs-access.json โ€” Score: Medium โ†’ Requests write access to /home โœ— shell-exec.json โ€” Score: Critical โ†’ Allows arbitrary command execution โ†’ No sandbox constraints specified
๐Ÿ™ Engine 7

GitHub Repo Scanner

Audit any GitHub repository before cloning. Scans workflows, commit history, and dependencies for supply-chain risks.

  • Checks GitHub Actions for secret exfiltration
  • Analyzes dependency tree for known CVEs
  • Flags repos with suspicious recent force-pushes
reposcan
$ claudearmor repo scan user/project โœ“ 23 workflows analyzed โœ“ 156 commits checked โš  .github/workflows/build.yml โ†’ Uploads secrets to external URL โœ“ Dependencies: 0 known CVEs โœ“ No suspicious force-pushes Overall: Moderate risk (1 issue)
๐Ÿ“ฆ Engine 8

Zero-Trust Micro-Sandbox

Execute suspicious files in a proprietary isolated micro-sandbox with strict network and filesystem boundaries for safe behavioral detonation.

  • Kernel-enforced process & memory isolation
  • Strict virtualized boundary confinement
  • Captures runtime behavior without host system risk
sandbox
Sandbox initialized Flags: --no-remote --disable-extensions Paths: Restricted to ~/tmp, ~/home โœ“ Network access: blocked โœ“ File system: restricted โœ“ Process isolation: active Running suspicious.js... โœ— Attempted network call to evil.com โœ— Attempted write to C:\Windows\ Result: Blocked ยท File quarantined
0
Files Scanned Daily
8
Security Engines
40+
Secret Patterns
AES-256
Quarantine Encryption
โš–๏ธ Compare
Why ClaudeArmor?
Purpose-built for developers using AI coding agents.
FeatureClaudeArmorTraditional AVManual Review
API Key Detectionโœ“ 40+ patternsโœ—Slow & error-prone
AI Agent Monitoringโœ“ Real-timeโœ—โœ—
MCP Tool Scanningโœ“ Built-inโœ—โœ—
GitHub Repo Auditโœ“ Pre-cloneโœ—โœ“ Manual
AI Security Reportsโœ“ Forensic AIโœ—โœ—
Encrypted Quarantineโœ“ AES-256โœ“ Basicโœ—
Developer-Friendlyโœ“ Built for devsGenericโœ“
โ Testimonials
What Developers Are Saying
โ˜…โ˜…โ˜…โ˜…โ˜…

"ClaudeArmor caught three hardcoded API keys in our repo that we'd missed for months. The AI report explained exactly why they were dangerous and how to rotate them."

AK
Amir Khalil
Security Lead, FinTech Startup
โ˜…โ˜…โ˜…โ˜…โ˜…

"The real-time file watcher is incredible. It flagged a suspicious npm postinstall script within seconds of npm install. Saved us from a supply-chain attack."

SL
Sarah Liu
Full-Stack Engineer, Vercel
โ˜…โ˜…โ˜…โ˜…โ˜…

"Finally an antivirus that understands developer workflows. ClaudeArmor doesn't just find threats โ€” it explains them in plain English and tells you how to fix them."

MR
Marcus Reyes
DevOps Engineer, Shopify
โšก Quick Start
Up and Running in Seconds
$ npx claudearmor@latest
$ brew install claudearmor
$ claudearmor scan .
# Download from Official Security Portal
$ curl -sSL https://claudearmorsec.com/download/installer -o ClaudeArmor-Setup.exe

Works on macOS, Windows & Linux. The installer handles Node.js and everything else for you.

โŸฉ Works With
Integrates With Your Stack
๐ŸŸฃ VS Code๐Ÿ™ GitHub๐ŸฆŠ GitLab๐Ÿ”ต Bitbucket๐Ÿ“ฆ npm๐Ÿ pip๐Ÿค– Claude Code๐Ÿง  Cursor๐Ÿ’Ž Windsurfโšก Copilot
โŸฉ Stay in the Loop

๐Ÿ›ก๏ธ Get Security Updates

Get notified about new features, threat patterns, and ClaudeArmor releases. No spam, unsubscribe anytime.