Critical n8n Vulnerability: CVSS 10.0 RCE Threatens Cloud Workflow Platforms

What Happened

CVE-2026-21858 is a content-type confusion vulnerability in n8n's webhook handling mechanism that enables complete server compromise without authentication.

The Vulnerability at a Glance

• CVE ID: CVE-2026-21858
• CVSS Score: 10.0 (Critical)
• Attack Vector: Network - No authentication required
• Affected Versions: n8n 1.65.0 through 1.120.4
• Patched Version: 1.121.0 and above
• Discovered by: Multiple security researchers (Orca Security, Upwind, Horizon3.ai)

Why CVSS 10.0? This vulnerability checks every box for maximum severity:

• No authentication required (attack complexity: low)
• Remote exploitation over the network
• No user interaction needed
• Complete confidentiality, integrity, and availability impact
• Enables full remote code execution with system-level privileges

Real-World Impact

n8n is a widely-deployed workflow automation platform used by thousands of organizations for:

• API integrations and data pipelines
• Cloud infrastructure automation
• DevOps workflow orchestration
• Business process automation
• Data transformation and ETL processes

These deployments typically have access to:

• Database credentials and API keys
• Cloud infrastructure credentials (AWS, Azure, GCP)
• Internal network access
• Customer data and business intelligence

A single compromised n8n instance can be the gateway to your entire cloud infrastructure.

Technical Details: How the Attack Works

Understanding this vulnerability is crucial for bug hunters looking for similar patterns in other platforms. Here's the complete attack chain:

The Content-Type Confusion Bug

n8n's webhook handler processes incoming HTTP requests to execute workflows. The vulnerability exists in how n8n parses the request body based on the Content-Type header.

Normal behavior:

1. Webhook receives request with Content-Type: multipart/form-data
2. Multipart parser validates file uploads
3. Files are stored securely with access controls

Vulnerable behavior:

1. Attacker sends Content-Type: application/json
2. JSON parser processes body containing crafted files object
3. Multipart validation is completely bypassed
4. Arbitrary file paths can be read/written

Attack Chain: From File Read to RCE

Step 1: Arbitrary File Read

The attacker crafts a JSON payload with a malicious files object:

POST /webhook/test HTTP/1.1
Host: target-n8n.example.com
Content-Type: application/json

{
  "files": {
    "config": "/home/node/.n8n/config"
  }
}

This bypasses validation and allows reading:

• /home/node/.n8n/config - Database credentials
• /proc/self/environ - Environment variables with API keys
• /etc/passwd - System user information
• /home/node/.n8n/.secret - Encryption keys

Step 2: Session Forgery

Using extracted secrets from step 1:

• Read ENCRYPTION_KEY from environment
• Read JWT_SECRET from config files
• Forge admin session token
• Authenticate as administrator without credentials

Step 3: Remote Code Execution

With admin access, attackers can:

• Create malicious workflows containing arbitrary code
• n8n executes workflows in Node.js environment
• Workflows can execute system commands via child_process
• Install backdoors, exfiltrate data, pivot to other systems

Why This Pattern Matters for Bug Hunters

Content-type confusion vulnerabilities are everywhere. Look for these patterns:

• Different parsers for different content types (JSON vs XML vs multipart)
• Security controls applied at parser level instead of business logic level
• Assumptions that "frontend always sends correct content-type"
• File upload handlers that trust Content-Type header

Impact: Complete Infrastructure Compromise

This isn't just about compromising one server. n8n instances typically serve as automation hubs with privileged access to your entire stack:

Immediate Impact

• Credential theft: Database passwords, API keys, service account tokens
• Data exfiltration: Customer data, business intelligence, source code
• Backdoor installation: Persistent access via system-level shells
• Workflow manipulation: Inject malicious code into existing automation

Lateral Movement Scenarios

AWS Environment:

• Extract AWS credentials from environment variables
• Access IMDSv1 (if not disabled): http://169.254.169.254/latest/meta-data/iam/security-credentials/
• Assume IAM roles with broad permissions
• Access S3 buckets, RDS databases, Secrets Manager
• Launch additional EC2 instances for crypto mining

Azure Environment:

• Extract Azure AD service principal credentials
• Access Azure Key Vault secrets
• Pivot to Azure SQL, Storage Accounts, Container Registries
• Modify Azure DevOps pipelines for supply chain attacks

GCP Environment:

• Extract service account keys from environment
• Access Google Cloud Storage buckets
• Query BigQuery datasets for sensitive data
• Compromise Container Registry for image poisoning

Enterprise Risk Scenarios

Financial Services: Payment processing credentials, customer financial data, transaction histories

Healthcare: Patient records, HIPAA-protected data, medical system credentials

E-commerce: Customer databases, payment gateways, inventory systems

SaaS Platforms: Multi-tenant data access, customer credentials, application secrets

How to Test for This Vulnerability

For Security Researchers: If you're testing applications with workflow automation, webhook handlers, or API integration features, here's how to identify similar vulnerabilities.

Step 1: Identify Webhook/Upload Endpoints

Use Burp Suite Professional to map your target:

• Enable proxy and browse the application
• Look for endpoints accepting file uploads or webhook data
• Note all endpoints with Content-Type validation
• Pay attention to workflow/automation features

Step 2: Test Content-Type Confusion

Baseline test: Send the same request with different content types

// Original request
POST /webhook/handler HTTP/1.1
Content-Type: multipart/form-data; boundary=----WebKitFormBoundary
...

// Test 1: Change to application/json
POST /webhook/handler HTTP/1.1
Content-Type: application/json

{"files": {"test": "/etc/passwd"}}

// Test 2: Change to application/xml
POST /webhook/handler HTTP/1.1
Content-Type: application/xml

<files><file path="/etc/passwd"/></files>

// Test 3: Remove content-type entirely
POST /webhook/handler HTTP/1.1

files[test]=/etc/passwd

Step 3: Test File Path Manipulation

If the endpoint accepts file references, test for path traversal:

• ../../../etc/passwd
• /proc/self/environ
• /proc/self/cmdline
• /home/user/.env
• /app/config/database.yml

Step 4: Look for Secondary Exploitation

If you achieve file read, escalate to:

• Credential extraction: Config files, environment variables
• Source code disclosure: Application logic, hardcoded secrets
• Session token forgery: JWT secrets, encryption keys
• Remote code execution: Template injection, command injection

Step 5: Test Adjacent Features

n8n vulnerability exists in webhook handling. Test related features:

• API integration endpoints
• Workflow execution APIs
• Data transformation handlers
• Custom function evaluators
• Template rendering engines

Mitigation: Protect Your Infrastructure

For n8n Users - Immediate Actions

1. Update Immediately

• Upgrade to n8n version 1.121.0 or higher
• Verify version: n8n --version
• Check Docker images: docker.n8n.io/n8nio/n8n:1.121.0
• Review upgrade notes at docs.n8n.io

2. Audit for Compromise

• Review logs: Look for suspicious webhook requests with Content-Type: application/json and files objects
• Check workflows: Review all workflows for unexpected modifications
• Inspect file access: Audit access to /home/node/.n8n/ directory
• Review credentials: Rotate all API keys, database passwords, cloud credentials

3. Network Segmentation

• Place n8n behind VPN or private network if possible
• Implement IP allowlisting for webhook endpoints
• Use Web Application Firewall (WAF) with strict content-type rules
• Monitor for unusual outbound connections

For Developers - Prevention Patterns

Content-Type Handling Best Practices:

• Validate at business logic layer: Don't rely on parser-level security
• Strict content-type enforcement: Reject unexpected types explicitly
• Separate handlers: Different code paths for JSON vs multipart vs XML
• Input validation: Validate all object properties, not just content-type

Example secure implementation:

// BAD - Parser-dependent security
app.post('/webhook', (req, res) => {
  // Relies on Express/body-parser to handle content-type
  const files = req.body.files; // Vulnerable!
});

// GOOD - Explicit validation
app.post('/webhook', (req, res) => {
  const contentType = req.get('Content-Type');
  
  if (!contentType || !contentType.includes('multipart/form-data')) {
    return res.status(400).json({ error: 'Invalid content type' });
  }
  
  if (typeof req.body.files === 'object') {
    return res.status(400).json({ error: 'Unexpected data structure' });
  }
  
  // Process with multipart-specific handler only
  handleMultipartUpload(req, res);
});

For Bug Bounty Programs

If you run a workflow automation or webhook-based platform:

• Prioritize content-type confusion testing in your security audits
• Add specific test cases for parser bypass scenarios
• Consider dedicated bounties for webhook security issues
• Document expected content-types in API specifications

Defense in Depth Recommendations

Cloud-level protections:

• IMDSv2 enforcement: Disable IMDSv1 in AWS (prevents metadata endpoint abuse)
• Principle of least privilege: Limit IAM roles to minimum required permissions
• Secrets management: Use AWS Secrets Manager/Azure Key Vault instead of environment variables
• Network policies: Restrict egress to known-good destinations only

Level Up Your Security Testing

Want to find critical vulnerabilities like CVE-2026-21858? These resources will help you master cloud security and bug bounty hunting:

Frequently Asked Questions

Key Takeaways

1. CVSS 10.0 is rare: CVE-2026-21858 represents a perfect storm of critical factors - treat this with maximum urgency
2. Content-type confusion is common: This vulnerability class exists across thousands of applications - learn the pattern
3. Workflow platforms are high-value targets: They have privileged access to your entire stack
4. Update immediately: If you're running n8n < 1.121.0, you're vulnerable right now
5. Defense in depth matters: Network segmentation and IMDSv2 would have limited impact even if exploited

For Bug Hunters:

• Test every webhook and file upload endpoint for content-type confusion
• Chain vulnerabilities: file read → credential theft → RCE
• Focus on workflow automation platforms in your bug bounty targets
• Document exploitation chains thoroughly for maximum bounty payouts

For DevSecOps Teams:

• Audit all workflow automation and webhook handlers immediately
• Implement strict content-type validation at business logic layer
• Enable IMDSv2 and disable IMDSv1 across all cloud infrastructure
• Rotate credentials if you suspect any exposure

Remember: This vulnerability demonstrates why cloud security isn't just about firewalls and IAM policies. Application-level vulnerabilities can bypass all your infrastructure security. Stay vigilant, test thoroughly, and never trust content-type headers. 🎯