Security

The CORS Proxy is designed with security as a core principle. This page documents the security features, threat model, and configuration options for hardening your deployment.

Security Features Overview

Feature	Purpose
SSRF Protection	Block requests to private networks and cloud metadata
Domain Allowlisting	Limit which domains can be proxied
Origin Validation	Restrict which websites can use the proxy
Rate Limiting	Prevent abuse and DDoS
HTTPS Enforcement	Prevent man-in-the-middle attacks
No Logging	Protect user privacy

SSRF Protection

Server-Side Request Forgery (SSRF) attacks trick a server into making requests to internal resources. The CORS Proxy blocks all requests to private networks and sensitive endpoints.

Blocked IP Ranges

Range	Description
`127.0.0.0/8`	Localhost/loopback
`10.0.0.0/8`	Private network (Class A)
`172.16.0.0/12`	Private network (Class B)
`192.168.0.0/16`	Private network (Class C)
`169.254.0.0/16`	Link-local (AWS/GCP metadata)
`::1`	IPv6 localhost
`fe80::/10`	IPv6 link-local
`fc00::/7`	IPv6 unique local

Blocked Hostnames

localhost
0.0.0.0
metadata.google.internal (GCP)
Kubernetes internal services

Blocked Requests Examples

# Cloud metadata endpoints (credential theft)
https://cors-proxy.pondpilot.io/proxy?url=http://169.254.169.254/latest/meta-data/

# Internal services
https://cors-proxy.pondpilot.io/proxy?url=http://127.0.0.1:8080/admin

# Private network scanning
https://cors-proxy.pondpilot.io/proxy?url=http://192.168.1.1/router-config

All of these return 403 Forbidden with SSRF protection error.

Redirect Blocking

The proxy blocks redirects to prevent SSRF bypass via redirect chains. A URL that redirects to a private IP is blocked even if the initial URL appears safe.

Domain Allowlisting

By default, the proxy only allows requests to a curated list of safe domains. This prevents the proxy from being used as an open proxy for malicious purposes.

Default Allowed Domains

# Cloud Storage
*.s3.amazonaws.com
*.s3.*.amazonaws.com
*.cloudfront.net
*.storage.googleapis.com
*.blob.core.windows.net

# Code Hosting
*.github.io
*.githubusercontent.com

# Public Data
data.gov
data.gouv.fr
blobs.duckdb.org
*.duckdb.org

Custom Domain Configuration

To allow additional domains, set the ALLOWED_DOMAINS environment variable:

# Allow specific domains
ALLOWED_DOMAINS="api.yourcompany.com,data.yourcompany.com"

# Allow wildcard subdomains
ALLOWED_DOMAINS="*.yourcompany.com,*.partner.com"

# Combine with defaults (not automatic, you must list all)
ALLOWED_DOMAINS="*.s3.amazonaws.com,*.yourcompany.com"

Wildcard Security

Wildcards match only one subdomain level for security:

Pattern	Matches	Does Not Match
`*.example.com`	`api.example.com`	`malicious.api.example.com`

This prevents attackers from registering deep subdomains to bypass restrictions.

Origin Validation

The proxy validates the Origin header to ensure only authorized websites can use it.

Configuration

# Allow specific origins
ALLOWED_ORIGINS="https://app.pondpilot.io,https://yourapp.com"

# Allow localhost for development
ALLOWED_ORIGINS="https://app.pondpilot.io,http://localhost:5173"

Requests without a valid Origin header or from non-allowed origins are rejected.

Rate Limiting

Rate limiting prevents abuse and protects against DDoS attacks.

Configuration

# 100 requests per minute per IP
RATE_LIMIT_REQUESTS=100
RATE_LIMIT_WINDOW_MS=60000

Default Limits

Setting	Default
Requests per window	60
Window duration	60 seconds (1 minute)

Rate-limited requests receive 429 Too Many Requests.

HTTPS Enforcement

In production mode (NODE_ENV=production), the proxy only allows HTTPS target URLs. This prevents:

Man-in-the-middle attacks on proxied data
Accidental exposure of sensitive data over unencrypted connections

# Only HTTPS allowed (default in production)
HTTPS_ONLY=true

# Allow HTTP (development only, not recommended)
HTTPS_ONLY=false

Credential Handling

By default, the proxy never forwards authentication headers:

Authorization headers are stripped
Cookies are not forwarded
API keys in headers are removed

Credential Forwarding (Advanced)

For corporate environments where you trust both the client and data source, you can enable credential forwarding:

ALLOW_CREDENTIALS=true

Even with credentials enabled:

Upstream Set-Cookie headers are blocked
The proxy never stores or logs credentials

No Logging Policy

The proxy does not log:

Request URLs
Request/response bodies
User IP addresses (beyond rate limiting)
Any personally identifiable information

This is enforced at the code level, not just policy.

Threat Model

Protected Against

Threat	Mitigation
SSRF attacks	Private IP blocking, domain allowlist, redirect blocking
DNS rebinding	Domain validation before fetch, redirect blocking
Unauthorized proxy use	Origin validation
DDoS/abuse	Rate limiting
Bandwidth exhaustion	File size limits, request timeouts
Open proxy abuse	Domain allowlist, origin validation
Credential theft	No auth header forwarding by default
Data exfiltration	Read-only operations, domain allowlist
Privacy violations	No logging policy

Not Protected Against

Threat	Why
Malicious content from allowed domains	The proxy is transparent; it doesn’t scan content
Allowed origin misuse	If an allowed origin is compromised, it can use the proxy
Physical server compromise	Standard infrastructure security applies

Production Hardening Checklist

Before deploying to production, verify:

NODE_ENV=production is set
HTTPS_ONLY=true (or using HTTPS reverse proxy)
ALLOWED_ORIGINS contains only your domains (no *)
ALLOWED_DOMAINS is appropriate for your use case
Rate limits are configured appropriately
Firewall blocks direct access to internal networks
Health monitoring is configured
HTTPS termination is properly configured

Testing Security

Test SSRF Protection

# Should return 403
curl "http://localhost:3000/proxy?url=http://169.254.169.254/latest/" \
  -H "Origin: http://localhost:5173"

curl "http://localhost:3000/proxy?url=http://127.0.0.1:8080/" \
  -H "Origin: http://localhost:5173"

Test Domain Allowlist

# Should work (default allowed domain)
curl "http://localhost:3000/proxy?url=https://blobs.duckdb.org/test" \
  -H "Origin: http://localhost:5173"

# Should return 403 (not in allowlist)
curl "http://localhost:3000/proxy?url=https://evil.com/malware" \
  -H "Origin: http://localhost:5173"

Test Origin Validation

# Should work with allowed origin
curl "http://localhost:3000/proxy?url=https://blobs.duckdb.org/test" \
  -H "Origin: http://localhost:5173"

# Should return 403 with unauthorized origin
curl "http://localhost:3000/proxy?url=https://blobs.duckdb.org/test" \
  -H "Origin: https://malicious-site.com"

Reporting Security Issues

If you discover a security vulnerability, please report it responsibly:

Do not open a public GitHub issue
Email security concerns to the maintainers
Allow time for a fix before public disclosure

See the Security Policy for details.