On December 3, 2025, the React team released a security advisory regarding a vulnerability, CVE-2025-55182, in the React server that could allow an unauthenticated, remote attacker to perform remote code execution on an affected device or system.For a description of this vulnerability,
Impact
There is a denial of service vulnerability in React Server Components.
React recommends updating immediately.
The vulnerability exists in versions 19.0.0, 19.0.1 19.1.0, 19.1.1, 19.1.2, 19.2.0 and 19.2.1 of:
- react-server-dom-webpack - react-server-dom-parcel - react-server-dom-turbopack
These issues are present in the patches published last week.
Patches
Fixes were back ported to versions 19.0.2, 19.1.3, and 19.2.2.
If you are using any of the above packages please upgrade to any of the fixed versions immediately.
If your app’s React code does not use a server, your app is not affected by this vulnerability. If your app does not use a framework, bundler, or bundler plugin that supports React Server Components, your app is not affected by this vulnerability.
References
See the blog post for more information and upgrade instructions.
Repository: migration-planner-ui-app Priority: High Location: src/ui/environment/views/AgentStatusView.tsx:50 (also CreateFromOva.tsx:196, Environment.tsx:131)
Description: Discovery agents self-report credentialUrl when registering. The UI renders this string as react-router-dom <Link to=> prop and plain <a href> with no scheme validation. react-router-dom 6.30.3 passes absolute-scheme URLs to DOM <a href>, and React 18.3.1 only warns in dev mode about javascript: - production builds render verbatim. Users are instructed to click the link to reach the discovery VM.
Security Impact:
Attacker registers agent with credentialUrl: "javascript:..."
Stored payload executes in http://console.redhat.com origin when any org user clicks link
Yields victim's RH SSO session, enabling cross-tenant data access and API actions as victim
Persistent compromise within organization
Fix Suggestion: Introduce safeExternalUrl() helper that parses URL and returns only if protocol is http: or https:. Gate all three render sites so non-http(s) credentialUrl doesn't render as link. See patches/f101.patch.
This isn't just one-off bad luck—it's a pattern of critical vulns + supply-chain fails hitting the platform behind millions of production apps (Next.js, Vercel hosting, etc.). Quick chronological rundown.
• March 2025: Next.js Middleware Authorization Bypass (CVE-2025-29927)
• December 2025: React2Shell / React Server Components RCE (CVE-2025-55182 + Next.js CVE-2025-66478)
• And Now this April 19-20, 2026: Vercel Internal Systems Breach (via Context.ai supply-chain attack>!)!<
Vercel moves fast on fixes, but three major hits in 14 months (plus DoS vulns in between) has devs rotating keys and auditing everything. If you're on Vercel: Mark all secrets sensitive. Rotate + revoke exposed creds. Check build logs & team access. Stay safe out there. Who’s next? 👀 #VercelBreach #NextJS #CyberSecurity
The traditional vulnerability disclosure timeline relies on a fundamental assumption: exploit development and vulnerability discovery take time. Over the last 12 months the integration of LLMs into offensive tooling has demonstrably broken this assumption. I recently published a technical write-up arguing that the 90-day disclosure window is effectively dead backed by three specific observations from recent incidents:
1. Automated Diff Analysis (30-minute n-days) : The safety net between a patch release and an in-the-wild exploit is gone. Taking a recent React security patch (CVE-2026-23870), I used an LLM to analyze the diff, identify the vulnerable path, and write a working DoS PoC in roughly 30 minutes. The human reverse-engineering bottleneck has been bypassed. 2. Vulnerability Convergence : I recently reported a critical P0 to a vendor and was told I was the 11th reporter in 6 weeks. LLM assisted scanners are causing independent researchers to converge on the same bugs simultaneously. An embargo no longer contains the vulnerability; it simply provides a head start to whichever threat actor also found it. 3. The Linux Kernel (Copy Fail & Dirty Frag) : The recent kernel exploits highlight this perfectly. Copy Fail (CVE-2026-31431) went from an automated AI scan to a public PoC to nation state weaponization in days. Shortly after the embargo for Dirty Frag (CVE-2026-43284 / CVE-2026-43500) was broken in hours because an unrelated third party independently discovered the same bug class using similar tooling.
The defense cannot operate on monthly cycles when the offense is operating in hours. The focus needs to shift to real-time, PR-level AI scanning to match the pace. can read the full technical breakdown and case studies on my blog:https://blog.himanshuanand.com/2026/05/the-90-day-disclosure-policy-is-dead/
I am curious if the researchers here are experiencing similar convergence rates or if you view this as a temporary anomaly while legacy codebases are scanned with new tools.