News · 11 December 2025
React2Shell: when a framework flaw lands on a Wednesday
On 3 December 2025 a critical vulnerability in React Server Components was disclosed, tracked as CVE-2025-55182 and nicknamed "React2Shell". It allowed unauthenticated remote code execution on servers running affected versions, because of unsafe handling of data sent to certain server endpoints. Frameworks built on that functionality, including Next.js, were also affected, and security vendors reported exploitation attempts shortly after publication.
What made it serious
Three features combine to make an incident like this stressful: the flaw is in a widely used framework, it requires no authentication and it leads to code execution. Many organisations did not know whether they used the affected feature, because it can be enabled indirectly by a framework default.
The inventory problem
The first question on any advisory day is "are we affected?" and many teams cannot answer quickly. Practical preparation includes:
- An inventory of applications and their frameworks, kept up to date automatically.
- An SBOM per application, so you can search by component and version.
- Clear ownership: someone who is responsible for each application and can patch it.
- A fast patch path for emergency changes, tested before you need it.
Scanner noise versus validated findings
After disclosure, scanners lit up with detections. Some were genuine. Others reported hosts that matched a version string but did not expose the vulnerable endpoint. Triaging by hand is tedious but valuable: it separates "installed" from "reachable" and "reachable" from "exploitable".
This is why we describe security testing as validated. A finding that a specialist has reproduced, with evidence and an impact statement, is actionable. A scanner line is a lead.
What to do after patching
Patching closes the door but does not tell you whether someone has already walked through it. Review logs for suspicious requests, look for unexpected processes and outbound connections, and rotate secrets that the application could read. Consider the application's privileges: a web process should not be able to reach the entire estate.
Habits that pay off
- Subscribe to advisories for your main frameworks.
- Test your emergency patch process quarterly.
- Keep application privileges minimal.
- Commission independent testing of internet-facing systems regularly, not only after an incident.
The next critical advisory is a question of when, not if. The difference is how many hours it takes you to answer "are we affected?".
Written in December 2025; details reflect the situation at that time.