The short answer: CVE-2026-111497, nicknamed Node2Shell (also Broccoli2Shell and HTTP2Shell), is a critical memory-corruption flaw in the decoder for Google’s Brotli compression format. A remote, unauthenticated attacker can send a crafted compressed stream that corrupts memory, leaks data, crashes the service, or potentially runs code. CISA added it to the Known Exploited Vulnerabilities catalog on October 27, 2026. Patch Node.js, .NET, and your proxies, restart the processes, rebuild your container images, and block Brotli-encoded requests anywhere you can’t patch yet.
The flaw itself is a fairly ordinary bug. What makes it a useful case study is where it lives: in a compression library most teams never chose, never listed, and never thought of as attack surface.
Why a compression library matters this much
Brotli is a compression format that browsers and servers use to make web traffic smaller. When a client sends a request with Content-Encoding: br, the body arrives compressed. Node’s core HTTP server passes those bytes through untouched, but many frameworks, body-parsing middleware, proxies, and gateways decompress them automatically, often before your application code or authentication ever runs. Wherever that happens, the decoder is processing untrusted input. So the question for each service is whether anything in front of it decodes Brotli, and at what point.
CISA describes the root cause as an unchecked return value. In plain terms, the decoder trusted a result it should have verified, and a carefully built stream can make it write outside the memory it owns. The Hacker News reports that the flaw was found with OpenAI’s new security model and rates it CVSS 9.2.
Where it hides in your stack
| Where | Why it’s exposed | What to check |
|---|---|---|
| Node.js services | Where the flaw was first identified. A service is reachable when a framework, middleware, or built-in feature decodes Brotli request bodies, often before authentication runs. | Runtime version against the Node.js security release; base images of every container |
| .NET applications | Microsoft shipped an emergency .NET update and warned of privilege escalation on vulnerable versions. | Installed .NET runtimes on servers and in images; self-contained app builds |
| Reverse proxies and HTTP servers | Anything that decodes Brotli on the way in, using a vulnerable implementation, is in scope. | Proxy, load balancer, and API gateway versions; Brotli modules |
| HTTP clients | Clients that decompress Brotli responses can be attacked by a malicious or compromised server. | Outbound integrations, crawlers, webhook consumers |
| Container images | A patched host does nothing for an image built last month. The vulnerable library ships inside the image. | Every image in your registry, not just the ones running today |
The fix, in order
- Patch the runtimes. Apply the Node.js, .NET, and proxy vendor updates. Then restart the processes: a patched library on disk does nothing for a process that loaded the old one.
- Rebuild and redeploy images. Rebuild every container image from an updated base and redeploy. Pull old tags out of circulation so autoscaling doesn't bring a vulnerable copy back.
- Block or disable Brotli decoding where you can't patch yet. Reject incoming requests with Content-Encoding: br at the proxy or firewall, or turn off Brotli decoding. First check which clients send Brotli-compressed request bodies: browsers rarely do, but some API clients and SDKs do, and they will get an error (often HTTP 415) rather than retrying with gzip until they are reconfigured.
- Find the transitive copies. Search your SBOMs and lockfiles for Brotli implementations pulled in by other packages. The library you didn't choose is the one you won't think to patch.
- Record the evidence. Note what was patched, when, and how you verified it. CISA set November 10 as the deadline for federal agencies; your customers' security questionnaires will ask a version of the same question.
The lesson: you own your dependencies’ bugs
Node2Shell landed within about a week of PHP2Shell and the WordPress updater race. All three sat in shared code underneath thousands of applications. None of them were in code the affected companies wrote.
That is the case for three unexciting habits:
- Generate an SBOM for every build, so “do we use Brotli?” is a search, not a week of asking around.
- Rebuild images on a schedule, not just when code changes, so base-image fixes reach production without a fire drill.
- Scan what is running, not just what is in the repo. The image deployed in March is still vulnerable even if the Dockerfile is fixed.
For SOC 2 and ISO 27001 programs, this is also the evidence trail auditors look for: a known-exploited vulnerability, a documented decision, a dated fix, and proof it was verified.
Find the Libraries You Didn't Know You Shipped
Scan Ninja scans your running applications, hosts, and cloud workloads, flags known-exploited flaws first, and tracks each one through to a verified fix with audit-ready evidence. Managed Vulnerability Remediation adds a team that does the chasing for you.
Sources
- CISA warns of hackers now exploiting critical Node2Shell flaw — BleepingComputer (technical details, KEV date and deadline, mitigations, .NET update)
- OpenAI launches GPT-6 Cyber after AI helps find 81 Firefox flaws, 948 Linux exploits — The Hacker News (CVSS score and discovery credit)