Fastify Express plugin can skip security middleware on some URLs
Under a specific non-default setting, a crafted web address can make the server run an admin page while the login and permission checks are skipped. Anyone on the internet could reach protected pages without logging in.
| Severity | CriticalCVSS 9.1 of 10 |
|---|---|
| Being exploited? | UnclearNo reports either way yet |
| Fix available? | Yes@fastify/express 4.0.9 · since 27 Oct 2026 |
| How sure is this? | Official sourceBacked by an official source |
How the attack works simplified
- 1Attacker sends requestUses an absolute-form URL target
- 2Slashes collapsedignoreDuplicateSlashes alters the path
- 3Router matches admin routeHandler is selected for the scoped path
- 4Auth middleware skippedExpress layer no longer sees the path
- ResultProtected admin handler runs without checks
What happened
In @fastify/express versions 4.0.5 through 4.0.8, when Fastify runs with ignoreDuplicateSlashes enabled, absolute-form request targets can cause path-scoped Express middleware to be bypassed while the matching route handler still runs.
Are you affected?
You are affected if you use @fastify/express 4.0.5 through 4.0.8 with Fastify's non-default ignoreDuplicateSlashes option enabled.
@fastify/express 4.0.5 through 4.0.8 (only when Fastify ignoreDuplicateSlashes is enabled)
What to do
- Update @fastify/express to version 4.0.9 or later
- If you cannot update now, enforce authentication in a Fastify preHandler hook that runs after routing
- Or turn off ignoreDuplicateSlashes if you do not need it
Timeline
- 27 Oct 2026, 12:17 UTCPublicly disclosed
- 27 Oct 2026, 12:17 UTCReported by NVD · NVD
- 27 Oct 2026, 12:18 UTCPicked up by Zero Radar
Sources 1 report, oldest first
Official vendor, government CERT or advisory database
- CVE-2026-112362NVD · 27 Oct 2026, 12:17 UTC
Technical details
- ID
- CVE-2026-112362
- CVSS
- 9.1 / 10
- EPSS
- 0.4% chance of exploitation within 30 days
- Disclosed
- 27 Oct 2026, 12:17 UTC
- On Zero Radar
- 27 Oct 2026, 12:18 UTC (1 min after disclosure)
- Last change
- 27 Oct 2026, 12:18 UTC
The plain-language parts and the attack diagram were written by an AI model (anthropic/claude-haiku-5.5) from the sources above. Facts like versions, exploitation and dates come from the sources themselves. How Zero Radar works