dynamic segments with un-decodable entities → error 400 in dev, 500 in prod

[neptunus]

Link to the code that reproduces this issue

https://github.com/neptunus/nextjs-bad-decode

To Reproduce

1. run pnpm run or pnpm build && pnpm start
2. navigate to the front page of the demo
3. click the links to see different test cases

-- OR --

In any running NextJs app, load any dynamic segment with %A0 in the slug—for example: example.com/foo/%A0

Current vs. Expected behavior

I expected the bad link, which contains an un-decodable entity, to resolve in the same way as the good ones; probably just escaping the percentage sign (as happens with valid encoded characters such as %20)

Loads fine and escapes entities:

foo/this%20is%20fine

Instead, dev resolves to a 400 error and prod resolves to a 500.

You won't see this because the app throws an error:

foo/bar%A0

I expect no un-decodable entities at all to cause a 500 error.

This gets picked up by error-logging apparatuses, causing noise.

Provide environment information

Operating System:
  Platform: darwin
  Arch: arm64
  Version: Darwin Kernel Version 24.6.0: Wed Nov  5 21:33:58 PST 2025; root:xnu-11417.140.69.705.2~1/RELEASE_ARM64_T6000
  Available memory (MB): 32768
  Available CPU cores: 10
Binaries:
  Node: 22.22.0
  npm: 10.9.4
  Yarn: 1.22.21
  pnpm: 8.15.9
Relevant Packages:
  next: 16.2.1-canary.26 // Latest available version is detected (16.2.1-canary.26).
  eslint-config-next: N/A
  react: 19.2.4
  react-dom: 19.2.4
  typescript: 5.9.3
Next.js Config:
  output: N/A

Which area(s) are affected? (Select all that apply)

Dynamic Routes, Error Handling

Which stage(s) are affected? (Select all that apply)

next dev (local), next build (local), next start (local), Vercel (Deployed), Other (Deployed)

Additional context

This problem can be dealt with by redirecting using proxy.ts, but that's a level of overhead I'd rather not have to handle.