Ask HN: How do you handle auth when AI dev agents spin up short-lived apps?
I’m working on AI agents used for software development. These agents automatically spin up short-lived app instances – for example per pull request, per task, or per experiment – each with its own temporary URL.
Auth is handled in the standard way:
- OAuth2 / OIDC
- external identity provider
- redirect URLs must be registered in advance and be static
This clashes badly with short-lived apps:
- URLs are dynamic and unpredictable
- redirect URLs can’t realistically be pre-registered
- auth becomes the only non-ephemeral part of an otherwise fully automated workflow
What I see teams doing instead:
- disabling real auth in preview environments
- routing all callbacks through a single stable environment
- using wildcard redirects or proxy setups that feel like hacks
This gets especially awkward for AI dev agents, because they assume infrastructure is disposable and fully automated – no manual IdP config in the loop.
So I’m curious:
1. If you use short-lived preview apps, how do you handle real auth?
2. Are there clean OAuth/OIDC patterns that work with dynamic URLs?
3. Is the static redirect URL assumption still the right model here?
4. What actually works in production?
Looking for real setups and failure stories, not theory.
5 comments
[ 2.4 ms ] story [ 22.2 ms ] threadYou haven't given us enough detail of what you are actually trying to do for anyone to give you concrete advice.
But the answer is probably a reverse proxy, checking the auth on the way. It's a standard solution for things like this, just ask your favourite AI for an example of how you'd do it.
That's what your teams are effectively already doing (redirecting through a single stable environment), they've just not realised they've setup one of their servers as an ad-hoc reverse-proxy.
Your teams could do with some senior engineers who know what they're doing.
For redirect URLs, some identity providers let you configure them via an API key.
Which resources are protected by OAuth that you want these AI agents to interact with?
Meaning: instead of trying to authenticate to the preview app, make the preview app only reachable via an encrypted application-layer channel. The "authentication" becomes having the encryption key to that channel.
For AI agents, this could simplify things:
Agent generates an encryption key when spinning up preview Preview app runs in isolated network (e.g., behind a customized Envoy proxy doing application-layer encryption without PKI/TLS) Agent communicates with preview over the secure channel No OAuth redirect URLs needed at all
The pattern we've been using is basically "encrypted lanes" between services - each preview gets its own lane(s), fully ephemeral. When the preview dies, the lane disappears.
Main downside is this doesn't help if you need humans to access the previews with their SSO credentials. But for agent-to-preview-app workflows, it's been cleaner than fighting OAuth's assumptions.
Curious if you've explored this direction or if there are constraints that make network isolation insufficient?
The main advantages for us:
- The application code is exactly the same in prod and preview environments. The only thing that changes is the OAuth provider configuration (endpoints, secrets, etc.), not the auth flow itself.
- The mock lets you specify a user ID / username directly on the sign-in screen, without real credentials or email verification. That makes it usable both for humans testing previews and for agents or automated test suites.
- It also lets us simulate third-party identity providers (Google, etc.) without actually integrating with them in previews. Dealing with things like captchas or provider-side enforcement in ephemeral environments is another source of friction we wanted to avoid.
It’s not real auth, but it keeps previews fully functional and avoids special-casing large parts of the app just to make OAuth work with dynamic URLs.
Some auth providers like Auth0 have support for it: https://auth0.com/docs/get-started/applications/dynamic-clie....
There are some caveats and security concerns when using DCR, so you should research more into that. Another option may be CIMD (https://www.ietf.org/archive/id/draft-parecki-oauth-client-i...) though I'm not totally familiar with it.
What you also mentioned, like using proxys, or wildcards, is also widely used, at least for what I've seen.