Tested
These we've run end to end.
| Part | Tested with |
|---|---|
| Web app | Next.js 16, with npm |
| Backends | FastAPI, running next to the web app |
| Database | Hosted Supabase Postgres |
| Coding agent | Claude Code |
| Sign-in | Supabase Auth, Auth.js (GitHub and Google) and Clerk |
| Phone | iPhone (iOS 26), added to the home screen from Safari |
Sign-in providers all work the same way: add your VM's address to the provider once, and use redirect sign-in rather than popups. The ones above are the ones we've tried, not a list to choose from. See your stack.
Should work
We haven't tried these, but nothing should stop them.
- pnpm, yarn and bun for the web app.
- Other backends: Express, Go, or anything else you start with a command. Your VM runs each one as its own service next to the web app.
- Workers and job runners that run as a process, like a queue worker or a scheduler.
- Other coding agents, like Codex. Telescope follows changes in your repo, so any agent that can make them should work. Setup is written for Claude Code today.
- Other sign-in providers, like PropelAuth or any hosted login that redirects back to an address you add.
- Other hosted databases, like Neon or Upstash.
Not yet
- Web apps that aren't Next.js.
- Docker: databases and caches from docker-compose don't run on the VM. Use a hosted dev instance instead (Neon, Supabase or Upstash all have free tiers).
- Popup sign-in. Use your provider's redirect mode.
Didn't work? Tell us
If setup fails or your app won't run, it's probably a stack we don't support yet, and we'd like to add it. Ask your agent:
Send Telescope feedback about what didn't workIt sends us what your stack is and what went wrong, without your code or secrets. That's how we decide what to support next.
Off the map
Running a fancier setup? Agents in remote sandboxes, orchestrators that hand work between agents, background agents on a schedule: that's the new classic stack, and we haven't tried it with Telescope yet. We'd love to know how it goes.
Point it at a repo with Telescope set up and see whether the changes show up on your phone. Then tell us either way, through your agent as above, and include:
- which agents you run, and where (laptop, cloud, a sandbox service),
- what hands out the work (an orchestrator, a queue, a schedule),
- what worked and what didn't.
Whatever doesn't work yet, we'll use to decide what to build next.