kure29/flyx
DevOps & Self-hostingA self-hosted panel for deploying proxy nodes, relaying traffic between servers, and generating client subscriptions.
FLYX is a self-hosted control panel for people running their own proxy fleet: it installs Xray, sing-box or Snell on remote servers via a small agent, wires up relays, and publishes subscriptions. If you already manage this with hand-written configs across several servers, the appeal is consolidation; the cost is trusting a young project with root-level server access.
The problem
Setting up a proxy usually means SSH-ing into each server, installing a kernel, writing an inbound config by hand, then repeating the process for every relay hop. Clients still need something usable at the end: a subscription URL that returns Clash, sing-box or Surge YAML. FLYX moves those steps into a browser panel: you register a server, paste a generated agent install command, create templates for VLESS Reality or Hysteria2, chain "front → IX → landing" hops, and share a filtered, grouped subscription.
What's different
The interesting part is the relay layer combined with subscription assembly. Nodes created elsewhere can be pasted in as share links or Clash/sing-box/Surge configs, so the panel can orchestrate hops even for servers it did not build. Subscriptions follow a Sub-Store-like model: pick nodes into groups, filter, rename, add flags, group by region, and hand out separate credentials that expire and disconnect automatically. Traffic forwarding gets quotas, rate limits and one-way or two-way billing.
That breadth is also the trade-off. The project bundles multiple kernels, ACME certificate issuance (HTTP, Cloudflare and Alibaba Cloud DNS), and a relay graph in one place; a misconfigured rule or an agent bug affects the node itself, not just a dashboard view. The README's disclaimer limits use to legal, personal learning scenarios, so anyone considering multi-tenant or commercial use should read that carefully.
Who should try it
Good fit: a solo operator with a handful of VPS machines who wants one place to provision nodes, chain relays, and give friends a subscription link without editing configs. The online demo lets you see the workflow before committing to an install. Poor fit: someone with one server and one client, where static configs are simpler, or anyone who needs a formal SLA, audit guarantees or multi-user permissions, none of which the README describes.
Requirements and limitations
The panel runs via Docker on a server; agent installation is per-server through a command generated in the panel, so each node you manage needs its own agent deploy and network reachability to the panel. The default login is admin_user / admin_user and the README says to change it immediately, so expose port 6366 only behind a firewall or reverse proxy you control. Installing kernels puts root-level software on your machines; the agent link means the panel is effectively a control plane for those hosts. Ongoing upkeep includes panel and kernel upgrades (there is an in-panel upgrade check), certificate renewal choices, and reviewing relay rules when topology changes.
Before relying on it, verify the FAQ and kernel docs cover your target OS and the protocols you need. The README does not state agent resource requirements or supported distributions, so confirm those against the documentation for your servers.
HotRepo's take
The only dated signal is a 2026-10-28 appearance on the NEW-REPOSITORY board at #22 with 100 total stars, 13 days after creation; that is not an overall Trending rank and has no comparable daily history behind it. With one snapshot, persistence of interest cannot yet be judged. Check the demo and FAQ against your OS and protocol list before deploying an agent to production hosts.
Written by AI from the project's README, updated October 28, 2026. The project's own docs are the source of truth.