Imported a subscription but the proxy page is empty? Usually the client didn’t get parseable Clash nodes—not a broken install. Change one variable at a time: refresh, network, plan, format, then logs. Separate “update failed” from “update succeeded with zero nodes.”
Errors/timeout/403 → update failed; “success” with zero nodes stays here. Logs: debug logs. Import refresher: import guide.
Read the card status first
| Card status | What it means | Where to look |
|---|---|---|
| Update failed / red | Feed download failed | Use the update-failed guide |
| Updated, 0 nodes | Parse / format / plan issue | Check format, plan status, filters |
| Nodes > 0, one group empty | Group / filter, not subscription | Use the blank-groups guide |
| Nodes > 0, Connections empty | System proxy, not subscription | Use the system-proxy guide |
Check in this order
- Click update. Import creates a card; nodes come from refresh. Confirm the card’s update time is just now.
- Can this network reach the feed? Try hotspot or “update via proxy” (useless if you have no working node yet—use backup URL/hotspot first).
- Plan expired / traffic zero? Fix on the provider panel. Choosing providers: choose a provider.
- Clash / Clash Meta format? A browser HTML page isn’t YAML. Pasting a Sing-box-only link often yields “success” with an empty list or parse errors.
- Backup URL, then logs. timeout ≈ network; 403 ≈ block; parse/yaml ≈ format. Don’t hammer the same bad URL dozens of times.
Panel looks fine, client still empty
HTML/login pages, empty CDN payloads, or local filters wiping nodes (disable filters—blank groups, latency). Compare panel counts vs the card; switch to a Clash Meta link instead of hammering a bad URL.
If the global node count is > 0 but one policy group is empty, that’s a group/filter issue—not “the subscription never arrived.” Confirm the full list exists before editing group references.
After nodes appear
Count > 0 only means parse OK. Then: latency not all timeout; system proxy shows connections; Tun only if needed (Tun). All-red latency is a speed/node issue, not “subscription still empty.”
Acceptance chain: update OK → nodes present → latency not all dead → Connections shows rows. Jump to the matching article for the missing step instead of reinstalling first.
Don’t do these together
Don’t reinstall while swapping feeds; don’t set update-proxy with zero nodes; don’t treat marketing pages as configs; don’t enable Tun hoping more toggles will invent nodes. If backup URL, network, and filters fail, send log lines to the provider.
Boundary vs system proxy
Nodes exist but Connections is empty → system proxy, not “no nodes.” Sites fail with Connections rows present → sites blocked. Builds: download center.