01 · Subscribe: Understand Subscriptions and Server Entries
Know the Three Types of Data
A server entry contains the address, port, protocol, and authentication details needed for a connection. Subscribe stores a subscription URL that can be fetched again to retrieve server entries in bulk. Config determines how traffic matches rules and selects policies. These objects are related but not interchangeable. Successfully importing a subscription only means its data appeared in the list; it doesn’t mean a server is selected or that rules are active as expected. When troubleshooting, first identify which type of data you’re viewing to avoid mistaking a rules issue for a subscription problem.
A subscription may contain multiple servers, while a single server sharing link usually describes just one entry. A subscription URL may look like an ordinary web address, but that doesn’t mean it belongs in the Config download flow. Likewise, a configuration file URL is not a server subscription. Check the stated purpose and format of the information you received before choosing an import method. If the URL has no explanation, confirm whether it’s a subscription, a configuration file, or a single-server sharing link. Don’t try it in every import flow and keep all the resulting entries.
Add a Subscription with Add Server
In Home, open Add Server, choose Type, then select Subscribe. You can also use the Subscribe option shown in your current interface. Enter the complete subscription URL, add a note that identifies its purpose, then save and update it. The goal is to create a subscription you can maintain, not to enter the URL as a regular server’s Address. Depending on the interface, the note may appear as Remarks or as the list name. Use the field according to its purpose rather than searching for an exact match to this guide.
Before pasting, check for spaces, line breaks, Chinese punctuation, or chat text at either end. A URL’s query parameters may contain credentials, so don’t remove anything after the question mark just because the URL is long. Don’t replace hyphens with lookalike characters, either. Copying a link’s displayed text may give you something different from copying its actual URL. After saving, reopen the edit screen and verify the address instead of relying on the short note shown in the list.
What to Check After the First Update
When the update finishes, check that servers appear under the subscription, the protocol types make sense, and the notes are recognizable. Then test one entry. A change in the number of entries is only a clue, not proof that everything is correct: subscription contents can change and may include informational items. If the list is empty, check for a request or parsing error in the app, then confirm that the URL is still valid. Don’t delete the old list right away; keeping it alongside the error can help distinguish between content that wasn’t retrieved and content that couldn’t be parsed.
Note the subscription’s purpose, such as everyday use, testing, or a backup, but don’t put passwords in the note. If the subscription updates but all its servers fail, check the server details, current network, and routing mode. If the subscription itself won’t update but existing servers still work, troubleshoot the subscription request separately. Keeping these layers distinct makes future updates and cleanup easier to manage. For the entry flow, see the quick-start guide; for common error messages, see FAQ.
02 · Add Server: Enter Protocol and Connection Details Manually
Choose Type Before Filling in Shared Fields
Manual entry is useful when you already have complete server details and need to verify each field. In Home, open Add Server and choose Type first. The protocol determines what the following fields mean and which options are available, so don’t fill in the address and password before selecting a Type at random. Shadowsocks, VMess, VLESS, Trojan, Hysteria2, WireGuard, HTTP, and SOCKS5 use different authentication methods. Fields that look alike aren’t necessarily interchangeable. If required details are missing, check the original information instead of guessing.
Enter the server’s domain name or IP in Address and the supplied port in Port. For a regular server entry, Address should not contain a webpage path, explanatory text, or a complete sharing link; use an import option to parse a link. Match every character in the domain and every digit in the port. Remarks are for identifying the entry. You can note its purpose and protocol, but a name alone can’t confirm the server’s location or performance. Check the address and test the connection.
| Field | What to Check | Common Mistakes |
|---|---|---|
| Address / Port | Match the domain or IP and port to the original details. | Adding a URL path to Address; missing a digit in the port. |
| Authentication | Enter the Password, UUID, username, or key for the selected Type. | Pasting the note instead; adding spaces before or after a credential. |
| TLS / Transport Settings | Check applicable fields such as Server Name, Host, and Path against the supplied details. | Treating the server domain, Host, and Server Name as the same field. |
| Remarks | Helps identify the entry without changing its connection target. | Using the name instead of checking the connection details. |
Check Authentication, TLS, and Transport Separately
For Shadowsocks, the encryption method must match the Password. VMess and VLESS often require a UUID and may include additional transport settings. Trojan typically uses a Password and involves TLS. Hysteria2 may include bandwidth or obfuscation settings. WireGuard uses keys, addresses, and Peer details. These examples describe field types; they aren’t universal default settings. If an option is missing from your interface, check that the selected protocol matches the supplied details. Don’t move unrelated values into other fields.
Server Name is used for TLS-related name handling; Host and Path may apply to specific transports. These values may or may not match Address, so follow the supplied details. Changing a note or importing again won’t fix a certificate error, and disabling certificate verification should not be your first routine troubleshooting step. Check the device’s time and make sure the target name matches the connection details, then verify the server configuration. Disabling verification changes the connection’s trust boundary and may hide a simpler issue, such as an incorrect address.
Save and Test a Single Entry
Save one entry first, then reopen it to confirm that the details were retained, especially if defaults might have replaced supplied values. Test the entry, connect, and visit the intended destination. Use any error message to identify where the connection failed. Only use it as a template for other entries after it passes these checks. When copying, verify each entry’s address, port, and authentication details; otherwise, different names may point to the same server. Name manual entries separately from subscription entries so you don’t mistake a subscription update for a change to your manual settings.
03 · Scan QR Code & Import from Clipboard: Check Before and After
A QR Code Only Carries Information
Scan QR Code reads the text encoded in a QR code, then the client tries to identify its format. A code isn’t necessarily a server: it could contain a single-server sharing link, a subscription URL, a webpage address, or unrecognized text. Before scanning, decide what kind of item you intend to import. After scanning, check whether the result is a server or Subscribe. Treating “QR code recognized” as “correct server imported” skips the most important step: checking the contents.
Open Scan QR Code in the current interface and grant camera access when prompted. Fit the entire code in the viewfinder, clear of chat bubbles, cropped edges, or glare. If you can scan from your photo library, choose a sharp original rather than a heavily compressed thumbnail. If scanning fails, check image quality and permissions before changing connection details. If the result is just a webpage address, check what the information is for; don’t assume the webpage content is an importable subscription.
Clipboard Import Needs the Complete Text
Clipboard import works well when you have a standard sharing link or subscription URL. Copy the complete text, then use the clipboard import option in the current interface. If the system asks whether to allow pasting, respond based on this import. Copying a chat message may include a title, numbering, quotation marks, and line breaks that the app can’t parse. For more reliable results, select just one complete link, import it, and check the result before handling the next one. Avoid pasting a long message that mixes several formats.
If nothing appears after copying, check whether the clipboard is empty or whether a later copy operation replaced its contents. Some apps copy the message text; others copy the link target. When long-pressing a link, pay attention to the action you select. Don’t add or remove a protocol prefix to force recognition, and don’t submit text containing passwords to an online conversion site. A recognition failure doesn’t necessarily mean the server is unavailable; the text may simply not use a supported sharing format.
Check the Imported Item Before Testing the Connection
After importing, check the new item’s Type, Address, Port, and note, then confirm its group. Some sharing formats include only the fields needed to connect, may omit a name, or may require you to check special transport settings. If you imported the same information by both QR code and clipboard, compare the entries so you don’t mistake duplicates for separate servers. You can temporarily label an entry “Needs review,” then decide which copy to keep after testing.
If scanning succeeds but the connection fails, distinguish parsing from connecting: parsing only means the text was converted into an entry; connecting also depends on the address, authentication, and network. If manually entering the same details works but a QR import doesn’t, compare the fields and look first for missing transport settings or encoding differences. If both methods fail, check the server details and network. Change one thing at a time and keep the original entry so you can compare results. Don’t reset permissions, recreate the entry, and replace Config all at once.
05 · Subscription Update Strategy: Separate Fetching, Parsing, and Replacement
Establish a Baseline Before Updating
Updating a subscription isn’t just adding a few lines to a list. The app fetches the subscription again and adjusts local entries based on its contents. Before updating, note the server you’re using, the subscription name, and its last known working state. Back up important manual settings. If you’ve edited the address, authentication, or transport settings of an entry from a subscription, save the details you need separately as an independent entry. Don’t assume local edits will persist; an update may regenerate entries with values from the subscription.
Try updating one subscription manually first. Confirm that the request and parsing both work before considering any automatic update option offered in the interface. Whether an automatic update runs may depend on the app’s state, system background scheduling, and network conditions; an interval isn’t a guarantee of exact timing. Before travel, a network change, or an important connection, check manually if needed. Don’t keep refreshing just because nothing changed in a short time; repeated requests can make it harder to reconstruct what went wrong.
Troubleshoot by Failure Stage, Not by Repeatedly Deleting Data
If an update fails, first check whether the app received a valid response. A network timeout, a failed domain lookup, and a server refusing the request point to different causes. A webpage-style response isn’t necessarily a valid subscription. If the app reports a parsing error, check the subscription format and returned content instead of changing every server password. Error details may include addresses or identity information, so redact them when recording the issue. The time and network type are usually more useful than a note that simply says “failed.”
Subscription requests and server connections may use different network paths. A working server doesn’t guarantee that the subscription URL is reachable, and a successful update doesn’t mean every server is usable. Keep working entries in place, note the current Global Routing mode and network, then change one condition at a time. If results differ across networks, investigate the request path or network restrictions first. Avoid rebuilding the entire list over a temporary network issue.
Check What Changed After an Update
After a successful update, don’t stop at the confirmation message. Check that the selected entry still exists, whether its group changed, and whether any manual settings you need were overwritten. Server names can change, and entries with the same name may point to different addresses. Check the actual connection fields, not just the note. If the current connection fails after an update, test a single server whose details you’ve verified, then check routing. Don’t change Config at the same time; you need a clear before-and-after comparison.
For important changes, work in this order: back up, update one group, test a sample, review the differences, then move to the next group. If an update creates many duplicates, check whether the same address was added as multiple Subscribe entries before deleting servers in bulk. Restoring an old snapshot only restores the saved data; it doesn’t roll back the remote subscription. Older snapshots may contain expired addresses or credentials, so retest after restoring. Past success doesn’t guarantee a connection will work now.
06 · Organizing Multiple Subscriptions: Groups, Names, and Duplicates
Map Each Source Before Changing the Display Order
The first step in managing multiple subscriptions isn’t sorting every entry by latency; it’s knowing which Subscribe manages each group of servers. Give subscriptions clear, consistent names based on purpose, such as everyday use, testing, or backup. Use a separate naming convention for manual entries. Don’t put a complete subscription URL, password, or UUID in a name. List the subscription objects first, then check their servers to distinguish a subscription added twice from two subscriptions with similar entries.
If the interface supports collapsing groups or changing their order, put frequently used groups where they’re easy to find. Display position doesn’t determine connection priority. Don’t assume the top entry becomes the active server, or that sorting the list changes a rules policy. After organizing, check the selected server and the policy references in Config. When maintaining several groups, preserving clear group boundaries makes update changes easier to track than mixing everything into one list.
Compare Connection Details to Identify Duplicates
Two entries with the same name aren’t necessarily duplicates, and entries with different names may be identical. Compare at least the protocol type, Address, Port, authentication identity, and applicable transport settings. Comparing only the server IP may lead you to delete entries with different ports or credentials; comparing notes is even less reliable. Entries for different purposes may use the same destination but different Host, Path, or Server Name values. Those differences can affect the connection, so don’t merge entries just to shorten the list.
For suspected duplicates, inspect the full details of a pair first and confirm how each was imported and which subscription owns it. If the same subscription URL was saved twice, remove the extra object at the subscription level instead of deleting every server beneath it; otherwise, the entries may return on the next update. If two different subscriptions contain the same server, you can keep both groups if that suits your workflow. The ability to maintain and trace entries is often more important than having the shortest possible list.
| What You See | Check First | What to Do |
|---|---|---|
| Two groups are nearly identical | Was the same Subscribe added twice? | Back up, then organize the subscription entries. |
| Same name, different details | Type, port, authentication, and transport fields. | Clarify the notes; don’t merge yet. |
| Entry returns after deletion | Is a subscription update recreating it? | Check which subscription manages it. |
| Manual changes revert | Was an entry within a subscription edited? | Save a separate copy for independent management. |
Clean Up in Small Batches
Export or save a copy of the current data before cleanup, and avoid updating several groups at once. Work on one subscription or a small number of entries at a time. After each step, check group names, item counts, and the current selection before continuing. If you’re unsure whether an entry has stopped working, mark it or move it somewhere easy to identify, then retest on another network before deleting it. One timeout doesn’t prove a server is permanently unavailable, and an unused subscription doesn’t automatically need to stay. What matters is knowing its purpose and how it’s maintained.
When you’re done, keep a brief record of which subscriptions are still maintained, which manual entries need independent changes, and what you removed. Don’t include sensitive URLs; local notes are enough. This makes it easier to spot unexpected additions or returning duplicates after future updates. For more detailed grouping and cleanup steps, see Organizing Multiple Subscriptions. If a connection fails after cleanup, restore the previous selection first rather than immediately restoring the entire backup.
07 · Connectivity Test: Latency, Sorting, and Routing Checks
Test Results Apply Only to Specific Conditions
Connectivity Test shows whether entries can connect under the current network conditions. Results depend on the test method and destination. A test value isn’t the same as download speed and doesn’t predict performance across every website or app. The network path, server load, and device state can all change. For a useful comparison, test on the same network, with the same method, at roughly the same time. Sorting results from home Wi-Fi and mobile data together rarely produces a meaningful comparison.
Before testing, confirm that the server details are correct and note the current network and Global Routing mode. If you have many servers, test the commonly used group first, then verify candidate entries with an actual connection. Don’t keep testing the entire list just to get a lower number; short-term variation is normal, and repeated refreshes exaggerate incidental differences. If an entry times out, wait a moment and test it by itself, then try another network if needed. The failure stage helps distinguish an unresponsive destination from a wider network-path issue.
Sorting Doesn’t Change the Selected Server or Rules
Sorting by test results usually changes only the display order. It doesn’t mean the client has switched to the first entry. After sorting, check which server is selected and verify it with an actual visit to a destination. If your Config uses policy groups, rules and group selection may determine traffic routing rather than the name shown in Home. First identify whether you’re testing a server, the current policy, or access to a particular destination. Mixing these up can make a working server appear to be the problem when the destination is what’s failing.
| Mode | What It Does | What to Check |
|---|---|---|
| Config · Configuration | Routes traffic according to the rules and policies in the current Config. | Which rule matches the destination and where its policy points. |
| Proxy · Proxy | Tests destination traffic through the selected proxy. | Whether the current server can reach the destination. |
| Direct · Direct | Tests access to the destination through the device’s direct network connection. | A successful direct connection doesn’t prove the proxy server works. |
Check the Rules Path with Minimal Changes
In Config, rules are generally checked in order, and matching traffic uses the corresponding policy. DOMAIN-SUFFIX matches domain suffixes; IP-CIDR matches IP ranges; GEOIP relies on an address-location lookup; FINAL handles traffic not matched by earlier rules. The snippet below illustrates how these components relate; it isn’t a recommendation to replace your Config. Save the original file before making changes, and confirm what PROXY points to in your configuration. A line can be syntactically valid without selecting the server you expect.
[Rule]
DOMAIN-SUFFIX,example.com,PROXY
IP-CIDR,192.168.0.0/16,DIRECT
GEOIP,CN,DIRECT
FINAL,PROXY
For testing, stick to one verified server and one destination. Confirm the connection, then check the relevant connection records and rule match. If Proxy works but Config doesn’t, focus on rule order and policy selection. If both modes fail, recheck the server fields, authentication, and network. If Direct works, that only confirms the direct route; it doesn’t show that the subscription is working. Change one condition at a time to get useful troubleshooting results.
After testing, restore your usual routing mode and check whether On Demand will trigger a connection again under its configured conditions. For details, see the On Demand setup guide. Keep the final sort order focused on finding and maintaining entries: clear groups, an explicit selection, and repeatable test results are more useful than simply chasing the lowest latency.
08 · Deleting & Backing Up: Limit the Scope and Verify Restores
Confirm the Item and Scope Before Deleting
Before deleting anything, identify its level: an individual server, an entire Subscribe entry, a Config, or an exported file. Each has different effects. A server removed from a subscription may return on the next update; removing Subscribe may affect all entries it manages; removing Config may change which rules are in use. Read the confirmation prompt and check the item name and scope instead of guessing from the current screen. If a bulk action’s scope is unclear, cancel it and save a copy first.
Check whether the server currently in use belongs to the group you plan to delete and whether any policies in Config refer to its entries. Select a verified replacement before removing old data to reduce interruptions. Don’t delete every entry in a group after just one timeout. Likewise, if you’ve confirmed that a subscription is no longer needed, don’t just clear its server entries while leaving the subscription active. Align cleanup with how the data is maintained, or later updates may cause confusing changes.
Back Up Both Servers and Config
Before backing up, list what you may need to restore: manual servers, Subscribe URLs, the current Config, required rule files, and any settings you need to record separately. Exporting a configuration file may not include all server data; exporting servers may not include all app settings. Check exactly what the export function includes and verify each item during restoration. A file named “backup” isn’t necessarily a complete backup, and its size alone doesn’t prove it covers everything.
When using export or sync from Settings or the current data management interface, save to a location where you control access. Subscription URLs, passwords, UUIDs, and keys may be included in exported data, so treat backup files as sensitive. Cloud sync is convenient across devices but isn’t a substitute for an independent snapshot: deletions and mistaken changes may sync to other devices too. Before bulk cleanup, keep a separate copy that won’t change along with your current data. That gives you a reliable recovery point if something goes wrong.
Restore in Order to Avoid Conflicts Between Sync and Import
When moving to a new device or rebuilding data, check whether synced content is already present before choosing a snapshot to import. Syncing and manual import at the same time may create duplicates or make it hard to tell where entries came from. Restore a small number of servers or one subscription first and verify them. Then restore the Config you need and check Global Routing and On Demand. The current selection after a restore may differ from the one on your old device, so confirm it rather than relying on old habits.
After restoring, check that all subscription entries are present, manual authentication and transport settings were retained, the intended Config is selected, and its rules still point to valid policies. Addresses and credentials in an old backup may have changed. A successful import doesn’t prove that a connection works. Test one entry with an actual destination, then update subscriptions one group at a time. This makes it easier to tell whether a problem came from the snapshot or a change in remote content. For migration details, see Restoring on a New Device with iCloud and Exported Backups.
Finally, remove temporary exports, QR code images, and duplicate snapshots you no longer need, but keep one usable backup with a clear purpose. If you shared material containing credentials somewhere uncontrolled, replace the affected credentials instead of merely deleting the local copy. A simple maintenance routine is: back up before changes, verify afterward, review after updates, and clean up periodically. Keeping each action scoped gives you a reliable recovery point and reduces the need to rebuild the entire list.