Official Manual · Reference Guide

Shadowrocket
Subscription and Server Management Manual

For users who already have their own Subscribe link or server details. Use these eight topics to review importing, updating, organizing, and recovery. Check the target before each operation, then save your changes.

If this is your first time using Shadowrocket, start with the Quick Start Guide to import and verify a connection. Then return here to review field differences, duplicate entries, and ongoing maintenance. This page is a reference guide, not a step-by-step first-connection walkthrough.

Shadowrocket is available for iPhone, iPad, Mac (Apple silicon and Intel), Windows, Android, and Linux. This page uses the iPhone / iPad interface labels Home, Add Server, and Subscribe as a reference; on other platforms, check the corresponding fields rather than assuming the controls are in the same place. Visit the Download Shadowrocket page to get the app.

Suggested order: If an import fails, start with the first three chapters. If an imported entry behaves unexpectedly, check the link format and update result. For ongoing maintenance, start with organizing multiple subscriptions. All addresses and configuration snippets are examples for learning and cannot be used as connection details.

01 Subscribe: Entry point, URL, and import results

First, tell subscription URLs from server addresses

Subscribe takes a URL that retrieves a server list. It is not a server hostname or an account management page. A subscription URL usually includes a path and query parameters used to identify an account. A server address, by contrast, must be combined with a port, protocol, and authentication details to form a connection. Putting one in the other's field can result in a subscription that saves but will not update, or an entry missing required parameters. Before proceeding, identify which type of information you have; don't judge by whether the URL contains a domain name.

A subscription is a data source that can be fetched repeatedly. The first import creates a local list; later updates may change its addresses, names, or entry count. This differs from importing a single server once. An existing list does not mean its subscription URL is still valid, and an accessible URL does not guarantee that the client can parse its response. When troubleshooting, record separately whether the content can be fetched and whether it produces the expected entries. Don't treat these as the same step.

Save a subscription through Add Server

  1. Open Home, go to Add Server, and select Subscribe under Type. If your platform has a separate subscription entry point, make sure you're opening the form for saving a subscription URL, not the form for a single server's protocol details.
  2. Enter your complete subscription URL in the URL field. Preserve the path, query parameters, and letter case. Check that there are no leading or trailing spaces, quotation marks, or line breaks.
  3. Enter a recognizable name in Remarks. After saving, wait for the list to finish loading, then return to Home and check the group and its entries.

The example below illustrates the URL structure only. Content after the question mark may identify your account; don't delete it just because it looks like an optional note. Likewise, don't add parameters you found online. If your provider offers multiple export formats, use the existing URL specifically intended for Shadowrocket and keep the original details for reference.

https://example.com/sub?token=xxxx

Before saving, check the entire URL, not just the domain. Messaging apps may copy the punctuation at the end of a sentence, and rich text may show different text from the actual link. You can inspect the first and last characters in a private text editor and check for a full-width colon, full-width question mark, or missing middle section before pasting into the client. Don't submit the full URL to a public testing site.

Use the import result as your baseline

After the first fetch, check that the group name matches the subscription you just saved. Then inspect a server's protocol, address, and port for plausibility. An entry name is usually just a remark; it doesn't prove where the server is located or what it can do. If the list includes expiry or usage details, treat them as information returned by the provider, not as criteria for choosing a server. Don't skip checking the parameters just because the name looks reasonable.

If the list is empty, leave the existing subscription settings in place while checking that the URL is complete, the current network can fetch it, and the response isn't a login page or error message. Don't add the same URL repeatedly; that can turn one fetch failure into duplicate groups. If you already have a working configuration, keep it and test the new subscription separately. Changing both the active connection and subscription source at once makes it harder to compare results.

Your baseline doesn't need to include sensitive details. Record the group's purpose, import method, whether it updates, and the outcome of the last test. If you track the number of entries, record what you observed rather than an expected count. Lists can change during normal updates; determine whether a change is unexpected by checking the fields and testing the connection.

02 Add Server: Manual entry and protocol differences

Choose Type to match the details you have

When using Add Server to create a single server entry, first select the Type that matches your existing details. Shadowsocks, VMess, VLESS, Trojan, HTTP, SOCKS5, WireGuard, and Hysteria2 use different authentication and transport structures. They aren't interchangeable just because they share a domain and port. Select the protocol before filling in the fields to avoid losing information when switching forms. If a parameter wasn't provided, check the original instructions rather than filling it in based on another protocol.

Manual entry is useful for checking a small number of entries or pinpointing differences after an automatic import. If you already have a complete share link, importing it first usually preserves additional parameters more reliably; if you only have a list of parameters, enter them one by one in the matching fields. Either way, reopen the edit screen after saving and review the result. Text appearing in a field doesn't necessarily mean it was interpreted as the intended parameter.

FieldWhat to enterCommon mistakes
TypeThe protocol specified for the serverChoosing a different protocol with a similar name
Address / PortServer domain or IP address and portEntering a webpage path or full URL in Address
Password / UUIDOriginal authentication detailsLeading or trailing spaces, missing characters, or changed letter case
TLS / SNIOriginal encryption and certificate settingsReplacing SNI with the server address without checking
Host / PathRequirements for the selected transportMissing path slash or a Host value that doesn't match
RemarksA local remark that helps you identify the entryMistaking a remark for an authentication parameter

Check connection and transport fields separately

Address usually takes a domain name or IP, not a protocol prefix, subscription path, or query parameters. Enter Port as provided; don't guess based on commonly used ports. A domain and an IP aren't always interchangeable: using an IP may change certificate validation and name resolution. If you need to test a DNS issue, make a separate troubleshooting entry and keep the original for comparison. After identifying the cause, decide whether to change the original.

Password and UUID are both authentication details, but they aren't the same field. Shadowsocks also requires a matching encryption method; fill in VMess and VLESS according to their own parameters, and don't apply another protocol's authentication structure to Trojan. When copying authentication details, avoid autofill, smart quotes, and input-method substitutions. Compare the character count and first and last characters before saving; there's no need to show the full value in a screen recording or screenshot.

TLS determines whether the connection uses that layer of protection. SNI is the name used during the handshake, while Host and Path may be part of the transport request. These values may or may not match each other; don't set them all to Address by default. With transports such as WebSocket, a correct address and authentication may still not be enough to establish the connection. Check the transport type and its additional fields together.

WireGuard uses keys, interface addresses, peers, and routing ranges; a single password cannot substitute for its configuration. Enter Hysteria2 authentication and related options according to the details you have. If an advanced field isn't shown, first check that Type and the transport are correct. Then decide whether a different import method is needed instead of putting a parameter in an unrelated field just to save the form.

Run a small-scope check after saving

Give a manually entered entry a Remarks value that distinguishes it from subscription entries. Return to the list and make sure it wasn't added to the wrong group. Select it, run a Connectivity Test, then visit a target you know works. If the test fails, check the address and port, authentication, transport, and TLS in that order, changing only one layer at a time. If you change several things at once and it starts working, you won't know what fixed it.

If an imported entry with the same details works but the manually entered one doesn't, compare fields one by one before blaming the network. Pay particular attention to default switches, Path, SNI, and encryption method. If both fail, check the server status, current network, and rules path. See the glossary for protocol and rule terminology, or the FAQ for general troubleshooting.

Before editing a subscription-generated entry, check whether the next update will overwrite your changes. To keep a personal adjustment long term, save it as a separate, clearly named entry and note its purpose. Don't give a temporary fix exactly the same name as the original subscription entry, or it may be difficult to select the right one when sorting, testing, or cleaning up later.

03 QR code and clipboard import: Steps and troubleshooting

First, check what the QR code or clipboard actually contains

QR codes and the clipboard are just ways to transfer text; they don't guarantee that the content is complete server information. A QR code might contain a single share link, a subscription URL, or an ordinary webpage. Clipboard text may include instructions, line breaks, or multiple items. Identify the type of content before importing, then check the result. Successfully scanning a QR code doesn't prove that the server parameters are correct.

Import a single server link as a standalone server; save a subscription URL in Subscribe so it can be updated later. Treating a subscription URL as ordinary text may not create an updatable group. Putting a single-server link in the subscription URL field may fail during fetching. If you're unsure, inspect the prefix and full structure, then compare them with the next chapter. Don't cycle through every import option and create duplicate entries.

Scan QR Code: steps

  1. From the add menu on Home, select Scan QR Code. Follow the system prompt to grant camera access needed for scanning.
  2. Fit the whole QR code in the viewfinder, at a suitable distance and brightness. If the client can scan from your photo library, use a clear, original image.
  3. After scanning, check whether the item to be created is a single server or a subscription. Then verify the name, protocol, and address before saving.

Don't crop the blank border around a QR code or use an image that's been compressed repeatedly, has glare, or is obscured by interface buttons. QR codes generated from long links are usually denser, so screen brightness, zoom, and focus can affect scanning. Try enlarging the original image or displaying it on another screen. You don't need to change the text inside the QR code to reduce its density, and you should never remove query parameters.

If the scanner won't open, check whether the system allows Shadowrocket to use the camera. If access was denied, change the permission in system settings and reopen the scanner. If the camera opens but can't recognize the code, check image quality first. If the code is recognized but its content isn't supported, inspect the text structure. Handle permission, image, and format issues separately; repeatedly tapping Scan won't restore missing protocol parameters.

Remove surrounding text before importing from the clipboard

When copying, select only the complete link, without bullets, code fences, or text such as “link below.” If the system asks to allow pasting, confirm according to your intent. If the current screen offers a clipboard import action, use it to read the content. If it only shows a manual-entry form, put the content in the appropriate fields based on its type. Menu locations vary by platform, so follow the labels shown in your app.

Multiple links are usually separated by line breaks. Check each entry after importing; don't assume the client will ignore all unrelated text or that a failed batch import saved nothing. Return to the list and check which entries appeared before trying again. Otherwise, a partially successful first attempt followed by a full second import can create duplicates that are hard to identify.

Compare before and after importing

Before you start, note the target group and existing entries. After importing, find the new content and open its edit screen to check the key fields. If the QR code produced an unexpected Remarks value, don't just rename it and stop. Also check whether Type, Address, Port, and authentication belong to the server you intended to import. The name may come from part of the link or be generated during import; it isn't reliable proof of identity.

If import fails, keep the original text and don't repeatedly edit it in place. Make a copy to check for leading or trailing spaces, line breaks, and escaping, correcting only formatting issues you can confirm. If the link came from your provider and its structure is complete but it still won't parse, note the error, import method, and a redacted version of its structure for later review. Don't post the full link publicly to ask for help.

When finished, clear clipboard contents and temporary QR code images you no longer need. Pay particular attention to cross-device clipboard and photo library sync, and make sure they meet your storage requirements. Clearing temporary copies doesn't invalidate credentials that have already been exposed. If you shared them publicly by mistake, change the relevant credentials or subscription URL, reimport the updated details, and remove the old records.

05 Update strategy: Separate subscription refreshes from Config changes

What can change during an update

A Subscribe update fetches the subscription content again and updates the local list according to how the client handles it. It may add servers, remove old entries, or change names and connection parameters; it isn't just a refreshed timestamp. Avoid updating every source on impulse while relying on an existing connection for an important task. Keep the records you need and make changes when you can afford to interrupt service, so the result is easier to review.

Before updating, confirm the target subscription's name and URL. Don't replace the entire subscription just because one server stopped working. A single failure may be related to that entry's authentication, target port, or current network; an empty group is more likely to involve fetching or parsing. Keeping the old list for comparison helps you tell whether the problem began before the remote content changed or after the local refresh, instead of attributing every change to the same cause.

Update manually in small batches

  1. Note the currently selected server and Global Routing setting, and make sure you have the recovery information you need.
  2. Update one group of subscriptions and wait for a clear result. Don't trigger the same update again while it's still in progress.
  3. Review changes to the list, then test a familiar server. Once it works, move on to the next group.

Repeated refreshes won't fix an incorrect URL and may increase requests to the remote service. If your platform offers automatic update settings, choose a frequency and scope that fit your usage; more frequent isn't necessarily better. You can check subscriptions you use less often manually when needed. For lists that change frequently, use stable group names so you can still identify each source after an update.

If fetching fails, check whether the URL changed, authentication expired, or the current network can't retrieve the subscription. If the update URL is accessible only through your current connection, first make sure that connection still works. If it depends on an entry that has stopped working, use other working connection details you have retained. Don't expose the full URL to verify that fetching works.

Subscriptions and Config operate at different levels

Server subscriptions usually manage selectable connection entries, while Config may contain rules, policies, and other settings. Updating a subscription doesn't automatically fix all rules, and switching Config doesn't update server authentication. Troubleshooting should answer two separate questions: Are the server details correct? And are requests being routed as intended? Otherwise, a rule sending traffic direct can lead you to repeatedly reimport the subscription for no reason.

The snippet below illustrates rule order; it isn't a complete configuration you can use as-is. PROXY must resolve to a valid policy in the active configuration. Rules are evaluated in order, so a request matched earlier won't continue to the later FINAL rule. Pasting a rule snippet into a subscription URL or server Address has no meaning; interpret it in the context of Config.

[Rule]
DOMAIN-SUFFIX,example.com,PROXY
IP-CIDR,192.168.0.0/16,DIRECT
GEOIP,CN,DIRECT
FINAL,PROXY

If the connection path changes after switching configurations, check the active Config, rule order, and policy targets before deleting a server. DOMAIN-SUFFIX matches domain suffixes, IP-CIDR matches address ranges, and GEOIP and FINAL each have their own matching conditions. Renaming a server remark doesn't change what these rules mean. Keep a copy of the original file before editing the configuration.

Handle overwrites, empty lists, and recovery

Subscription-generated entries may be recreated during updates; whether personal changes survive depends on how the client handles them. Save any special adjustment you need to keep as a separate entry and note its purpose, rather than repeatedly editing a remote entry. If the previously selected entry is no longer there after an update, select a valid one and test it. Don't assume that an entry with the same remark is still the same connection.

If a group becomes empty after an update, stop further cleanup and check whether you have the original URL or a backup to restore from. Don't immediately delete and recreate the group; that removes useful information for comparison. If only one entry changed, keep a redacted comparison of its fields before and after. If the whole source is affected, troubleshoot the URL, fetch result, and content format in turn before deciding whether to restore local records.

A reliable update strategy answers three questions: Which group was updated? At what level did the change occur? How can you recover if it fails? A brief note is enough; don't save a full error log containing authentication details. For more connection troubleshooting, continue to the FAQ. Investigate subscription management and rule behavior separately, then cross-check them.

06 Organizing multiple subscriptions: Groups, names, and duplicates

Keep sources distinct before tidying the list

The primary goal of managing multiple subscriptions is to know where each entry came from and which updates maintain it, not to squeeze every server into the shortest possible list. Different addresses can share a name, and the same address can use a different port, authentication, or transport. Before reorganizing, keep each subscription in a clearly identified group. Then sort and remove duplicates carefully so you don't delete entries that look alike but serve different purposes.

Group names should describe how you use them, such as “Everyday,” “Testing,” or “Needs review.” Don't include passwords, subscription tokens, or complete account details. Keep names short and stable; add a brief remark for temporary states if needed. Renaming an entire group after every update makes it hard to compare past issues, while using the default name for every group makes it easy to act on the wrong one in bulk operations.

Use a naming scheme you can maintain

Give manually created entries and subscription-generated entries distinguishable names. For manual entries, you can note “manual” and the purpose in Remarks; for subscription-generated entries, keep the remote name where possible and identify the source by group. Don't assume a batch rename for tidiness will survive the next update. For automatically generated content, a clear source boundary is usually more reliable than polishing every entry.

As you organize, list the subscriptions you actually use, whether they update, which entries you added manually, and what you've backed up. This management index doesn't need authentication details. Before deleting a group you haven't used in a while, make sure it doesn't contain an entry you still rely on. Not selecting it recently doesn't prove that the configuration no longer uses it.

What you seeWhat to compareSuggested action
Identical remarksProtocol, address, port, and sourceDistinguish them first; don't delete yet
Same address and portAuthentication, transport, TLS, and purposeDeduplicate only after confirming they mean the same thing
Two groups from the same subscriptionComplete subscription URLs and management purposeKeep one working source and check what depends on it
Duplicates return after an updateWhether the remote source keeps returning the entryAdjust your organization method; don't keep deleting blindly

Judge duplicates by connection behavior

To determine whether entries are truly duplicates, compare at least the protocol, server address, port, authentication, and key transport settings. Comparing names alone can delete the wrong entry; comparing IP addresses alone isn't enough either, because one host can run multiple services. Authentication details are sensitive, so compare them on your own device. Don't export a whole group to a public spreadsheet or online deduplication tool and expose more information while organizing it.

Adding the same subscription URL more than once is a common cause of local duplicates. After confirming that two groups point to the same URL, choose one, update it, and verify the connection. Check whether the other group contains personal changes. Save anything you need to keep separately, then delete the extra group. Don't delete one first and only afterward discover that the one you kept was an old URL or a stale copy.

If the remote list itself contains duplicate entries, deleting them locally may only work until the next update. Treat this as an issue with the source content and management strategy, not as a sign that every update has damaged the list. If you don't need to remove every duplicate, preserve the source and reduce clutter with groups and sorting. If you need to manage entries independently, make clear which ones will no longer follow updates from the original subscription.

Clean up with small, controlled changes

Handle one group or one type of duplicate at a time. After each change, check that the selected entry and Config policies still work. Pay particular attention to configurations that refer to entries by name: after a rename, deletion, or update, a text reference may no longer point to the intended object. A shorter list isn't automatically a better configuration. The cleanup is done when entries are identifiable, update correctly, and can be restored.

For temporary test groups, establish a clear manual cleanup routine: record the result, remove unneeded copies, and make sure you haven't selected an entry you're about to delete. If you can't tell whether an entry is a duplicate, mark it for review instead of deleting it immediately. A few well-documented entries are easier to maintain than a growing list of unidentifiable ones.

Latency tests can help when organizing multiple subscriptions, but they help with selection; they don't prove that two entries are identical. For a step-by-step example of organizing groups, sorting, and removing duplicates, see the multiple subscriptions guide. After organizing the list, follow the next chapter to verify the actual request path rather than judging by how tidy the list looks.

07 Connectivity Test: Latency, sorting, and actual routing

A test result only answers what it measured

Connectivity Test helps assess whether a server is reachable. The time shown in the interface isn't the same as end-to-end service speed and can't guarantee future connection performance. Results can vary with the test target and method, your network, and server load. Don't rank numbers taken at different times or on different networks as if they were directly comparable. First make sure the test conditions match, then use the results to narrow down your choices and verify the final selection with a resource of your own.

A timeout for one entry doesn't always mean the server has permanently failed; a particular test method may be restricted even when the protocol can establish a connection. Conversely, a response doesn't prove the target website, DNS, TLS, and rules are all working correctly. Treat the test as one piece of evidence and consider it alongside error messages, request routing, and repeat tests. Don't delete an entire group based on one result.

Use a test sequence with a baseline

  1. Keep the device, current network, and test method unchanged. Test a few familiar entries first to confirm the test itself works.
  2. Then compare candidates in the same group. If necessary, repeat the test later to see whether the results are consistent.
  3. After choosing a candidate, check Global Routing and test a known target of your own. Then decide whether it suits the task at hand.

Avoid running repeated batch tests over a weak network. Test requests use network and server resources and may make short-term results less stable. If the whole group fails, compare against one known entry, then check device connectivity and subscription content. If only a few entries fail, check those entries' addresses, authentication, and transport. This is more likely to narrow down the cause than retesting the entire list over and over.

Global Routing determines how requests use the list

Selecting a server doesn't mean every request will go through it. The Config, Proxy, and Direct options in Global Routing affect rule processing and the connection path. Before troubleshooting, note the current setting. If you switch it temporarily to isolate a cause, restore it afterward so you don't leave your everyday routing in an unintended state.

Option and interface labelWhat to checkHow it helps during troubleshooting
Configuration ConfigRequests are handled using the active configuration and matching policiesCheck rule order and policy targets
Proxy ProxyApplicable requests use the proxy modeHelp distinguish a server issue from a rules issue
Direct DirectApplicable requests are sent directlyEstablish a baseline for local network access

These three options don't guarantee the behavior of all system traffic; consider which apps the client can handle and any platform limitations. If a target works under Proxy but fails under Config, check rules and policies before entering your password again. If it also fails under Direct, consider the target itself, the device network, and DNS. Switching options only isolates variables; it doesn't fix incorrect server details.

Sorting helps you choose; it doesn't replace ongoing maintenance

Before sorting by test results, check whether you're looking at one group or all sources. After sorting, verify each entry's name and group instead of relying on its previous position. The order may change after updates or another test. The first entry isn't a permanent recommendation. Identify frequently used connections by their remarks and configuration relationships, not by remembering that they were “third from the top.”

A lower time usually means a quicker response in that particular test; it doesn't necessarily mean the entry is better for large downloads, long-lived connections, or a specific service. Test with something close to your actual task, such as opening a known-working page or completing an expected request, and check that it remains stable. A slow webpage doesn't automatically mean high latency; page content, the destination server, and local caching also affect the experience.

If the test passes but actual access fails, check the selected entry, Global Routing, active Config, matching policy, and target domain resolution in that order. A REJECT rule may actively deny a connection; DIRECT may send a request outside the proxy; and an incorrect policy reference can make the result differ from what you expect. For details on REJECT rule boundaries, see REJECT rule scope.

Keep test notes with the conditions and outcome, such as the network, test method, whether you repeated it, and whether the actual target worked. Don't save a number without context. If an entry consistently fails, then proceed to delete it. Keep the original connection details or a working subscription URL so you can test it again if circumstances change, rather than rebuilding the parameters from scratch.

08 Deleting and backing up: Recovery materials and sensitive information

Before deleting, identify the item and scope

Deleting one server, deleting an entire subscription group, removing Config, and resetting app data have different effects. Read the item name and confirmation prompt before proceeding. Don't turn “remove failed entries” into a purge of all your management information. If you're using an entry that's about to be deleted, switch to another verified entry first, or make the change when you can tolerate an interruption. Then check whether your configuration still refers to the original name.

A subscription-generated entry may reappear after an update if you delete it; a manually created entry may not be recoverable from the subscription. Deleting a group also doesn't invalidate its subscription URL; it only removes the local management entry point. Understand the difference to decide whether you need to hide clutter, remove a local copy, stop maintaining a source, or replace remote credentials that have been exposed.

A backup is more than a server list

Recovery materials should include at least your subscription URLs, details for standalone servers, any required Config files, and a short index explaining the purpose of each group. If the client offers an export or backup option, check what it actually includes. Don't assume a single file covers subscription relationships, personal settings, and every rule. Decide what you need to restore before choosing a backup format.

MaterialWhat it restoresWhat else to check
Subscription URLRecreate an updatable listWhether the URL and authentication still work
Standalone entryRestore a manual entry or a special adjustmentWhether transport and TLS fields are complete
Config fileRestore rules and policy relationshipsWhether policy references match existing objects
Management indexExplain the purpose and recovery orderAvoid including complete authentication details

Store backup files somewhere you control and can protect from unauthorized access. File names can describe their purpose and scope, but shouldn't include tokens, passwords, or full account identifiers. If you sync between devices, make sure you control both the receiving device and the sync account. Shared photo libraries, public notes, and temporary online text pages aren't suitable places to store connection credentials; even a temporary copy increases the number of places they exist.

Verify recovery before bulk deletion

The most reliable backup check is a small recovery test: import one server into a separate test group or recreate a subscription, check key fields and connection results, then remove the test copy. Simply confirming that a file exists isn't enough. It may be missing fields, its URL may no longer work, or the configuration may depend on a policy name you didn't preserve. Test recovery before deleting the original.

During bulk cleanup, work through one group at a time. After each group, check the current connection and Config. If you're not sure what an entry is for, keep a management note and check its dependencies instead of deleting it immediately. A manual adjustment may be the working version left after troubleshooting. If it has the same name as a subscription entry, compare the parameters rather than relying on its update time or list position.

Restore in dependency order when moving to a new device

On your new device, first choose the right platform on the Download Shadowrocket page, then restore your own connection details. The iPhone / iPad version costs $2.99 as a one-time purchase on the App Store; the Mac (Apple silicon and Intel), Windows, Android, and Linux versions are free. Check the App Store listing for the iPhone / iPad system requirements. For first-launch system permission steps on desktop or Android, follow the download page instructions.

First restore subscriptions and standalone entries, then import and select the required Config. Next, check policy references, Global Routing, and the actual connection. Don't restore everything while also updating all sources in bulk; changing the backup and new content at the same time makes failures harder to trace. Once the basic connection works, update subscriptions one group at a time and repeat any necessary tests and sorting.

Don't assume every setting will transfer unchanged between platforms. Share links mainly carry connection details; platform-specific settings and system permissions still need to be checked on the destination device. Confirm that Config policies match the restored objects as well. If a setting has no equivalent control on the new platform, keep the original materials and consult the relevant instructions rather than deleting unknown fields to force the migration.

Responding to exposure and closing the loop

If a subscription URL, share link, or QR code is accidentally posted publicly, deleting the local entry or removing the message doesn't guarantee the credentials can't still be used. Use your own account management tools to replace the URL or authentication details, update your devices, and remove old copies. When asking for help, share only the protocol name, import method, redacted structure, and error; never send complete details that can establish a connection.

A regular maintenance routine can follow this order: keep recovery materials, check subscriptions, update one group at a time, test connections, verify the configuration, and then clean up. Change one main variable at each step and record the result in an index that contains no sensitive information. For a first-time setup, follow the Quick Start Guide; for a specific issue, see the FAQ. This page explains how these tasks relate and where the recovery boundaries are.