Official Manual · Settings Reference

Shadowrocket
Settings Reference

Look up what each setting does, its current state, when to adjust it, and how to verify the change.

Use this page to review settings after a connection is already established. To get started by importing your existing subscription, choosing a server, and connecting, read the Quick Start Guide first, then return to the relevant section. Interface terms are based on iPhone and iPad; settings locations and availability on other platforms may differ.

Record the original value Change one setting at a time Keep the change only after verification

Baseline Settings and Config Scope

Before changing settings, establish a baseline you can restore. Shadowrocket's behavior may depend on Settings, the selected server, the active Config, and the system's network state. An enabled switch alone does not prove that it determines where a request goes. In Home, note the server name and connection status, then check Global Routing and confirm which Config is active. Record only names that help you identify items; never paste subscription URLs, passwords, or keys into public notes. Start each test from the same set of conditions so you can tell which change caused the result.

Defaults: Use the Current Installation as Your Baseline

This guide does not assume that every device starts with the same settings. A fresh installation, restored backup, imported configuration, or system update can change what you see; some options appear only when certain conditions are met. Checking the default state means recording the current value and its source before changing it—not resetting everything first. Recommendations depend on your goal: leave settings as they are unless you have a specific problem, change one item to test a hypothesis, and restore it if there is no benefit. Don't change a connection that already works just to match the examples here.

Global Routing: Check the Three Modes First

Global Routing determines the overall routing mode, but it does not replace checking individual connections. Config routes requests according to the active configuration; Proxy can help compare the server path; Direct lets you observe access over the current network without the proxy. These are troubleshooting conditions, not speed tiers. With Proxy, local network access, server reachability, and DNS still matter. Direct does not guarantee that every app will work. For a fair test, use the same destination, network, and similar timing so changes at the destination itself don't look like mode differences.

ModePrimary useWhat to check
ConfigRoute traffic according to the configuration rulesActive Config, matched rules, and policy destination
ProxyCompare the server routeSelected server, protocol settings, and connection result
DirectCompare against the current networkNetwork authentication, direct access, and DNS results

Rule Snippets and Configuration File Scope

Rules are usually matched in order, so a broad rule near the top may take effect before a more specific rule below it. The snippet is only a syntax example, not a complete configuration to replace your file; PROXY must refer to a valid policy or definition in your Config. DOMAIN-SUFFIX matches domain suffixes, GEOIP applies conditions based on IP geolocation, and FINAL handles requests that matched no earlier rule. Save a copy of your Config before changing rule order, and edit the copy. If a connection to a domain is already open, start a new test request after the change so you don't keep observing an existing session.

[Rule]
DOMAIN-SUFFIX,example.com,PROXY
GEOIP,CN,DIRECT
FINAL,PROXY

Record Results, Not a Stack of Changes

For each test, record five things: the original state, the setting changed, the test destination, the result, and whether you restored the original value. For example, change only Global Routing while keeping the server and network the same. If there is no difference, restore the original mode before testing DNS. Keep subscription updates separate from manual edits; otherwise, a refresh can make it hard to tell whether your changes remain. If you encounter an unfamiliar keyword, check the Terminology Guide before removing anything in bulk. After each test, check an app that normally works to make sure the change hasn't affected your usual traffic.

DNS: Resolution and Connection Paths

DNS converts a domain name into an address that can be used for a connection. Resolving a name and connecting to the resulting address are separate steps. A page may fail to load because no address was returned, because the address was returned but couldn't be reached, or because a later encryption handshake failed. Don't attribute every connection error to DNS. First check whether the problem affects just one domain, then test the same destination on the current network, and finally use Log to identify the stage that failed. If your setup works, keep its current DNS configuration rather than replacing it just because another resolver is described as faster.

Check Defaults in Both Settings and Config

Before testing, review DNS-related options in Settings and DNS entries in the active Config separately. An address shown in the interface may not be the one used for every query: the Config, system network, and specific rules can all affect the result. Record the existing addresses, their order, and related switches. If an option isn't clearly shown, don't guess a value. If an imported Config includes DNS settings, read its documentation before changing them. During troubleshooting, change one identifiable source at a time and confirm that the Config is active. Editing an inactive file won't change current connection behavior.

Know the Difference Between Regular and Encrypted DNS

Regular DNS and encrypted DNS use different transport methods. Getting an address from both does not mean the configurations have the same requirements. With an encrypted DNS endpoint specified by domain name, you also need to resolve that endpoint and make sure the current network can reach it; a significantly incorrect system clock may affect certificate validation too. Use a resolver configuration whose source and transport you understand and whose endpoint is reachable on your network. Avoid stacking several unverified addresses to troubleshoot: the results and delays become harder to interpret. After changing networks, verify the same resolver setup again.

IPv4, IPv6, and Rule Matching

A domain can return both IPv4 and IPv6 addresses. If IPv6 isn't reachable along a particular route, connections may take longer or some apps may fail while others work. That alone doesn't mean you should disable all IPv6 features. Reproduce the issue on the current network and another available network, then check the address type in the connection record. If your Config uses IP-CIDR or IP-CIDR6, confirm that the actual address falls within the intended range. Domain-based and address-based rules show different parts of the picture; keep the domain and its address type linked when troubleshooting.

SymptomCheck firstDon't assume
Only one domain failsThe domain's DNS result and destination serviceAll DNS is unavailable
Works again after changing networksNetwork authentication and resolver endpoint reachabilityThe server settings must be wrong
Connection times out after resolutionDestination address, routing, and transport pathAdding more DNS addresses will fix it

Verify One Change at a Time

Stop the request you're testing and note the active Config. Change only one DNS-related setting. Save, reconnect as directed by the interface, then make a new request to the same domain and check Log for resolution and connection entries. Apps, the system, and browsers may cache results, so refreshing repeatedly may not trigger a new DNS query. If no new query appears, don't assume the change had no effect. Close the test app and try again, then compare the timestamps. If more domains start failing, restore the recorded value and check for a mistyped address, protocol, or endpoint path instead of making further changes.

Keep Results You Can Explain

After changing DNS, test both a destination that worked before and one that failed, to make sure the change hasn't broadened the problem. If the issue occurs only on one Wi-Fi network, complete its sign-in first. If only one service is affected, check the matched rule and the service itself. Don't substitute a server IP address for a website when testing: HTTPS also depends on the hostname and certificate, so the results aren't equivalent. If the cause is still unclear, follow the troubleshooting categories on the Q&A page and keep the error stage and network conditions with sensitive details removed. Don't share your entire Config.

On Demand: Triggers and Automatic Connections

On Demand applies connection behavior when its conditions are met. It doesn't decide whether a server suits your needs and can't replace Global Routing. On Demand answers “when should a connection start or change?”; Global Routing answers “how should requests be handled after it connects?” If a manual connection isn't stable yet, leave On Demand off while you verify the server and Config. Otherwise, automatic triggers, manual switching, and server failures can overlap, making “why did it reconnect?” and “why did access fail?” difficult to distinguish. Make sure the basic connection works before testing automation.

Check Current Switches and Conditions

In On Demand, note the main switch, existing conditions, and their order. Don't assume a new device starts with the feature off. A system restore or configuration migration may leave a different state than the original installation. If rules already exist, understand which networks each one covers before adding another; a Wi-Fi name may not be the only trigger. Start with as few conditions as possible, verify behavior on one clearly defined network, then expand to other scenarios. To troubleshoot a connection, you can temporarily disable automatic triggers and record the original state. Restore it when testing is complete.

Wi-Fi Names and Cellular Conditions

When matching a Wi-Fi name, check the exact network name, including capitalization, spaces, and similar-looking characters. The same name can appear in different places, so a name match does not verify that a network is trusted or make it a security boundary. Test cellular conditions after actually disconnecting from Wi-Fi; don't draw conclusions from a brief status-bar change. Dual SIM switching, hotspots, and captive portals can also change test conditions. First write down what should happen—for example, what state to expect when joining a network and after leaving it—then compare that with what actually happens instead of checking only whether the switch is on.

A Domain Condition Isn't a Complete Website Access Rule

Domain triggers are generally tied to DNS resolution or system network requests. They don't inspect every character typed into a browser's address bar. A webpage may contact several domains, and apps can reuse existing connections or cached results, so opening the same page won't necessarily trigger the same sequence every time. To test a domain condition, make a new, identifiable request and check when the event occurs rather than refreshing rapidly. Don't confuse an On Demand domain condition with a DOMAIN-SUFFIX rule in Config: they do different jobs and require separate checks for the trigger and the request's final route.

Rule Order and Fallback Conditions

When several conditions are present, check them in the order shown and confirm what each rule action means. A broad condition may take precedence over more specific ones, leaving an exception with no chance to apply. Define specific conditions and fallback conditions separately, then test each on a real network rather than assuming a complete-looking list is correct. If a platform or system version doesn't show the condition you expect, don't copy a menu path from another device. For more network-switching examples, see How to Set Up On Demand Wi-Fi and Cellular Rules. This page focuses on verification order and troubleshooting boundaries.

Troubleshoot Automatic Reconnection and Network Sign-In

If a connection returns shortly after you disconnect manually, check whether On Demand conditions are still being met instead of repeatedly tapping the main switch. To separate trigger behavior from connection failure, temporarily turn off On Demand and connect manually from Home. If that also fails, troubleshoot the server, DNS, and rules first. On Wi-Fi that requires a browser sign-in, complete authentication before checking automatic connection; otherwise, lack of access may be due to incomplete sign-in, not a rule error. Don't add more conditions to compensate for an unresolved network problem.

For final verification, test at least three cases: stay on the target network, leave it, and return. After each switch, allow the system to finish updating its network state. Record the network name, trigger condition, expected action, and actual action. Whether a change happens immediately after locking the screen may depend on system scheduling, so don't expect a fixed response time in seconds. Remove temporary conditions after testing and keep only rules with a clear purpose. Check them again after changing network names or moving to another device. Automatic connections should be predictable because their conditions are clear—not because every option is enabled.

Connectivity Test and Ping Results

Connectivity Test measures reachability under specific test conditions; it isn't an overall server-quality ranking. A time value shows what that particular test measured, not download throughput, video stability, or response times for every site. Interpret Ping results based on the actual test method; not every displayed delay is an ICMP round-trip time. Identify the selected test, destination, and network before interpreting a result. If conditions differ, two numbers may not be comparable. In particular, don't rank servers by comparing one result over Wi-Fi with another over cellular.

Confirm the Test Method and Destination

Before testing, check which Connectivity Test method is selected and whether the interface offers options such as a test address or timeout. Don't guess at parameters that aren't shown or copy values from another version. If the destination can be changed, choose an address you can access, that is reachable on your current network, and that responds consistently; record the original value. Keep the existing test configuration unless the destination is unreachable or its response isn't suitable for the test. A single failed destination shows that something is wrong with that destination and route combination; it doesn't prove the server is unavailable overall.

Connection Time vs. Full Request Time

Establishing a transport connection, completing encryption negotiation, sending an HTTP request, and receiving a response are separate stages. A test may cover one or several of them, so the same unit of time doesn't mean the same thing was measured. A successful connection to a server port doesn't confirm that authentication settings are correct; receiving a test response doesn't guarantee that a particular site will work. Use Connectivity Test as an initial check, then verify the real route with a lightweight request. To confirm protocol settings, check the protocol, address, port, and authentication details supplied by your service provider instead of repeatedly running latency tests.

What Timeouts and Unexpected Results Mean

A timeout means the expected result didn't arrive within the allotted time. The cause may be the local network, DNS, server, test destination, or return path. A timeout that's too short can mark a slow but usable route as failed; one that's too long delays the result without fixing the issue. Unless your test requires it, don't keep increasing the timeout just to get a success indication. Test a known-good route for comparison, then test the failing one. If all tests fail, first check network sign-in and whether the device can access external services at all instead of immediately deleting every server.

Result combinationNext stepKeep constant
Test passes, app or site failsCheck the domain's rules and connection logServer, network, and test time
Test fails, app or site worksCheck the test destination and methodDestination being tested and routing mode
Everything fails on the same networkCheck network sign-in and direct accessServer settings and a copy of the Config

How to Run Comparable Repeat Tests

Keep the network, server, and test method fixed. Pause bandwidth-heavy tasks, then repeat the test a few times under similar conditions. Look for persistent timeouts, large fluctuations, and whether real requests succeed; don't draw conclusions from the lowest number. Running many tests at once adds load to the device and network, which can change what you're trying to measure. Don't run a large batch while judging everyday performance. To compare two servers, use the same test conditions for each. If you switch networks, record it as a separate test rather than ranking it alongside the first set.

Cross-Check Test Results Against Rules

In Config mode, rules determine the route for each request. A successful server test doesn't prove that a particular site is using that server. Check the domain's connection record and policy first, then briefly switch to Proxy for comparison and restore Config afterward. If the test succeeds but a specific app fails, check whether it is reusing an old connection, using a different domain, or needs a transport method the server doesn't support. In your notes, specify which test succeeded and which request failed; don't just write “the network is fine.” This makes it easier to investigate the exact stage.

When testing is complete, restore the test address, timeout, and mode to their recorded values, or clearly note why you're keeping a change. You don't need to save every test result indefinitely; keep representative failure stages and reproduction conditions. If latency looks normal but connections keep dropping, check Log and network transitions rather than looking for answers in Ping numbers. For a server that remains unavailable, asking your existing service provider to confirm its status and settings is more direct than changing every local test option.

Log and Data: Find the Point of Failure

Log helps reconstruct the order of events, while Data shows traffic or connection statistics. Both make sense only in context of when the issue occurred. An increase in traffic means data passed through the measured scope; it doesn't prove that a webpage loaded completely. Missing expected log entries may be due to logging scope, filters, or a reused connection. First identify an issue you can reproduce, enable only the needed logging briefly, perform the action once, and review the result. Recording large amounts of normal activity adds noise and may retain sensitive information, so it isn't a good starting point for troubleshooting.

Check Logging Scope Before Increasing Detail

Open the relevant page and check whether logging is enabled, the current filters, the visible time range, and whether there are options to clear or export records. Available details vary by interface, so this guide doesn't prescribe one log level for every installation. Keep your existing logging settings for routine use. During troubleshooting, increase detail only enough to observe the issue, then restore the original setting. Before clearing old records, save any useful excerpts with sensitive details removed. Clearing logs won't fix a connection or replace a new request; it simply narrows the time window for your next check.

Use the Timeline to Find the Failure Stage

Read a request in stages: DNS resolution, rule matching, connection setup, authentication or encryption negotiation, and data transfer. Interpret errors in that sequence. If name resolution fails, check DNS first; if a connection is established but authentication fails, verify the server details; if the connection closes after data transfer starts, consider the destination service and network changes too. Use the exact error shown—don't treat different messages as the same cause. Identify the last stage that clearly succeeded, then investigate the next stage that failed. This is more likely to produce a verifiable result than changing every setting at once.

Matched Rules and Actual Routing

The policy assigned to a request is key to understanding Config behavior. If a record shows the domain, address, rule, and policy, read them together: the domain may match a suffix rule, the address may match a range rule, or an unmatched request may fall through to FINAL. After editing a rule, make a new request and check its new record; don't rely on seeing the rule in the file. A webpage may generate several connections to different resource domains. Check each one rather than assuming the entire page followed the same route because one connection did.

[Rule]
DOMAIN,example.com,DIRECT
DOMAIN-SUFFIX,example.com,PROXY
FINAL,PROXY

This snippet illustrates rule order: the exact domain example.com can match DOMAIN first, while other matching requests continue to the DOMAIN-SUFFIX rule. Before using it, confirm that PROXY is valid in your Config and merge the snippet in the correct location; don't replace the entire configuration with it. If you change the order, expect the suffix rule to process the exact domain first too, then verify with a new request. This example explains configuration behavior; it isn't a recommendation for accessing any real site.

Data Doesn't Confirm That a Request Completed

Upload, download, and connection statistics in Data may include background requests, retries, and other active tasks. To check a specific action, note the starting state, run one test, then review the change. If the interface lets you reset statistics, first confirm what the reset covers so you don't lose history you still need. The measurement scope may differ from your carrier's usage records, so a difference between the numbers doesn't prove something is wrong. Use Data as a clue to whether traffic occurred or requests kept retrying, then use Log and the app's result to determine whether the request completed.

Export, Redact, and Stop Logging

Before sharing logs, remove subscription URLs, authentication details, private domains, internal network addresses, and identifiable request paths. Hiding only the displayed server name may not be enough; query parameters, error context, or nearby records can still contain sensitive information. When describing a problem to your existing service provider, share the network type, approximate time, protocol name, and a redacted failure stage first. Don't send a full backup. When troubleshooting is complete, restore the logging options and remove temporary files as needed. Before clearing app data, read the iCloud Sync and Backup section and make sure you've saved a usable Config copy.

Widget: Status and Quick Actions

A Widget is a system-provided shortcut for displaying information or taking an action, not a separate connection configuration. What it can show, how often it refreshes, and which actions it offers depend on the system and the available widget type. First confirm the connection in Home, then add the Widget. If its status differs from the app, reopen the app and use its current state as your reference. An old indicator on the Widget doesn't prove the server has failed, and a delayed refresh doesn't mean the whole device is offline. Check the display and the connection separately.

Use the Widget You Have Added

Whether a Widget appears on the home screen depends on the device layout and what was restored; don't assume installing the app adds one automatically. Check whether Shadowrocket appears in the system's widget picker, then review the available sizes and descriptions. If the system doesn't offer a particular Widget, don't look for a matching button shown on another device. Choose a size that clearly displays the information you need rather than one with the most fields. After adding it, leave its settings unchanged at first and confirm which app and Config it refers to. If the Widget can be edited, note the selected item; it may not always follow the selection in Home.

Refresh Timing and System Scheduling

The system may limit background refreshes, so Widget updates can be delayed. Screen lock, low-power conditions, recent app activity, and system resource scheduling can all affect when the display refreshes. Open Shadowrocket to check its status, then return to the home screen and see whether the Widget updates. If the connection is stable in the app but the Widget is behind, troubleshoot the display path first instead of changing DNS or the server. Don't rapidly tap the Widget to force an update: those taps may trigger connection changes, creating a new connection issue where there was only a display delay.

Quick Actions and On Demand

An action from a Widget and an On Demand trigger may happen close together. If you disconnect and the Widget soon shows a connection again, check On Demand conditions instead of deleting the Widget. To test the shortcut, temporarily disable automatic triggers, confirm the initial state in Home, perform one action from the Widget, then return to the app and check the result. Restore the original On Demand state after testing. The goal is to verify whether the action was carried out, not to require every label on the Widget to update instantly. Record the action and the visual refresh separately to identify which layer has the issue.

Check References After Changing Config

After adding a server, deleting an old item, or restoring a Config, check what the shortcut refers to. If the Widget lets you choose a server or action, open its edit screen and confirm that the selection still exists. If there is no such option, follow the Widget's own instructions. Two servers with the same name aren't necessarily identical; a subscription refresh may change their details. Give frequently used items names you can recognize without exposing sensitive information. After making a change, test from the app, then check the shortcut. This helps distinguish a reference problem from a connection issue.

Privacy on the Home and Lock Screens

A Widget may reveal connection status, traffic, or item names to anyone who can see the screen. The details depend on the Widget type and system settings. Check its preview before adding it, and don't put private service addresses or account details in display names. Before sharing a device screenshot, check whether the Widget reveals usage patterns. Lock screen visibility and action permissions also depend on system settings; don't assume an action available on the home screen will work from the lock screen. If the device is managed for work, follow its management requirements instead of changing system restrictions you don't understand.

If a Widget stays blank, check in order that the app opens normally, the Widget type is still available, and the referenced item exists. Then try removing and adding the Widget again. Re-adding it troubleshoots the display layer; don't start by deleting the app or clearing its Config. If it still fails but Home connections and access work, use the app for now and keep the conditions needed to reproduce the issue. There's no need to rebuild all network settings for one shortcut. Keep troubleshooting focused on the layer where the problem occurs.

iCloud Sync, Backup, and Restore

Sync transfers state between devices; a backup keeps a historical copy that can be restored. They aren't interchangeable. Sync can propagate changes or deletions, so enabling iCloud doesn't guarantee you can recover something you delete. Before changing a major Config, export a recognizable copy using the method provided by the app, then check the sync status. Don't assume an exported Config includes every server, subscription, or app setting; verify what's included. Before moving to a new device, treat restoring the app, restoring Config, and rechecking service details as separate steps.

Check Both System and App Requirements

In system settings, first confirm that you're signed in to the intended account and that iCloud features are available. Then check the sync options in Shadowrocket. The exact location depends on the current system and app interface; don't treat menu paths on another device as universal. Note which categories are explicitly marked for sync and which are stored only on the device. Don't assume an item will migrate automatically if the interface doesn't say so. While the old device is still available, test a small item—for example, see whether a Config name and its contents appear on another device. An enabled switch alone doesn't confirm that recent changes have synced.

Check Export Contents and Sensitivity

When exporting a Config, make sure you selected the active file, not an older copy with the same name. After saving, check that the file opens and the rules are intact, and give it a name that identifies its purpose. A Config may contain server addresses, authentication details, or remote resource URLs, so store backups somewhere you control with appropriate access restrictions. Don't put the complete file in a public shared folder or send it in full to ask for help. Keep a brief note explaining the file's purpose, active mode, and anything else that needs restoring. Don't put sensitive credentials in the note's title.

Prevent Sync Conflicts by Saving Copies First

If two devices are offline while editing the same item, syncing again may create a conflict or an unexpected result. Stop editing on both devices and export the copies you can still access before comparing their contents. Don't assume files are identical just because their names match, or that the file with the latest modification time is correct; device clocks, sync delays, and local edits can all affect the result. Choose a tested baseline first, then restore it using the sync or import options currently available in the app. If there's no clear conflict-resolution prompt, don't repeatedly toggle sync in the hope of getting the result you want.

Verify Each Layer When Moving to a New Device

After restoring on a new device, first confirm that Shadowrocket opens. Then check the server list, subscription entries, Config files, and Global Routing. Test On Demand, Widget, and other optional behavior last. Your existing subscription may need a refresh, and some access permissions may depend on the device or service status; check with your existing service provider. Start with a manual connection from Home and confirm basic access before enabling automatic connections. Conditions from the old device may not apply to the new one. A restore isn't complete until you verify actual requests, not just confirm that items appear in a list.

LayerWhat to verifyWhy keep the old device
Service detailsSubscription, server, and authentication availabilityMakes missing items easier to spot
Configuration fileRules, policies, and active file matchMakes it easier to compare actual contents
Automatic behaviorOn Demand and Widget work as expected in the new environmentAvoid carrying over old network conditions blindly

What to Check When Sync Fails

If you don't see an update, first check that both devices are signed in to the expected account, the network is available, cloud storage has space, and the app shows no error. Then open the app on each device, allow sync to finish, and compare one specific item before changing anything else. Don't immediately sign out or delete the app; either action could make unsynced local content harder to recover. For the full step-by-step process, see iCloud Sync, Config Export, and New Device Restore. Keep the old device and a separate backup until the new device's Config and connections have passed verification.

Proxy Ports and Settings Recovery Order

A local Proxy port is an entry point for other programs to connect to; it is not the remote server's port. The remote port is part of your existing service's connection details. A local listening port is for the current device or programs authorized to access it. Even if the numbers match, they mean different things. If you don't have a clear need for a local proxy, leave these options as they are rather than changing them for a supposed speed boost. Support for listening, available protocols, and access from other devices depends on the options offered on your current platform. Don't copy settings from a desktop version or another device.

Check the Port, Protocol, and Listen Address

When checking settings, record the port, proxy protocol, listen address, and whether LAN access is allowed. Don't assume HTTP and SOCKS5 share an entry point, and don't treat example numbers as fixed product defaults. The protocol selected in the client program must match the listening endpoint; otherwise, the port may accept a connection but reject the request format. Also avoid conflicts with other local services. If enabling the listener immediately reports that the port is in use, identify the source of the conflict instead of trying a series of new numbers. Keep a record of the original settings, identify which program needs the entry point, and make the smallest necessary change.

Local vs. LAN Access

A listener on the loopback address is generally available only to the current device. Allowing other devices to connect also involves the listening scope, network isolation, firewall, and whether the device stays online. A successful connection from the current device doesn't prove another device on the same Wi-Fi will work. Use the narrowest access scope that meets your needs; if only the current device needs it, don't expose the listener to the LAN. For cross-device access, first confirm that the network is trusted and that access controls and authentication meet your requirements, then restore the settings when finished. Don't assume that other devices on a public network are authorized to use your local proxy.

Avoid Proxy Loops and Repeated Forwarding

When configuring a program to use the local proxy, make sure its connection won't be sent back to the same entry point in a loop. Clues include repeated retries, an unusual rise in connection counts, or all related requests stalling as soon as the proxy is enabled. Start by configuring just one test program and checking its destination and port. Don't change the system proxy, app proxy, and Config at the same time. To undo the test, first remove the proxy setting from the test program, then restore Shadowrocket's listener options. That way, the program won't keep pointing to an address that is no longer available.

Restore Settings by Failure Stage

If several changes make the cause unclear, restore the original value from the most recent recorded change first. Then check from the outside in: the current network works for direct access, Home can connect manually, Global Routing is clear, the correct Config is active, DNS results make sense, and the target request matches the expected policy. Restore On Demand, Widget, and other automatic or display features only afterward. Don't start by deleting all data; that removes both evidence and usable settings. If you need to import everything again, first export anything worth keeping and confirm that you can retrieve your existing subscription and server details.

Handle App Downloads Separately from Installation Issues

If you need to download the app again, check the distribution options on the Downloads for All Platforms page. The iPhone and iPad version is a one-time $2.99 purchase on the App Store. The Mac versions (Apple silicon and Intel), Windows, Android, and Linux versions are free; desktop and Android installers are available from this site under /dl/. Check the App Store page for iPhone and iPad system requirements. Verify the developer, Shadow Launch Technology Limited, and app ID 932747118. Reinstalling doesn't verify server settings or automatically fix an expired subscription, so save your Config before proceeding.

If a system blocks the app the first time you open it on desktop or Android, first confirm that the file came from this site's download page, then follow the instructions there. On macOS, check Privacy & Security and select “Open Anyway.” On Windows, review the SmartScreen prompt and select “More info.” On Android, allow “Install unknown apps” only for the browser that downloaded the file; you can turn off the permission afterward. Choose the Mac installer based on the chip information in About This Mac; don't guess the device architecture. See Installation Permission Guide for the full process. Don't troubleshoot system installation prompts by changing DNS, Ping, or Proxy ports.

Write Down a Result You Can Reproduce

When you finish troubleshooting, describe the change that worked and the conditions in one sentence—for example, “Restoring the original DNS on the affected network allowed the request to complete again”—rather than “Everything worked after I turned on all the switches.” Then test an app that already worked, switch networks once, and reconnect manually. Remove test rules, temporary logs, and local entry points you no longer need. If the problem persists, continue troubleshooting on the Q&A page with the failure stage and sensitive details removed. To repeat the full setup, return to the Quick Start Guide. This settings guide is here to make every change understandable, verifiable, and reversible.