iPhone / iPad: $2.99 one-time purchase on the App Store · Mac, Windows, Android and Linux: free

See the two pricing tiers →

Official manual · Organized by stage

ShadowrocketUser Guide

Start with getting and installing the app, then learn about existing subscriptions, Global Routing, connection tests and settings maintenance. Availability covers all platforms; the steps that follow use the iPhone / iPad interface as an example. Menu locations may differ across operating systems.

Use the tutorial to make your first connection. Use this manual to understand settings, check edge cases and establish a maintenance routine.

SubscribeConfig / Proxy / DirectDataSettings
  • 08 usage stages
  • 02 pricing tiers
  • 03 Global Routing modes
  • CONF rule snippet guide

Stage 01 / Check your device before getting the app

Getting and installing Shadowrocket by platform

Shadowrocket is a rule-based proxy client available across platforms. This chapter explains how to get the app; later chapters cover in-app setup. The iPhone / iPad version is a $2.99 one-time purchase on the App Store. The Mac (Apple silicon and Intel), Windows, Android and Linux versions are free. These prices apply to the client only and do not include any services from a provider you already use. A one-time client purchase is not a service plan. This site does not provide, sell or recommend servers or subscriptions. After installation, you still need your own server details or an existing subscription to follow the connection steps in this manual.

iPhone / iPad: Verify the App Store listing before downloading

Open the store from the iPhone / iPad section of the download guide, or visit the Shadowrocket App Store page. Verify the developer name, Shadow Launch Technology Limited, the official app icon and app ID 932747118. Don’t rely on the name in search results alone. Check the App Store page for system requirements and the current storefront for the price at checkout. The pricing tiers remain a one-time purchase for iPhone / iPad and free versions on other platforms.

If the app doesn’t appear in search, first check whether it’s available in your current storefront region, then check the account signed in for purchases. The app isn’t available in the mainland China storefront. Before changing regions, review the account’s requirements, including any balance, active subscriptions or payment details that may need attention. If you’ve already purchased the app and are reinstalling it, go to Purchased under the same account. Don’t assume a purchase made with a different account is available to the one currently signed in. When iPhone and iPad use the same purchase account, you can redownload the app from its purchase history.

Mac, Windows and Android: Choose the right build, then handle system prompts

Get versions for other platforms from the all-platform downloads page. Installers are served from this site at /dl/shadowrocket/. On Mac, check Chip or Processor in About This Mac and choose the Apple silicon or Intel build accordingly; the choice isn’t based on appearance or purchase year. On Windows, choose the 64-bit installer. On Android, use the arm64 package listed on the page. On Linux, choose deb, rpm or AppImage based on your distribution and installation preferences. Don’t mistake a subscription file for an installer.

The desktop and Android installers aren’t code-signed, so the first launch may trigger a system warning. Continue with the steps below only after confirming that the file came from this site’s download page. Stop if the source is unknown, the file path doesn’t match or the system reports a different risk. Allowing the system to run a file doesn’t import a server or configure a network connection.

First-install and system security prompts
SystemAction and limitations
MacMove the app to Applications. If you see “Apple cannot verify the developer,” go to “System Settings → Privacy & Security” and select “Open Anyway.” If you see “app is damaged,” confirm the source, then run xattr -cr /Applications/Shadowrocket.app on the actual app path.
WindowsIf SmartScreen appears, confirm the download came from this site, then select “More info → Run anyway.” If your organization’s policy blocks the app, contact your device administrator.
AndroidEnable “Allow from this source” for the browser or file manager you’re using to open the installer. You can revoke this permission for that source after installation. If a download opened from WeChat doesn’t respond, use the system browser to open this site instead.
LinuxFor AppImage, run chmod +x with the actual file path. For deb, run sudo apt install ./ followed by the actual filename. Install rpm packages with your distribution’s package manager, and review any dependency prompts.

Verify the path and filename before running a command. The command prefixes in the table aren’t complete install commands. Managed devices may restrict app installation, VPNs or network extensions; this manual can’t replace administrator approval. After installation, open the app once and confirm that Home loads before continuing. For detailed store recovery and system security steps, see the installation guide. Don’t keep downloading the file again and again instead of addressing the system prompt.

Stage 02 / Establish a controlled starting point

First launch and permissions: Understand the connection switch

The following chapters use iPhone / iPad as an example. When you first open Shadowrocket, learn how Home, Config, Data and Settings differ: Home lets you view servers, select an item and control the connection; Config manages configuration files and their rules; Data provides traffic monitoring; Settings contains app preferences. The exact layout and available options may vary with the system and app version, so follow the interface on your device. Don’t enable every option just to make your screen look like one in an article.

When does the VPN configuration prompt appear?

The first time you try to connect, iOS / iPadOS asks to add a VPN configuration and may require your device passcode or system authentication. This allows the system to route the relevant network traffic through the client. It’s local authorization, not a request for an account from a service provider. Check that the request comes from Shadowrocket before allowing it. Afterward, return to the app and watch the switch and connection status. If you deny authorization, selecting a different server usually won’t help; try connecting again and complete the system prompt.

If the system never shows a confirmation prompt, check whether another VPN is connected, whether device management restrictions apply and whether a related configuration already exists in VPN settings. Disconnect the current VPN before testing to avoid two tools changing the network route at once. Don’t start by resetting all network settings; this also affects saved Wi-Fi networks and other connection details. On a managed device, check the network access policy rather than trying to bypass it.

Grant permissions for the task at hand, not all at once

When you paste an existing subscription from the clipboard, scan with Scan QR Code or read a code from Photos, the system may separately request clipboard, camera or photo access. Grant permissions for the operation you’re performing: the camera is needed to scan a code, but manually entering server details doesn’t require access to your entire photo library. If you deny an import permission, enter the details manually or change the relevant permission in system settings. Don’t confuse an import failure with an unapproved VPN configuration.

The system may also ask separately for local network access. Whether it’s needed depends on whether the feature accesses local devices or performs a related check; remote connections don’t all need the same permissions. If you encounter a permissions issue, note the action that triggered it, the prompt text and the current network, then check Shadowrocket’s permissions in system settings. This lets you change only the relevant setting instead of changing several things at once and losing track of the cause.

Keep a simple baseline for your first test

Leave On Demand off, select a server from a known source, and hold off on complex remote configurations or simultaneous changes to DNS, rules and transport settings. The goal is to confirm that the system allows a connection, not to set up your complete daily routing policy. Note the current Global Routing mode and the active file in Config. Even if you don’t plan to use rules yet, know which file will be read when you switch to Config.

If the connection status differs from what you expect after switching to another app, distinguish between a change in the system status bar, a stale app screen and an actual interruption to requests. Return to Home, check the selected server and switch, then test the same destination again. Battery saving, network changes and on-demand connections can all affect what you see, so keep the Wi-Fi network and device state stable during the first test. Don’t switch to cellular data midway through.

By the end of this chapter, the app should open normally, you should know where permissions came from, and you should be able to control the connection switch. Other VPNs should be disconnected, and key settings should remain largely unchanged. If you don’t yet have your own server or subscription, you can continue reading to understand how the app works; an empty list doesn’t mean installation failed. The next chapter explains how to use details you already have. For a shorter set of steps, start with the quick-start guide, then return here for settings details.

Stage 03 / Add details you already have to the client

Add Server and Subscribe: Identify the three input types

Before importing anything, identify what you have. A single server share describes one connection. A Subscribe URL retrieves a group of servers and may be updated later. A Config file can also include rules, DNS and other settings. These serve different purposes; don’t paste every link into the same field. The steps below assume you already have details from your own provider. This site doesn’t provide connection resources or rate server quality by name or location.

Add Server: Check each protocol parameter

From Home, open Add Server, select the Type that matches your details, then enter Server, Port and the authentication fields. Server is usually a domain name or IP address; don’t paste a full web URL with its path. Enter only the port supplied by the service. For Shadowsocks, enter the encryption method and password; for VMess or VLESS, enter the appropriate ID fields; for Trojan, enter the password. Fields with similar names aren’t necessarily interchangeable.

Key checks for common parameters
Parameter typeWhat to checkCommon mistakes
Server addressCheck Server and Port against the source details.Entering a web page path as Server or leaving out the port.
AuthenticationCheck the password, UUID or other authentication fields for the protocol.Pasting leading or trailing spaces, or overwriting a new value with an old credential.
TLSCheck whether it’s enabled and which Server Name is specified.Confusing the display name with the domain used for certificate verification.
TransportCheck the required transport and parameters such as Host and Path.Entering only the address and authentication details, while omitting required transport settings.

The display name helps you identify an entry; it doesn’t determine where the server connects. The TLS Server Name, transport Host and Server can also be different, so don’t change them to the same value just for consistency. Unless your provider’s details explicitly require it, don’t enable options such as skipping certificate verification to hide an error. Check the device time, destination name and original configuration first. After saving, reopen the entry to confirm the fields were saved rather than left on the edit screen.

Subscribe: Check the saved URL and update result separately

To add an existing subscription, paste the complete URL into the relevant Subscribe type or subscription import field. Give it a recognizable name, save it and run an update. Check that the beginning, end and query parameters are intact; don’t copy chat instructions, line breaks or punctuation along with the URL. Subscription credentials grant access, so don’t show the full URL in public screenshots. A QR code is simply another way to provide details: after using Scan QR Code, check whether the result is a single server or a subscription.

Saving a subscription name doesn’t mean the server list was retrieved successfully. After updating, check whether entries appear, whether their number is broadly consistent with the information from your provider, and whether any errors are shown. If the list is empty, check whether the URL can be retrieved on the current network, whether the provider has disabled it, and whether the returned content is a subscription format supported by the client. An information page that opens in a browser may not contain subscription data. Don’t keep adding the same URL; that only creates duplicate groups.

Keep subscription updates and configuration files separate

A subscription update may replace server names, addresses and parameters from that source. Before making long-term manual changes to individual subscription entries, check whether a later update will overwrite them. To keep personal notes, use clear groups and names instead of changing connection fields. If you have several subscriptions, update one source at a time and check the result before moving on. Before deleting an old source, make sure the currently selected server no longer depends on it.

If a subscription still won’t parse, use the subscription parsing troubleshooting checklist to check the input and response. If the list is cluttered, see how to group and clean up multiple subscriptions. This stage is complete when the list contains at least one server from your own details with verifiable parameters, and you know whether it was added manually or came from a subscription—not merely because a name appears in the list.

Stage 04 / Choose how requests should be handled

Global Routing: Config, Proxy and Direct

Global Routing determines the overall handling mode for requests. In the interface, Config, Proxy and Direct are the three modes; rule actions are written in uppercase as PROXY, DIRECT and REJECT. These terms are related but operate at different levels: Global Routing selects the overall mode, while a rule action determines what happens when a specific rule matches. Check the overall mode before investigating whether an individual rule works. This helps avoid blaming a rule file that isn’t active.

When to use each Global Routing mode
ModeHow it worksWhat it’s useful for checking
ConfigProcesses requests according to the rules and policies in the active configuration.Everyday routing and which rule matches a specific destination.
ProxyRoutes ordinary requests handled by the client through the proxy.A brief check of the server route; it doesn’t verify rule-based routing in Config.
DirectRoutes ordinary requests handled by the client directly.Comparing local network behavior; it doesn’t turn off the system VPN.

Config: Rule order matters more than the number of entries

In Config, check which file is active, then select Config in Global Routing. Rules are generally evaluated in order, and the first match determines the result, so put more specific conditions before broader ones. FINAL is the fallback action when no earlier rule matches and usually belongs at the end. If a catch-all action comes first, later, more specific rules may never take effect. Keep a copy of the original file before editing so you have a baseline to return to.

[Rule]
DOMAIN,example.com,DIRECT
DOMAIN-SUFFIX,example.net,PROXY
DOMAIN-KEYWORD,example,PROXY
IP-CIDR,192.0.2.0/24,DIRECT,no-resolve
GEOIP,CN,DIRECT
FINAL,PROXY

The snippet above explains syntax and ordering. It isn’t a complete configuration with all network settings, nor a routing recommendation for any particular website. DOMAIN matches an exact domain; DOMAIN-SUFFIX matches a domain and its relevant subdomains; DOMAIN-KEYWORD matches a keyword in the domain and is generally broader. IP-CIDR matches an address range; no-resolve in the example avoids an extra DNS lookup just for that IP rule. GEOIP relies on a geographic database and doesn’t indicate where a website’s business is based.

Proxy and Direct: Isolate variables during testing

If a destination fails in Config, briefly switch to Proxy and test it again using the same server, network and destination. If it works, return to Config and check the matching rule, policy target and DNS behavior; don’t conclude that the entire configuration is broken. If it still fails in Proxy, investigate the server connection and authentication. After testing, restore the original mode and verify again instead of leaving a temporary troubleshooting setting in place without remembering why.

Switching to Direct changes how the client routes requests; it doesn’t turn off the connection switch on Home. The system may still show a VPN, and the client may continue handling requests under the current settings. To compare client-handled traffic with a completely disconnected state, turn off the connection separately and record that step. Label all three test states clearly so you don’t mistake a Direct result for one taken after disconnecting.

How rule actions relate to the selected server

When a rule matches PROXY, there still needs to be an available proxy target or valid policy defined in the configuration. A correctly written rule can’t fix incorrect authentication details. DIRECT sends the request directly; REJECT blocks it. If you see a rejection record, changing servers won’t fix the issue. If the configuration refers to a specific policy name, check that it exists and points to a valid entry. A spelling mismatch, deleted group or changed name after an update can all affect the result.

When testing rules, start with a small number of conditions you can explain instead of importing many entries from different sources at once. More rules don’t necessarily make decisions more accurate, and broad keywords can also catch subdomains you intended to route directly. For details about individual settings and their relation to rules, see the settings guide. This chapter establishes three things to check—order, scope and action—before the connection tests in the next chapter.

Stage 05 / Verify step by step instead of switching repeatedly

Connection and testing: From server to actual request

After importing your details and choosing a routing mode, return to Home, check the selected server, turn on the connection switch and complete any system authorization prompt. Confirm that the app stays connected before making an actual request. A switch color, one Connectivity Test or the system status bar icon can’t confirm that every setting is correct; each reflects a different aspect. A full check covers the system connection, server reachability, authentication, domain resolution and rule result. Keep test conditions consistent at each step.

Record a baseline before running Connectivity Test

Note whether you’re on Wi-Fi or cellular, the selected server name, the Global Routing mode and the active configuration. Then run Connectivity Test and check the test method and result. Different methods may check only a port response, a protocol handshake or a particular request; they may not cover everything a real app does. A fast response doesn’t prove that large files will transfer quickly. Consider the test target and current network before interpreting a timeout, and don’t immediately assume that every server in the subscription is unavailable.

If the server list has several entries, choose one with a clear source and keep it fixed through the entire test. Change only one condition at a time: keep the network the same and switch servers, or keep the server the same and switch networks. If you update a subscription, change DNS, switch configurations and restart the device all at once, you won’t know which step affected the result—even if the problem disappears—and it will be hard to reproduce later.

Use real requests to separate connection issues from app issues

Once connected, open a familiar page that should currently be accessible, then check a second, different destination. Use a new request where possible. If needed, close and reopen an existing page, since browser cache and existing connections can make results lag behind. Also distinguish browser pages, in-app web views and background tasks: they may use different connections, so success in one doesn’t prove every app takes the same route.

If only one app fails, check whether it relies on multiple domains, accesses different destinations during sign-in and content loading, or has relevant request records around the failure time. Routing a main page domain through PROXY doesn’t mean image, authentication and API domains match the same rule. Add specific conditions based on observable requests rather than using a broad DOMAIN-KEYWORD to catch every similar name. Existing connections may also keep their old route; after changing a rule, start a new test request.

Choose the next step based on what you observe
SymptomCheck firstAvoid for now
Connection switch won’t stay onSystem authorization, VPN conflicts and whether the active configuration is usable.Repeatedly changing rules for service domains.
Port is reachable but requests failAuthentication, TLS, transport parameters and DNS.Assuming reachability means authentication succeeded.
Proxy works, Config doesn’tActive file, matching rules and policy references.Deleting every subscription and starting over.
Wi-Fi fails, cellular worksNetwork conditions, captive portal sign-in and DNS differences.Immediately overwriting all server parameters.

Troubleshoot DNS, TLS and network changes separately

If a domain can’t be resolved, the certificate name doesn’t match or a handshake fails, troubleshoot the relevant layer first. DNS resolves a domain to an address; TLS verifies the remote identity and negotiates encryption; protocol authentication may happen at another stage. Manually entering an IP address changes the resolution path temporarily, but doesn’t automatically resolve certificate name or server routing requirements. An incorrect device date can also affect certificate validation. Correct the system time before considering changes to app settings.

Public Wi-Fi may require captive portal sign-in. If every destination fails, disconnect the client, confirm that the local network itself works and complete the portal sign-in before testing again. Switching between Wi-Fi and cellular may interrupt and reconnect existing sessions, so wait for the status to settle before making a new request. Don’t treat one failure during the switch as a prolonged outage, and don’t overlook a pattern that appears only on a particular network.

When reviewing error logs, keep only the necessary entries near the failure time, and hide subscription URLs, authentication fields, private domains and device details. When contacting your provider, include the protocol, error text, network environment and single-variable tests you’ve tried; this is more useful than simply saying “it won’t connect.” Continue troubleshooting common issues on the Q&A page. Once requests are stable, move on to Data and learn how to interpret traffic readings.

Stage 06 / Treat statistics as evidence, not a verdict

Data and traffic: Understand readings and their limits

Data shows traffic activity the client can record; it isn’t a substitute for your provider’s bill or your carrier’s usage figures. In-app statistics, system cellular usage and remote account records may cover different time periods, define traffic direction differently or use different billing methods. Before interpreting a number, check the current view, its starting point and whether it has been reset. If the interface offers views by server or app, compare them under the same conditions; don’t add totals from different scopes together.

Read upload, download, totals and rates separately

Upload and download show the direction of traffic. A total is the volume over a particular reporting period, while a live rate reflects changes over a shorter time window. A rate near zero just after opening the page doesn’t mean there was no earlier transfer. A rising total doesn’t mean a download is still in progress. Units matter too: a byte count and a per-second rate are different measures. Check the units shown in the interface before comparing them; they aren’t direct measures of performance.

Loading a page usually creates several connections for content, images, fonts, API requests and background refreshes. The system and other apps may continue making requests after you close the visible page, so Data may not stop changing immediately. For a controlled observation, note the starting value, keep other activity steady where possible, complete one clearly defined task, then check the difference. A day’s total from mixed activity can’t tell you the complete route of a single request.

Limits of viewing traffic by server or app

Viewing traffic by server can help show whether the selected server is handling expected traffic, but interpret it in the context of the routing mode. In Config, some requests may go Direct or be rejected, so they won’t all count against the current proxy server. Little traffic on that server doesn’t mean an app isn’t working. If a subscription update renames, replaces or regroups servers, historical entries may not map to current ones by name alone.

When viewing by app, first check whether the current system and build provide that attribution. Some requests may appear under system processes, shared components or an unspecified source; an in-app web page may not be attributed as you expect. Don’t keep changing rules to force a complete list for every app. Traffic attribution is one way to observe activity; rule matching requires checking the actual destination and handling result. One can’t replace the other.

Use a controlled request to check traffic routing

Note the Data reading, selected server and Global Routing mode before testing. Open one specific destination, wait for it to finish loading, then note what changed. To compare Config with Proxy, use the same destination and similar steps in both modes, and account for less traffic on the second attempt due to caching. If you need to reload, make sure a new request is actually sent; don’t generate pointless traffic by refreshing repeatedly. Avoid large files as a routine test, especially on limited mobile data.

If client records differ from your provider’s usage, first check that the reporting periods start at the same time, whether other devices use the account, and whether remote figures include protocol overhead, retries or different traffic directions. Switching to another server after a subscription update may spread local readings across entries while the provider still totals usage under one account. The difference can be assessed meaningfully only after aligning the scope, period and number of devices.

Keep necessary context before resetting statistics

A reset usually changes the local measurement baseline. It doesn’t clear usage already recorded by your provider or reset your account allowance. Before clearing anything, record the data you need for the current period and check what the interface says will be affected. A statistics reset isn’t a connection fix: it can’t change server authentication, correct rules or restore an expired service. If you just want to start an experiment, recording the starting value is often enough; you don’t need to clear long-term statistics.

If traffic rises unexpectedly, pause unnecessary tasks, check for apps syncing or updating or repeatedly retrying, then investigate in the context of the current routing mode. Don’t infer which service caused the usage from the server name alone, and don’t publicly share statistics screenshots that reveal private destinations. For more detail, see how to read Data by server and by app. Once you have a consistent way to observe traffic, you can assess whether a change in Settings made a real difference.

Stage 07 / Make every settings change for a reason

Common Settings: DNS, On Demand and diagnostics

Don’t treat Settings as a checklist you have to adjust from top to bottom. Available options, their locations and descriptions may vary across systems and builds. Read the interface on your device and decide whether an option relates to the issue at hand. Network behavior may also be controlled by the configuration file, so Settings may not be the only factor. Before changing anything, note the original value, active Config and network environment. Afterward, test one specific destination and keep the change only if it helps.

DNS: Separate address resolution from rule matching

DNS resolves domain names to addresses. Routing rules determine how requests are handled. The two can affect each other, but they aren’t the same thing. Domain rules may match before an address is resolved; IP rules need the relevant address information. The configuration’s resolution method, cache and rule parameters can all affect results. If a website won’t open, don’t start by randomly changing DNS addresses. First identify whether the failure is domain resolution, a server connection or a rule sending the request down the wrong route.

If the Config file explicitly sets DNS, check its contents and documentation before changing resolution options elsewhere in the app. Keep a working baseline and avoid changing the upstream resolver, fallback behavior and rule order at the same time. After making a change, send a new request and remember that the cache may temporarily retain an earlier result. Clearing the cache can help create a fresh test, but it isn’t a universal connection fix and usually won’t resolve an authentication failure.

On Demand: Separate trigger conditions from connection status

On Demand manages or triggers connections under specified conditions. First make sure a manual connection is stable; otherwise, automatic triggers make failures harder to observe. Review each condition in the current interface, including network type, Wi-Fi scope or other applicable settings, before enabling it. Start with one condition that’s easy to test, then check behavior in environments where it is and isn’t met. Don’t add many exceptions at once and judge the result from a single status change.

If the connection reconnects automatically after you disconnect it manually, first check whether the trigger condition still applies instead of immediately deleting the VPN configuration. Temporarily turn off On Demand while troubleshooting, then return to the manual connection baseline. Review the conditions again before re-enabling it. On-demand connections don’t guarantee that all background system tasks run when you expect, and they don’t ensure that a particular destination uses a proxy; requests still follow the overall mode and active configuration.

Local network access and connection test settings

To access local devices, check the system’s local network permission, LAN-related handling in the configuration and the current subnet—not just whether the remote server works. Wi-Fi printers, storage devices and router admin pages may use private IP addresses. If your rules don’t send them Direct as expected, remote pages may work while local devices remain unreachable. Before adding an IP-CIDR rule, verify the actual subnet so you don’t change routing for a wider address range than intended.

If you can configure the Connectivity Test target, method or timeout, choose settings that match the problem you’re investigating. A shorter timeout reports failure sooner, but is also more prone to false alarms on an unstable network. An unavailable test target can make several otherwise working servers appear to fail at once. Record the new conditions after changing test settings. Don’t directly compare readings from before and after the change as evidence that server performance has changed.

Keep log and advanced-feature changes to a minimum

Enable detailed logs only when needed for diagnosis, then check the relevant time range soon after reproducing the problem. Logs may include destination domains, server addresses or authentication context. Before sharing, include only the necessary lines and hide sensitive details. If an advanced feature requires extra certificates, scripts or remote resources, first confirm your specific use case, source and scope. The first-connection steps in this manual don’t require these features; don’t add extra variables just to match a screenshot.

After changing the Config file, check network behavior again rather than assuming your previous Settings changes still take effect in the same way. For help with a specific option, consult the Settings page and Q&A by topic. Completing this chapter doesn’t mean changing every option; it means being able to explain what each change is for, how to test it and how to restore its original value.

Stage 08 / Keep sources, baselines and recovery steps on hand

Routine maintenance: Updates, backups and troubleshooting notes

Routine maintenance isn’t about constantly changing configurations. It’s about keeping sources, active files and change history traceable. Periodically check that subscriptions still update, the selected server belongs to a valid source and Config uses the intended file. When removing duplicates, reorganizing groups or updating the app, handle one type of item at a time and run a brief test afterward. If something goes wrong, you can revert the last known change instead of rebuilding every setting from scratch.

Treat the three types of updates as separate tasks

An app update changes the client. A Subscribe update changes the server list or parameters from that source. A remote Config update may change rules, resolution behavior or other settings. They aren’t inherently synchronized. A successful subscription update doesn’t mean the client is up to date or that the active configuration has new rules. In your maintenance notes, specify which update you ran, whether it produced an error and which destination you tested afterward. Don’t use “updated” to describe three different actions.

Manage app updates for iPhone / iPad through the App Store. For Mac, Windows, Android and Linux, get the appropriate installer again from this site’s download page, using the site’s /dl/shadowrocket/ path. Before replacing an installer, preserve any needed data; handle first-launch system prompts as described in Chapter 1. The pricing tiers remain $2.99 one-time on the iPhone / iPad App Store and free on the other platforms listed above. Getting the app and the term of a service you already have from a provider are managed separately.

Keep configuration, credentials and statistics separate in backups

Check the export or backup options in the current interface and what each one includes. Don’t assume a single file contains configuration files, manually added servers, subscription URLs, settings preferences and Data records. Even after a successful export, confirm that the file is readable and that you can identify its contents. Keep original sources for important details too. Before moving to a new device, note the active configuration and overall mode on the old one, then test the restored setup. This helps avoid discovering after migration that subscription authentication details are missing.

Subscription URLs, server passwords, UUIDs and some shareable QR codes may contain credentials, so protect them like account details. Store backups somewhere you control with appropriate access protection; don’t put them in public photo libraries, shared documents or group chats. When showing someone a rule, share a separate snippet without authentication details where possible. Before sharing a full configuration, check each remote URL and account field—not just the display names visible on screen.

Organize multiple subscriptions and long-lived exceptions

If you have multiple subscriptions, give them names that identify their source and purpose rather than relying on temporary sort order. Before cleaning up, check for duplicate imports from the same source, retired URLs and servers or policies that still reference entries in a subscription. After deleting a source, check the list and active selection before deleting anything else. For more on grouping, see the guide to managing multiple subscriptions. You don’t need to reinstall the app to organize the list.

Add a brief note explaining the purpose of each rule you create, and review it periodically. Broad rules added temporarily for troubleshooting are easy to forget after the issue is fixed, potentially changing routes for unrelated destinations. A remote Config may overwrite local changes, so keep track of differences before updating. If behavior changes afterward, compare rule order and policy references before replacing the entire file with an older copy. A rollback is repeatable only when its scope is clear.

Keep a reproducible troubleshooting record

At a minimum, record the exact action that failed, the error text, network type, Global Routing mode, active Config, server source and recent changes. Then troubleshoot in order: manual connection baseline → server connection → actual request → rule match → DNS and app differences. Retest the same destination and note whether each step changed the result. Restore the original value after a failed test before trying the next change; don’t pile unverified changes on top of one another.

When contacting your provider, share only the details needed to diagnose the issue and hide credentials. For client or system problems, use the exact error text to continue through the official manual. Follow the setup tutorial for your first connection; use the download guide to get or reinstall the app, and the Q&A for specific questions. The eight stages on this page can be read independently, but troubleshooting works best from basic conditions through rules and settings. Don’t change higher-level policies before confirming the server details.