If you have a working configuration, use Config for everyday routing. Switch to Proxy briefly to check whether routing rules are causing an issue, or use Direct to test whether your local network can reach a destination directly. This guide is for users who have imported their own subscription or nodes. It uses the Home, Config, and Settings screens on iPhone and iPad; Mac, Windows, Android, and Linux users can apply the same principles but should follow the controls in their own apps.
Understand Global Routing, nodes, and the connection toggle
Shadowrocket’s Global Routing setting determines how connections are routed; it does not provide a route or server. The selected node in Home determines which proxy endpoint handles traffic that needs proxying. The connection toggle controls whether the proxy tunnel is active. Selecting Proxy does not mean a connection is already established, and selecting Config does not guarantee that the active configuration is correct.
Break each connection into three questions: Does the app’s traffic enter Shadowrocket? How does the selected mode route it? Can the chosen endpoint reach its destination? Checking these layers first is usually more useful than repeatedly switching nodes when a site won’t load.
The interface may also offer Scene, which applies routing settings based on specified conditions. It is not a fourth fixed mode for routing everything directly or through a proxy. For an initial test, choose Config, Proxy, or Direct explicitly to avoid automatic scene changes affecting the results. Once you’ve confirmed the basic behavior, check the Scene conditions separately.
Config, Proxy, and Direct: differences and when to use each
Config
RecommendedRoutes connections according to the rules in the active configuration. Connections matching DIRECT go directly; those matching PROXY or a policy use the specified endpoint. Connections that match no earlier rule are handled by a fallback such as FINAL—not automatically sent directly.
Best for: Everyday access to both local and remote resources with a verified configuration.
Proxy
Sends all eligible traffic through a proxy endpoint instead of applying ordinary routing rules to choose DIRECT or PROXY for each connection. If a node is unavailable, switching to this mode won’t fix the node.
Best for: Briefly testing a proxy endpoint or checking whether Config rules are causing different access results.
Direct
Sends eligible traffic directly to its destination without relaying it through the selected proxy node. This does not turn off the connection toggle, and it is not a substitute for testing your network with the tunnel stopped.
Best for: Checking whether your current Wi-Fi or cellular network can reach a destination directly.
“All traffic through a proxy” and “all traffic direct” describe routing modes. They do not guarantee that every system connection on the device is captured or that DNS queries follow the same path as web traffic. DNS settings, tunnel exclusions, local network access, and platform networking behavior can all affect the result. Check the recorded destination and endpoint for the connection you’re troubleshooting.
Switch modes in Home and check the active Config
These steps use an iPhone or iPad as the reference. Before you begin, make sure you already have a service and complete node details of your own; this guide does not cover finding or recommending routes. A node list and a set of routing rules are separate things: successfully importing a subscription does not prove that Config contains suitable rules.
- Open Home and note the selected node and connection-toggle status. If the node belongs to a policy group, note which member is currently in use so the endpoint doesn’t change during testing.
- Go to Home → Global Routing and check whether Config, Proxy, Direct, or Scene is selected. Choose the mode that fits your test, then return to Home.
- To use Config, open Config and check which configuration file is active. Review its rules and policy names. Importing a file without enabling it does not mean it is being used for routing.
- Return to Home, confirm the intended node is selected, and turn on the connection. If the system asks you to authorize adding a VPN configuration the first time, follow its prompts.
- Retry the destination. If needed, close and reopen the page or app to clear its existing connection. An established long-lived connection may not switch endpoints immediately when you change modes.
How Config matches rules: from DOMAIN-SUFFIX to FINAL
Config is not simply a matter of separating “domestic” and “international” traffic. It matches connections against rules in the order defined by the configuration. In general, put specific exceptions first, broader conditions later, and a fallback last. The examples below show rule syntax and order only. They use reserved example domains and private address ranges, and are not a complete replacement for your existing configuration.
[Rule]
DOMAIN-SUFFIX,example.com,PROXY
IP-CIDR,192.168.0.0/16,DIRECT,no-resolve
GEOIP,CN,DIRECT
FINAL,PROXY
DOMAIN-SUFFIX,example.com,PROXY: Matches example.com and its subdomains, then sends matching connections to PROXY. Other domains won’t use a proxy just because of this rule.IP-CIDR,192.168.0.0/16,DIRECT,no-resolve: Matches this private IPv4 range and connects directly.no-resolvemeans the client won’t actively resolve a domain just to match this IP rule.GEOIP,CN,DIRECT: Matches the destination IP using GEOIP data and connects directly. A domain may resolve to different IP addresses and therefore match different IP rules; this does not determine location from a website’s language.FINAL,PROXY: Handles connections that didn’t match any earlier rule. Changing it toFINAL,DIRECTchanges how unmatched traffic is routed, so it’s not accurate to say that Config sends everything else directly.
The rule action may also use a policy name defined in the configuration. Check not only that the domain is correct, but also that the policy exists, its referenced node works, and no broader rule earlier in the list matches first. Verify each item in your existing configuration; don’t replace the entire file with the four example rules just for testing.
Subscription updates usually refresh node information included in the subscription. Whether rule files update depends on the configuration source and its update settings. A change in the number or names of nodes in Home does not mean that DOMAIN-SUFFIX, GEOIP, or FINAL rules in Config have also changed. After updating, return to Config and verify the active file and its actual rules.
Use two comparisons to distinguish rule issues from endpoint issues
Keep the device, network, node, and destination the same throughout testing, and change only one variable per test. Don’t update a subscription, change DNS, switch protocols, and change Global Routing all at once. Even if access returns, you won’t know which change fixed it.
Test plan: compare routing modes first, then networks
Same network: compare Config with Proxy
- On Wi-Fi, select an existing node and reopen the destination in Config.
- Switch to Proxy, keep the same node, and start a new connection.
- If only Proxy works, first check Config’s matching rule, policy reference, and DNS result.
- Note whether the failure occurs during name resolution, connection setup, or page transfer.
Same mode: compare Wi-Fi with Cellular
- Keep Proxy and the same node selected, then record the result over Wi-Fi.
- Switch to Cellular, reconnect, and open the same destination.
- If only one network fails, check whether that network can reach the node’s address and port.
- Also use Direct to test a site that should be reachable directly and confirm the local network is working.
If Config fails but Proxy works, that points to a rule or resolution path worth checking first; it doesn’t prove the rules are wrong. Confirm the result in the record for a new connection.
For example, if the same HTTPS page times out in Config but loads in Proxy, check whether Config routes it to DIRECT or to another unavailable policy. If both modes fail but Direct opens a site that is normally accessible without a proxy, that only shows that some basic network access works. Check the proxy endpoint and destination site separately.
Keep the ports distinct, too. HTTPS websites commonly use destination port 443; the node’s service port depends on your existing settings and may not be 443. Global Routing cannot correct an invalid node port, authentication details, or TLS parameters. A successful latency test also doesn’t guarantee that DNS resolution, the TLS handshake, and data transfer for a web page will all work.
Common questions: what to check when switching modes doesn’t help
Why does everything still seem to use a proxy in Config?
Open Config and check the active file’s rule order and FINAL rule. If most requests match no earlier rule and the fallback is FINAL,PROXY, they will mostly use the proxy. You may see similar results if all members of a policy group point to the same node. Check the match record for a new connection; don’t infer the route from the page language or loading speed alone.
A page loads in Proxy but not Config. Should I keep using Proxy?
You can leave Proxy on briefly to complete a necessary comparison, but then check the target domain’s rule, policy reference, and resolution result in Config. Save a copy of the current configuration before making changes, change one thing at a time, and create a new connection to test each change. Choose a mode for the long term based on your needs; bypassing rules temporarily does not mean they have been fixed.
Why is the VPN indicator still visible after switching to Direct?
Direct changes the routing mode; it does not turn off the connection toggle in Home. To compare your network with the tunnel completely stopped, turn off the connection in Home and check whether Settings → On Demand reconnects automatically. The system status indicator shows tunnel status; by itself, it doesn’t prove that a particular request used a proxy node.
Will switching to VLESS or Hysteria2 change Config rules?
No. Switching protocols does not automatically change the rules. Shadowsocks, VMess, VLESS, Trojan, Hysteria2, and WireGuard refer to different connection protocols or tunneling methods; Config, Proxy, and Direct are routing modes. Changing protocols requires server support and the correct node settings—it is not a substitute for changing Global Routing. First confirm that the node works, then test routing.
Checks to run before returning to everyday use
When troubleshooting is done, restore the intended operating state rather than just confirming that one page loads temporarily. If you have suitable rules, return to Config for everyday use. If you specifically need all traffic proxied or routed directly, you can choose Proxy or Direct, understanding the effects of each.
- In Home → Global Routing, confirm the intended mode is selected—not a temporary test option.
- In Config, confirm the expected file is active and check FINAL and important domain exceptions.
- In Home, confirm the intended proxy node or policy member is selected, not a test endpoint.
- In Settings → On Demand, restore the automatic connection conditions you need.
- Test a direct-access destination, one that requires a proxy, and any relevant local network resources. Confirm that each behaves as expected.
This guide covers client settings; it does not provide, sell, or recommend nodes or subscriptions. Shadowrocket is a closed-source commercial app, and getting the app is separate from arranging your own routing service. The iPhone and iPad app costs $2.99 as a one-time purchase on the App Store. The developer is Shadow Launch Technology Limited; check the App Store listing for system requirements. The Mac version (Apple silicon and Intel), Windows, Android, and Linux versions are free. See the site’s download guide for platform-specific options.