Official Manual Rules & Configuration

Shadowrocket Rules Reference

From configuration sections to the first matching rule, find the conditions, policies, and limits relevant to your question.

This reference is for users who know how to connect and are ready to edit a configuration. New to Shadowrocket? Start with the Getting Started guide, import your existing subscription, and verify the connection before returning to the rules. The examples assume you already have your own subscription or server details. Get subscription links from your provider.

Shadowrocket is available for iPhone / iPad, Mac, Windows, Android, and Linux. The iPhone / iPad version costs $2.99 as a one-time App Store purchase; Mac / Windows / Android / Linux versions are free. A one-time app purchase does not include a service plan. See the download guide for platform-specific options. For iPhone / iPad, check the App Store listing for system requirements. Instructions below use the iPhone / iPad interface; menu locations and available settings on other devices depend on the app as currently displayed.

Configuration file structure

Shadowrocket rule configurations are typically saved as .conf text files, with square-bracketed headers marking different sections. Understanding the sections matters more than memorizing a whole configuration: connection details specify how to reach a server, rules specify how to handle each request, and DNS resolves domain names to addresses. These settings can affect one another, but they are not interchangeable. Pasting a subscription link into a rules section or putting a domain rule in the Host section will not produce the intended result.

Start by distinguishing three things

Server entries in Home are used to connect; subscriptions manage groups of these entries; configuration files in Config control routing and related behavior. Updating a subscription does not necessarily change the active configuration, and selecting another configuration does not necessarily change the active server. When troubleshooting, note the subscription name, current server, and selected configuration separately so you do not mistake changes to different items for one operation.

Open the target configuration in Config and check whether you are editing a local copy or a remote configuration. Fetching remote content again may overwrite edits made directly to it. For changes you want to keep, make a local copy first and give it a recognizable name. After saving, confirm that the file is selected; a configuration appearing in the list does not mean it is active.

SectionPrimary purposeDoes not replace
[General]Overall settings such as DNSServer credentials and per-request routing
[Rule]Match requests and assign a policyCreating a working proxy server
[Host]Static address mappings for specified domainsChoosing whether a connection uses PROXY or DIRECT
[URL Rewrite]Processing URLs in requests the app can inspectGeneral network routing or reading arbitrary encrypted content

Read the minimum configuration

[General]
dns-server = system

[Rule]
DOMAIN-SUFFIX,example.com,PROXY
IP-CIDR,192.168.0.0/16,DIRECT,no-resolve
FINAL,PROXY

This snippet demonstrates how overall settings relate to the rules section. dns-server = system means to use system DNS; the rules below handle an example domain, a private IP range, and all remaining requests, in that order. There are no Host or rewrite sections because empty sections are not required. A shorter configuration makes it easier to connect each setting to its actual effect and provides a useful baseline for troubleshooting.

Commas in rules separate fields, so use standard ASCII punctuation. List the type, matching value, and policy in that order, with optional parameters at the end. Do not append explanatory text in Chinese or any other language to a policy. Put notes on a separate comment line to avoid copying invisible spaces or full-width commas. The domain field takes a hostname, not a URL with a protocol or a full web address.

Saving is not the same as validating

After editing, check that section headers are intact and rules remain in the correct section. Then save and reselect the target configuration. If the app reports a parsing issue, check the fields on the line you just changed before adding anything else. A successful save only means editing is complete. To confirm routing, make a new request and inspect the connection record.

Some changes affect only new connections; existing long-lived connections may continue using the old route. Before testing, close and reopen the test page, and reconnect Shadowrocket when appropriate. Do not repeatedly switch configurations during an important transfer. If a rollback does not appear to take effect immediately, rule out connection reuse and app caching before concluding that the rollback failed.

A complete configuration may contain more sections and settings than this guide covers. If you find an unfamiliar item, leave its value unchanged until you understand its purpose; do not concatenate complete files from different sources. Duplicate sections, conflicting settings, and remote updates overwriting local edits can be harder to diagnose than a typo in one rule. The examples below are for reference or local changes; you do not need to enable them all at once.

Matching order and Global Routing

Where a rule appears matters as much as what it says. In Config mode, the Rule section is typically checked from top to bottom. Once a request matches a rule, its policy applies; the app does not continue searching for a supposedly more specific rule. Shadowrocket does not automatically move DOMAIN ahead of DOMAIN-SUFFIX or reorder rules based on how specific they look.

Check Global Routing first

The three common Global Routing modes are Config, Proxy, and Direct. When learning custom rules, first confirm that Config is selected. If Proxy or Direct is active, requests may not follow the Rule section, but that does not by itself mean the rule syntax is wrong. The mode determines how the configuration is used; actual traffic is also subject to the system routing scope and what the app can handle.

Interface labelHow it worksUseful for checking
ConfigRoute traffic using the current configurationRule order and policy matches
ProxyRoute intercepted traffic through the proxyBriefly checking whether the proxy route works
DirectRoute intercepted traffic directlyChecking whether the local network can reach the destination

These modes are diagnostic variables, not performance ratings. If a page opens after switching to Proxy, that only shows that changing the route changed the outcome; it does not prove every DNS setting is correct. If a page fails in Direct mode, the local network may simply be unable to reach the destination. After each comparison, switch back to Config and verify the rules you intend to keep, so you do not mistake a temporary mode for a permanent fix.

Put specific exceptions first

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

When you visit login.example.com, the first rule matches, so the second one is never checked. For www.example.com, the first rule does not match and the second one applies. If you swap the first two lines, the suffix rule catches the login domain first, and the DIRECT exception never runs. This is an ordering issue; adding the same domain repeatedly will not fix it.

A common layout puts explicit exceptions first, followed by narrower service rules, broader address or region conditions, and FINAL last. This is a review-friendly convention, not a fixed law. When conditions overlap, set the order according to the desired result. For example, if a domain matches a keyword rule but has its own exception, put the exception before the keyword rule.

Separate rule matching from connection success

A rule match only means that a policy was selected. A PROXY match can still fail because the server is unreachable, authentication fails, or the transport is restricted. A DIRECT match can also fail if the destination refuses the connection. Treat “matched the right rule” and “the request succeeded” as separate conclusions. This avoids trying to fix server or website issues by endlessly changing rules.

A failed connection does not make Shadowrocket automatically try the next Rule. Later rules are not a failover list. To switch connections, check the server selection or policy group behavior you actually configured. Do not add several rules for the same domain with different policies and expect them to be retried in order. PROXY here refers to the current proxy selection; it does not mean automatic selection of the fastest route.

On Demand controls whether a connection is established under specified conditions; it does not replace per-request matching in Rule. Scene settings are also separate from the three Global Routing modes described above. If the connection state differs from what you expect after a network change, check the automatic connection conditions first, then inspect traffic routing. For more examples of first-match behavior, see Custom Rules and Priority.

Domain and request conditions

Domain rules are useful for specifying which policy to use for a service and are often easier to maintain than tracking changes to a website's IP addresses. First determine whether you have a hostname, a full URL, or a name shown by an app, then choose the rule type. A page title, app name, and the hostname used for a connection are not the same thing. Putting a Chinese display name into DOMAIN will not identify the corresponding network service.

DOMAIN and suffix boundaries

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

DOMAIN matches an exact hostname. The first rule applies to api.example.com, but does not also cover www.example.com or deeper subdomains. Enter only the hostname, without https://, a port, or a path. If you need to handle one specific endpoint, an exact rule has a more predictable scope and makes later domain changes easier to spot.

DOMAIN-SUFFIX matches the specified domain and its subdomains. The second rule can match both example.org and img.example.org, but it is not an arbitrary string-ending search; notexample.org is not a normal match. Do not add an asterisk in front: *.example.org is not the suffix syntax used here.

Consider the domain boundary before using a suffix rule. Applying one policy to a broad suffix on a public hosting platform may affect many unrelated sites. When you can specify a service's actual domain, do not widen the scope to a shared suffix. Conversely, if a service uses several subdomains, a rule for only the homepage hostname may not be enough; the homepage might work while images or login fail.

A keyword does not identify a service

DOMAIN-KEYWORD matches a keyword in a domain name and often has a broader scope than a suffix rule. The example keyword example may match several unrelated domains; that does not mean they belong to the same provider. Use keywords only when you know and can review the naming pattern. Common short words are not reliable classifiers. If a rule catches too much, replace it with a precise condition instead of piling on exceptions.

For example, directing ads to REJECT only demonstrates keyword blocking; it does not guarantee that only ad requests will match. Legitimate domains may contain those letters too. Check the actual domains in your connection records before deciding what to reject. A page loading does not prove its login, media, and submission endpoints are unaffected; test the actions you actually use.

What USER-AGENT can identify

[Rule]
USER-AGENT,ExampleClient*,DIRECT
DOMAIN-SUFFIX,example.com,PROXY
FINAL,DIRECT

USER-AGENT matches User-Agent content in requests the app can identify; the asterisk in the example indicates a wildcard. It does not identify a process or mean “all traffic from an app.” An app may use different request libraries and identifiers, while multiple apps may share similar identifiers. A single USER-AGENT rule cannot establish that you have complete app-level routing.

HTTPS encrypts request headers, and ordinary tunnel forwarding cannot freely inspect the User-Agent. Whether this rule can match depends on the information available to the app and the applicable processing conditions. Do not treat extra certificate trust or decryption settings as required just to make a rule work; this guide does not require enabling them. For routine routing, prefer verifiable domain conditions.

If a domain rule seems not to match, first check the actual destination used by the connection. Redirects, content delivery, separate authentication domains, and direct IP connections can invalidate your assumptions. If the client has no usable domain information, DOMAIN rules cannot infer it. Add conditions only when connection records show they are needed; do not assume every failure means the suffix rule is “not strong enough.”

For easier maintenance, keep a small set of well-founded rules for each service and put special exceptions first. After a service changes, remove obsolete domains so the configuration does not accumulate unexplained conditions. More domains do not mean more accurate coverage. A stable configuration is one where you can explain why every rule exists and what it should match.

IP and region matching

IP rules are more direct when a connection uses an IP address or when you need to handle local network addresses explicitly. They match the final destination address, not the spelling of a domain. Before using one, distinguish a single address, a subnet, and a regional database. A website may resolve to multiple addresses, and one address may serve multiple domains; IP routing may not align with a website's service boundaries.

IP-CIDR ranges

[Rule]
IP-CIDR,192.168.50.10/32,DIRECT,no-resolve
IP-CIDR,192.168.0.0/16,DIRECT,no-resolve
IP-CIDR,10.0.0.0/8,DIRECT,no-resolve
IP-CIDR,172.16.0.0/12,DIRECT,no-resolve
FINAL,PROXY

IP-CIDR handles IPv4 networks. The number after the slash is the prefix length, not a port or a count of addresses. /32 specifies one IPv4 address; /16 specifies a range with the first 16 bits fixed. Use the network boundary you actually need. Do not expand the range far beyond your local network just to make one device connect directly.

The first three example ranges are common private address ranges and can help explain local network routing, but they do not guarantee reachability on your current network. Corporate networks, guest Wi-Fi, and home routers may isolate devices or use different routes. A DIRECT match only tells the local network to attempt the connection; it does not grant access or automatically connect subnets that network equipment has isolated.

What no-resolve means

The point of no-resolve is to avoid actively triggering a domain lookup just to match that IP rule. It does not disable all DNS or guarantee that the rule will never see an address obtained through resolution. If the destination is already an IP, or the processing path already has the destination address, the rule can still evaluate it. This can avoid extra lookups made solely to check certain ranges, but the effect depends on the actual request path.

Suppose a request starts with only a domain name. If an earlier domain rule matches, there is no need to rely on a later IP rule to choose a policy. If no domain rule matches and the following IP rules all use no-resolve, do not assume the client will necessarily resolve the domain first and match a range. When writing or troubleshooting rules, distinguish between an address that is already available and a lookup that would need to be triggered for matching.

IPv6 and GEOIP

[Rule]
IP-CIDR6,fd00::/8,DIRECT,no-resolve
GEOIP,CN,DIRECT
FINAL,PROXY

IP-CIDR6 uses the same prefix approach for IPv6 addresses. The example uses part of a unique local address range to illustrate syntax; it is not a complete list of IPv6 local addresses. If a destination has both A and AAAA records, an IPv4-only range may not express your full intent. When testing, check which address family the connection actually uses rather than relying on the domain name alone.

GEOIP uses a regional database to classify the destination IP. In the example, CN is a region code; it does not refer to a Chinese-language website, a Chinese domain, or an account's region. Content delivery and deployments across regions can cause the same domain to resolve to addresses in different regions, and the database itself has update and classification limits. GEOIP is better suited to broad classification than to replacing explicit service exceptions.

For services that need a fixed policy, put domain or exact-IP exceptions first, then use GEOIP for the remaining traffic. A broad region rule at the top may prevent later service rules from matching. If you want GEOIP not to trigger resolution, understand how no-resolve may change which requests have an address available for matching; do not copy the parameter without checking the result.

When troubleshooting address rules, record the destination domain, actual destination IP, address family, and matched rule. Do not assume an IP you looked up in a terminal is the only address the app will use: caches, network DNS, and content delivery can all produce different results. If the same domain sometimes uses a different policy, check whether these four details are consistent before changing ranges. This is usually more useful than continually widening regional rules.

Policies and FINAL

The left side of a rule says what to match; the policy on the right says what to do when it matches. PROXY, DIRECT, and REJECT are three different actions, not server protocols or connection quality ratings. A rule expresses its full intent only when both the condition and the policy are correct. You can review each line as a sentence: when the destination meets this condition, handle it in the specified way.

PROXY and DIRECT

PROXY means to use a proxy policy. The actual route depends on the current proxy selection or the configuration's policy mapping. It does not select a working server from a subscription, and it does not guarantee the destination will accept that route. After a PROXY match, check the current server, protocol settings, and reachability. If the server details are outdated, adding more PROXY rules will not fix an authentication issue.

DIRECT means to connect to the destination over the local network without using the selected proxy server. It does not mean Shadowrocket is disabled or guarantee that all system networking mechanisms are bypassed. Within the traffic the app handles, DNS and other configuration settings may still affect access. Use DIRECT for destinations that should be reachable locally, but do not assume direct access works from every location or network.

PolicyActionWhat to check next
PROXYSend traffic over the proxy routeCurrent server, authentication, and destination response
DIRECTSend traffic over the local network routeLocal routing, network permissions, and destination reachability
REJECTReject the matching requestWhether a required endpoint or related domain is being blocked

Limit the scope of REJECT

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

This example rejects one exact domain first, preserves a DIRECT exception for the login domain, and sends other example subdomains through the proxy. A request matched by REJECT does not automatically fall back to DIRECT. If an app keeps retrying, some content is blank, or a submit button does nothing, check whether a required endpoint was rejected rather than checking only whether the homepage loads.

REJECT rules need especially clear justification. For domains with an unclear purpose, inspect their requests and relationship to app features before rejecting them. Analytics, authentication, and configuration requests may share a base domain; broad suffix or keyword rules can affect essential features. To troubleshoot an unintended block, temporarily disable one rule and repeat the same action. This makes cause and effect easier to confirm than deleting an entire group.

FINAL defines the default behavior

FINAL does not need a domain or network field. Common forms are FINAL,PROXY and FINAL,DIRECT. It determines what happens when none of the earlier conditions match, making it the default policy for the configuration. The right choice depends on your needs and network; there is no single answer for every device. Before changing it, consider where unknown domains, background requests, and new services will go.

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

This snippet uses the pattern “send explicitly listed domains through the proxy; connect directly to everything else.” Changing it to FINAL,PROXY means all other requests use the proxy by default, so you must list DIRECT exceptions earlier. Either pattern can work, but they require different maintenance: the first requires attention to proxy destinations you might miss, while the second requires attention to DIRECT destinations you might miss. Do not use both assumptions to explain the same configuration.

Do not set FINAL to REJECT for everyday use unless you are prepared to maintain and verify an explicit allowlist. A fallback rejection makes new, unlisted domains fail; even a newly added page endpoint can cause problems. While learning, use an observable default policy to establish a stable baseline, then narrow conditions for a specific reason.

If your configuration refers to a custom policy name, confirm that it exists and is spelled consistently. This guide uses built-in policies to avoid that extra ambiguity. Do not assume a server's display name from your provider is a built-in policy keyword. Final validation should answer three questions: Does the condition match correctly? Does the action express the intended policy? Does the route work? All three must be confirmed.

General and DNS

General describes overall network behavior, and DNS is often the key variable when rules look correct but a page will not load. DNS resolves a domain to an address; routing rules then select a policy using the available information. They can affect one another in practice, but troubleshoot them as separate concepts. A normal proxy latency result only confirms the parts of the connection covered by that test; it does not prove that the domain resolved or the page's resources loaded.

Establish a DNS baseline for comparison

[General]
dns-server = system

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

dns-server = system is an example of using system DNS, not a performance recommendation for every network. If you already have a specific DNS configuration, save the current value before comparing alternatives. Do not paste a long list of DNS servers found online and assume that more servers are more reliable. Different answers, network reachability, and timeout behavior can make the problem harder to explain.

When using custom DNS, first check that the address format matches a form supported by the current app, then confirm that your local network or the relevant request path can reach it. DNS server reachability, a usable response, and destination reachability are three separate checks. Even a successful lookup can return an address that does not work over the current route; even a working proxy does not mean the DNS request used the proxy as you expected.

Distinguish three points of failure

First, no address is returned, often due to a lookup timeout or a domain-related error. Second, an address is returned, but connecting to it times out. Third, the connection is established, but the destination returns a certificate or application-level error. Each calls for a different fix. A browser message such as “page can't be opened” does not distinguish them; use available Shadowrocket connection records and tests made at the same time.

If the same domain resolves to different IPs on two networks, do not assume one result is wrong. Content delivery may choose addresses based on network location and resolution path. The relevant questions are whether the current address is reachable under the selected policy and whether it belongs to the expected service. After switching networks, make a new request and account for cached addresses before attributing every difference to rule order.

IPv6 is another important comparison. If a domain returns both IPv4 and IPv6 addresses, the system may try one first; the actual choice also depends on network conditions. If you checked only IPv4 rules but the connection uses IPv6, the observed policy may differ. Check the actual address family before adding rules; disabling an address family is not a universal fix for DNS problems.

Avoid changing multiple overall settings at once

General may also include bypass ranges, address-family options, and other overall behavior settings. These operate at a different level from Rule: if a type of traffic never enters the expected handling path, more precise domain rules cannot explain it. If you find settings such as skip-proxy, first confirm what they do and their current values. Do not replace them casually as though they were ordinary DOMAIN exceptions.

Change overall settings more cautiously than adding a precise rule. If you change DNS, address family, bypass ranges, and FINAL all at once, even a temporary improvement will not tell you what worked or reveal side effects. Keep the server and Rule settings fixed while changing one DNS setting; then restore DNS and adjust the target rule. Record each change, test target, and result so you can repeat the comparison.

If some pages will not load, choose one fixed target and record the stage of failure instead of switching between several sites. If only one app fails, compare whether the browser uses the same domain and address family. For a more detailed sequence, see DNS Troubleshooting. This chapter explains the concepts; the article walks through diagnosis based on symptoms.

Host and URL Rewrite

Host and URL Rewrite can both affect a request, but neither is an alternative form of a regular Rule. Host specifies the address to use for a domain; URL Rewrite processes a URL when request information is available; Rule selects a policy for traffic. Keeping these responsibilities separate avoids assumptions such as “changing a domain mapping should automatically use the proxy” or “a redirect should fix every connection.”

Host maps a domain to an address

[Host]
test.example.com = 192.0.2.10

[Rule]
DOMAIN,test.example.com,DIRECT
FINAL,PROXY

This example maps test.example.com to 192.0.2.10 and uses a separate Rule to specify the domain's policy. This address is reserved for documentation examples and does not provide a real service; do not use it to test page reachability. For real use, enter an address you are authorized to use and have confirmed is appropriate for the domain. Document why you added a temporary mapping and when to remove it.

A static mapping bypasses address changes from normal DNS updates, which also means it cannot follow automatic routing changes. After a service moves or its content delivery changes, the old address may stop working. If a domain works on other devices but fails with the current configuration, check for an outdated Host entry before adding another DNS server. After testing a temporary mapping, decide whether to keep or remove it.

For HTTPS, certificate validation still relies importantly on the domain being accessed. Mapping a domain to an arbitrary reachable IP does not give the server a valid certificate or the corresponding website. If you see a certificate error, do not mask an address mismatch by ignoring validation. Check whether the mapped address actually serves the destination; if needed, remove the mapping and compare using normal DNS.

URL Rewrite fields

[URL Rewrite]
^http://old\.example\.com/(.*)$ http://new.example.com/$1 302

This is an HTTP redirect example: a regular expression comes first, followed by the destination URL and the redirect status code. ^ anchors the match to the start of the string; dots in the domain are escaped with backslashes; (.*) captures the rest of the content; and $1 in the destination refers to that capture group. This example illustrates syntax only; neither example hostname is guaranteed to host a working page.

Keep regular expressions as specific as possible about the protocol, host, and required path. If you do not escape dots in a domain, they match any character; without an anchor at the start, the expression may match too broadly. Before editing, identify one URL that should match and one that should not, then check that the expression distinguishes them. Testing only a positive example can miss an overly broad match.

A redirect makes the client request a new address, so check for redirect loops. If the destination matches the original expression again, or another rule redirects it back to the old address, the request may loop. The new domain also needs to resolve normally and is subject to its own Rule policy. A successful rewrite does not mean the destination is reachable. Check separately whether the original request was processed, whether a new request was made, and how it was routed.

Encryption boundaries and sensitive parameters

Ordinary HTTPS forwarding cannot inspect the full encrypted request path and headers. A regular expression alone does not make arbitrary rewriting possible. This example uses HTTP to avoid mixing certificate trust and decryption requirements with basic rule behavior. Without a clear need and full understanding, do not change device trust settings just to try a rewrite snippet; basic domain routing does not depend on it.

A URL may contain login state or other sensitive query parameters. In the example, (.*) carries everything in its match scope to the new destination. Before using this in practice, check whether it could forward parameters that should not cross domains. Do not paste real service URLs, authentication links, or private configuration into public discussions. Use example domains to describe the structure and remove sensitive values while keeping the field format.

If you configure both Host and rewriting, validate each layer separately. First disable rewriting and confirm the original domain's address and route. Then enable rewriting and observe whether the destination changes. Finally, check the new destination's resolution and policy. Restoring one layer at a time helps distinguish an outdated static mapping, a regex that does not match, and an incorrect destination route. It also prevents a complicated rewrite from obscuring a simple DNS issue.

Validation, troubleshooting, and maintenance

The goal of rule maintenance is not to make the configuration longer; it is to be able to explain why each important request uses its current policy. A reliable workflow includes a backup, a single change at a time, positive and negative tests, and a rollback plan. Keep a working configuration, then create a test copy. If the basic connection is already unstable, address the server or network first instead of rebuilding all the rules at once.

Validate in a consistent order

  1. Confirm that the current server in Home works, Global Routing is set to Config, and the target file is selected in Config.
  2. Add or change only one rule. Save, then create a new request to the target so an existing connection does not continue using the old route.
  3. Check the target domain, actual IP, address family, matched rule, and final policy, then record the access result.
  4. Test a target that should be unaffected to confirm the new condition has not expanded beyond its intended scope.
  5. Keep the change if the result matches your expectations. Otherwise, roll it back, restore the baseline, and test the next hypothesis.

A positive test should clearly match the new rule. Choose a negative test as similar as possible to the positive one, but make sure it does not meet the condition. For a suffix rule, test a hostname under that suffix and a different domain; for an exact-domain rule, also check another subdomain. This comparison can catch a false success where an earlier, broader rule already matched and the new rule never took effect.

Isolate conflicts with a minimal snippet

[Rule]
DOMAIN,api.example.com,PROXY
FINAL,DIRECT

If a long configuration is hard to understand, temporarily keep one exact condition and a fallback policy in a copy. Confirm that the minimal snippet works as intended, then restore the other rules in groups. If the result changes after a group is restored, check whether that group matches the target first. Do not delete large sections from the original and try to rebuild them from memory; a test copy and clear backup give you a reliable starting point.

Observed behaviorCheck firstDo not conclude immediately that
All requests appear to use the same policyGlobal Routing, selected configuration, rules that match earlierDomain rules are all unsupported
Rule matches, but access failsDNS, current server, destination responseAdding the same rule again will fix it
Only some page features failRelated domains, REJECT, rewriting, and cacheThe whole site is fine if the homepage loads
Changes disappeared after an updateRemote configuration overwrites, local copies, and selection stateA subscription update always changes every configuration

Tell the two types of update apart

Subscription updates primarily maintain server entries provided by your service provider. Remote configuration updates may replace rule text; these are not the same operation. Before updating, note the object in use and its name. Afterward, check the server list and configuration contents separately. If you only need to learn how to refresh server entries, see the subscription update guide. Do not repeatedly import an entire configuration as a substitute for a normal refresh.

For long-term maintenance, a one-line purpose for each rule group is more useful than a vague note saying “fixed.” Explain what it affects, the intended policy, and why it should remain; do not include real credentials. When your network environment or a service domain changes, first check whether the rule is still needed. Removing obsolete conditions reduces overlaps and helps anyone troubleshooting later identify the relevant lines.

When sharing troubleshooting details, keep the rule type, relative order, and stage of failure; remove server credentials, private subscription URLs, and personal network information first. Get subscription links from your provider, and check your existing provider details for authentication or server-side parameter issues. This manual explains client configuration and does not rely on server details from unknown sources as a prerequisite for fixing a problem.

Finish with a complete check: test actual tasks such as loading pages, signing in, and transferring media or files, then confirm the behavior still matches your intent after switching networks. If you only ran a latency test, limit your conclusion to that test; do not claim the whole configuration is working. To return to the quick-start workflow, read the Getting Started guide. For platform options and common download questions, see the all-platform download guide. A stable configuration should be reproducible, explainable, and easy to roll back.