A profile is a complete configuration, not a single node

In Clash GUI clients, a profile usually means an entry in the configuration list. It can come from a subscription URL or a local YAML file. When you activate an entry, the client loads that configuration and may layer its own settings on top. As a result, the configuration running in the core may not match the downloaded file exactly.

  • proxies defines nodes; proxy-providers lets the main config load nodes from a separate data source.
  • proxy-groups defines proxy groups. In a manual selection group, you choose a member; an automatic latency-testing group selects one according to its own test settings.
  • rules determines which policy handles a matching connection. A domain rule can point to a proxy group or specify a built-in policy such as DIRECT.

Keep these three actions separate: switching profiles replaces the entire configuration; switching a proxy group member changes the route within the active configuration; enabling the system proxy or TUN changes how app traffic reaches the core. Switching profiles does not enable the system proxy, nor does it guarantee that all apps use the client.

Remote subscriptions vs. local YAML: different update sources

How the two profile sources differ
SourceWhat gets importedWhat changes later
Remote subscriptionThe client fetches the config from a URL, and the entry usually retains the update URL.Refreshing fetches the upstream content again; edits to a downloaded copy may be overwritten.
Local configImport a YAML file from disk; some clients copy it to their own data directory.If you edit the original file but the client uses a copy, you may need to reimport it or edit the client-managed entry.

A subscription URL does not always return a main config that Clash can load. Some URLs return only a list of nodes; others show a login page when access expires. Even a file with a .yaml extension cannot replace a full config with routing policies if it contains only node definitions. Before importing, choose the provider's Clash or Mihomo format, then check which core the client actually uses. Mihomo (formerly Clash.Meta) supports fields and protocols that may not work with older Clash cores.

Keep multiple subscriptions: identify each entry and confirm which one is active

For example, in the desktop version of Clash Verge Rev, import a remote entry by URL from the Subscriptions page in the sidebar. To import a local file, use New on the same page, choose the local type, and import it. Menu labels and card actions vary by version and language. The key is to keep each config as a separate entry instead of overwriting an existing one with a new URL.

  1. Give each entry a recognizable name. For example, “Primary · Remote,” “Backup · Remote,” or “Rule Testing · Local.” Use the source to distinguish entries, and never put a subscription URL containing a token in the display name.
  2. Select the entry you want to use. Choose the config on the Subscriptions page and confirm that the active indicator has changed. Then open the Proxies page and check the target config's proxy group names and current members. If both configs have a group called “Proxy,” check the member list instead of relying on the group name alone.
  3. Start a new connection. Use an app configured to route through the client to revisit a test destination, then check the rule, proxy group, or outbound fields on the Connections page. Existing connections may continue using their previous route or disconnect during a reload, so a new connection is a more reliable test.

Having two subscriptions in the config list does not mean their nodes are automatically combined. Usually, one main config determines the active rules and proxy groups. To use nodes from both sources in one group, the core must support the relevant data format, and the same main config must explicitly reference both providers. A subscription URL that returns a complete main config should not be used directly as a node provider URL.

Subscription updates and local edits: keep them separate

Refresh a remote entry and check that the new config is loaded

Find the remote entry on the Subscriptions page and use its card action to update it. Check the download status first, then confirm that the active config was applied successfully. Some clients reload an active entry automatically; others may require you to select it again. A URL can return a new file that the core still rejects because of unsupported fields or invalid policy references. Check the first config error in the logs; it is usually more useful than refreshing repeatedly.

Keep personal rules in an override or a separate local file

To keep receiving upstream updates, use the client's override or extension-config feature if available, and check whether it applies to the current entry or globally. Clients merge arrays differently: rules may be appended or replaced wholesale. Keep a copy of the original config, then inspect the final rule order. Rules are generally matched from top to bottom, so put specific rules before general rules that could match first.

If you just need an independently editable snapshot, save a copy of the remote content as local YAML and import it under a new name. The snippet below shows rules to add to an existing config; it is not a complete config. Put specific rules before the fallback and keep the existing fallback policy. Do not add a second rules key to the same YAML document.

rules:
  - DOMAIN-SUFFIX,example.com,DIRECT
  - MATCH,DIRECT

The example's MATCH,DIRECT sends any remaining matching traffic directly and is included only to illustrate rule order. When editing, keep the existing fallback if it points to “Proxy,” and make sure that group is still defined under proxy-groups. YAML uses spaces for indentation. If parsing fails, check the log's line number and make sure list items align with their parent key.

Back up and migrate configs: include everything they depend on

  • Back up the main config and overrides separately. Save the local YAML, extension scripts, or override files, and note which entry each belongs to. Saving only a subscription URL cannot recover older content that has since been replaced upstream.
  • Check file references. If a provider depends on local files, save those data files too. After migrating, make sure path still resolves relative to the core's working directory in the new environment. Do not assume absolute paths work across systems.
  • Record client settings separately. The system proxy and TUN switches, listening port, and current proxy group selection may be saved by the client independently. If another app points manually to 127.0.0.1:7890, check which port the new client actually listens on after migration. Do not assume the example port is the default.
  • Restrict access to backups. Subscription URL query parameters, node authentication fields, and some client exports may contain credentials. Store backups somewhere access-controlled, and redact sensitive details before sharing troubleshooting information.

To restore, import the backup as a new entry first and keep the working entry as a fallback. Confirm that the config loads, the proxy groups are present, and dependent files are readable. Then enable the traffic routing you need and test a new connection. If only one domain uses the wrong policy, check which rule matched. If no new connections appear in the client, first check whether the app is routed through the system proxy or TUN. These issues require different troubleshooting steps.