Importing a Clash Subscription Link: Format Detection, Update Failures, and Leak Response

If the client offers direct link, configuration link, URL Scheme, and other entry points, start with an ordinary remote link. If one method reports client error(Connect), try another import entry documented by the client, but do not leave “allow invalid certificates” enabled.

Clash cannot import an arbitrary web address; it needs a Clash- or Mihomo-compatible subscription from the provider. When the client accesses the link, the server may return YAML with proxies, proxy-groups, and rules, or it may return encoded text, a sign-in page, or an error page. The response content determines whether parsing succeeds.

Some subscription services return different formats based on User-Agent. A browser displaying a string does not guarantee that a Clash request receives the same content; conversely, a web page in the browser does not necessarily mean the link is invalid. Give the provider the client version, update time, and status code instead of repeatedly deleting the configuration.

Errors such as parse, yaml: unmarshal, or field not found mean the network request probably completed and the client failed while reading the content. Common causes include a server response formatted for another client, damaged YAML indentation, or fields unsupported by the current core.

Switching nodes usually does not help here. Preserve the first specific error line and line number, verify the client's Mihomo version, and ask the subscription provider for a compatible format. Do not manually repair dozens of fields in a remote configuration; the next update will overwrite every change.

If the remote update reports success but the Proxies page has no nodes, the response may contain only rules, all nodes may have been filtered out, or the old configuration may still be active.

Check the selected mark in Profiles, then whether the configuration contains proxies or proxy-providers and whether proxy-groups references them. Also check whether an override script removed every node.

Nodes can exist yet all sit in an unused group, leaving the primary policy on DIRECT. Choose a policy group that actually contains nodes and pin one, then observe a request in Connections. This reveals more about whether the configuration works than the latency button alone.

Clash normally continues using the last successfully downloaded local copy. Existing nodes may therefore keep working temporarily after a subscription expires or the server begins returning an error. Judge update status by the “last updated” time and the current error, not simply whether web pages open.

Do not rush to delete a working old configuration. Export the error, verify the account and link status, and remove the old entry only after the new address updates successfully. Deleting the sole working copy turns a diagnosable issue into an immediate outage.

Select the new Profile, pin a node in Proxies, and keep Rule mode. Enable the system proxy on a Windows/macOS desktop client. On Android, start the VPN through the client’s connection control and grant permission. Open a familiar page and check its domain, rule, and outbound route in Connections. A latency test does not replace a real request.

If a request reaches Clash but reports timeout, try another node. If there is no record, check the desktop system proxy or Android VPN. If it shows DIRECT, inspect the rule. A working page with a record that matches your selection confirms that the subscription has been downloaded, parsed, activated, and used to forward the request.

The token in a subscription address usually acts as a read credential. Troubleshooting needs only the error status, response type, and time, not the complete URL. Before sharing chat screenshots, screen recordings, or logs, hide credentials in the path or parameters. If you are unsure which parts are sensitive, hide the entire subscription address.

Changing only the display name of a Profile does not invalidate the old credentials. If the complete subscription has been made public, reset the address through the provider’s panel, then replace old links on devices using the steps below.