The short version
- When a conversion request takes your original subscription URL as input, the converter touches that input. Project documentation also supports file-path input, so not every conversion method means submitting a URL to a public site.
- A conversion request produces a target subscription format. It is not a general repair for plan state, route state or client faults.
- The shortest path is usually the provider's native format into the matching client. Consider conversion only when a format genuinely does not fit.
- If node count, rules or update behaviour changes after conversion, record the difference. Current public material does not provide a complete compatibility matrix, so do not assign a specific technical cause.
- Mask subscriptions, conversion addresses, QR codes, logs and screenshots. If you suspect exposure, rotate the token in the provider panel — deleting a message is not rotation.
What a conversion request actually carries
In the open-source subconverter documentation, a request names a target format and supplies the subscription to convert through a url parameter. Merging several subscriptions means several URLs are handed to the program as input. That describes what the program needs in order to read upstream content. It says nothing about whether any particular public deployment is trustworthy.
| Input or action | What the converter may do | What that does not prove |
|---|---|---|
| Original subscription URL | Fetch nodes and configuration content | That the service keeps no URL or logs |
| Target type | Request output for a supported target value | That every protocol or field converts without loss |
| Multiple subscriptions merged | Aggregate several upstream payloads | That nothing duplicates, conflicts, or leaks more tokens than before |
Mihomo's proxy-providers documentation is often cited here, and it is worth reading for what it actually solves: remote providers, save paths, filtering and health checks. It is a way to organise and refresh configuration. It is not an argument that a third-party online converter is required.
Decide whether you need conversion at all
- Check whether the provider panel already offers a subscription adapted to your client.
- Import the native address once and record what happened: an HTTP error, a parse error, or an empty node list.
- Identify what the address returned: a complete configuration, a node list, a login page, or an error page.
- Only after establishing “the upstream is valid but this client cannot parse this format” does conversion become a reasonable step.
- If your goal is only to change rules or rename nodes, check first whether the client supports overrides, providers or a local configuration file.
Native format, local conversion, self-hosted, public site
The table lists the main checkpoints in the decision. It is not a complete inventory of who can access data on any specific deployment — the actual parties depend on that deployment's configuration, logging and permissions.
| Method | Who can touch the original subscription | Advantage | Cost or main risk |
|---|---|---|---|
| Provider's native format | Provider and the local client | Shortest path, clear responsibility for format | Provider may not offer every client format |
| Client reads it directly | Provider and the local client | No extra processing step | Some fields or protocols still unsupported |
| Local conversion | Your machine and the program you chose | The original subscription never goes to a public site | You install, update and verify the tool yourself |
| Self-hosted conversion | Provider, your server, your client | You control deployment and logging policy | Server security, updates and logs are your responsibility |
| Public online conversion | Provider, the conversion site, your client | Convenient | Page promises cannot establish logging, caching or access control |
“Local” and “self-hosted” are not automatically safe either. Verify the project source, the version, the listening address, access control, logging, update path, and whether the output URL is publicly reachable.
When the output differs from the input
Record the input format, target format, converter and version, whether an external configuration was applied, and the exact observed difference — then consult the documentation for that version. Public material does not supply a full format compatibility matrix, so “it changed after conversion” is not a diagnosis, and a missing node is not automatically a dead node.
Change one variable at a time: validate the original address, then the conversion request, then the client import. Do not swap client, rules, DNS and system proxy together.
| Symptom after conversion | First thing to record | What remains unproven |
|---|---|---|
| Fewer nodes | Protocols supported by the target core, the filter rules, count before and after | Whether the missing nodes are dead, filtered, or unsupported |
| Rules behave differently | Whether an external rule set or template was applied | Whether the converter or your template caused it |
| Client refuses the output | The first line the client rejects and its parser version | Whether the upstream was valid in the first place |
| Still no traffic | One real HTTPS request after import | Anything about the plan or the server |
A record you can actually check later
Keep the date, client and version, upstream format, target format, conversion method, converter and version, node count before and after, skipped protocols, the first error text, and whether external configuration was used. For the subscription URL, record only the domain and a masked identifier; never write the token, UUID, password, QR code or full query string into a public record.
If you suspect the subscription leaked
The conversion input, the output address, the QR code and any public log all fall inside the exposure scope. The full handling sequence — remove public content, rotate the token, update clients, check for anomalies and validate the old address — belongs to the subscription-leak procedure rather than being repeated here.
Scope and limits
This page provides a framework for deciding whether conversion is needed and where the trust boundary sits. It does not assess whether any specific public converter logs data, and it does not promise that any protocol converts without loss. Project documentation describes functionality; it is not an endorsement of a third-party deployment, its operator, or its logging policy.
Sources: the subconverter README for target, subscription URL and multi-subscription inputs; the Mihomo proxy-providers documentation for provider paths, filtering and health checks.