1. Scope, goals, and the split-routing model
This guide is for users who already run Clash Verge Rev, Clash Verge, Mihomo, or another Clash-compatible client and want a cleaner streaming setup for Netflix and Disney+. The objective is not to send every connection through one overloaded VPN route. Instead, you will create a dedicated policy group for streaming traffic, place the relevant domain rules above broader regional rules, and keep local websites, banking services, work tools, and nearby services on DIRECT whenever that is the better choice.
Streaming platforms are more complicated than a single video domain. A Netflix page may contact separate hosts for authentication, artwork, subtitle files, playback manifests, telemetry, and content delivery. Disney+ behaves similarly, with account services, static assets, DRM-related requests, and video segments sometimes being delivered through different hostnames or CDNs. If only the homepage is routed through a proxy, the page may load while playback fails, subtitles remain unavailable, or the application repeatedly returns to its sign-in screen.
A split-routing design gives you three useful controls. First, streaming domains can use a selected relay or a streaming-specific group. Second, local and domestic services can remain direct, which reduces latency for ordinary browsing and prevents local sites from taking an unnecessary international path. Third, all unclassified traffic can use a predictable fallback policy instead of accidentally inheriting the streaming route.
This article does not guarantee access to every catalog, and it does not replace the terms, licensing rules, or account policies of Netflix or Disney+. Catalog availability can depend on account country, billing information, device location, IP reputation, and platform-side decisions. Treat the routing configuration as a way to make your network behavior consistent, not as a promise that a particular title will appear in a particular region.
2. Why a simple “VPN on” toggle often fails
A binary VPN switch is convenient, but it hides the decisions that matter during streaming troubleshooting. When everything is tunneled, local search, office portals, payment pages, software repositories, and video calls all compete with the streaming session for the same exit. When everything is direct, the streaming service may see an unsuitable route, and the application can fail before the player even requests a video segment.
The first common symptom is a working login but a failed playback attempt. The browser may reach the main Netflix or Disney+ page through one route while a media or license request follows another route because of a hostname that your configuration does not cover. The second symptom is inconsistent behavior between devices. A browser may follow the system proxy, while a smart TV application, Android app, or desktop media process uses its own network stack. The third symptom is poor performance: a relay that is excellent for a short web request may be a poor choice for a long, high-bandwidth video session.
DNS handling is another important layer. With Mihomo fake-IP mode, the client may return synthetic addresses and then associate them with domain names internally. That can work well when the rules, DNS mode, and traffic mode agree, but inconsistent settings can produce certificate errors, region mismatches, or requests that appear in logs under unexpected destinations. If you change DNS mode while testing, restart the relevant browser or application; old connections and cached resolutions can otherwise make a correct rule look ineffective.
Streaming also uses long-lived HTTPS connections and may prefer HTTP/2 or HTTP/3. A route can be technically reachable but still unsuitable if it suffers packet loss, unstable UDP handling, or repeated connection resets. Measure playback quality rather than relying only on a ping result. A node with a low ICMP latency may perform poorly during sustained downloads, while a slightly slower node may provide a much better start time and fewer quality drops.
The advantage of Clash is that these choices are visible. You can inspect the matched rule, change only the streaming policy group, compare relay nodes, and leave unrelated traffic untouched. That makes troubleshooting more precise than repeatedly switching an all-or-nothing VPN connection.
3. Inventory Netflix and Disney+ traffic before editing YAML
Start with observation rather than copying a large hostname list from a forum. Open the service in a clean browser profile, sign in, start a title, change subtitle or audio settings, pause and resume the video, and note the requests that appear during each action. In Clash Verge Rev, the Connections or traffic view can show the domain, process, rule, and selected proxy group. Record the domains that actually appear on your device and configuration.
For Netflix, you may encounter domains associated with the main web service, authentication, artwork, player assets, and CDN delivery. Common examples seen in different environments include netflix.com, netflix.net, nflxvideo.net, and nflximg.net. These names are representative rather than permanent instructions. Netflix can add, remove, or redirect traffic through different providers, so a suffix that worked last year may be unnecessary or incomplete today.
Disney+ traffic can include disneyplus.com, disney-plus.net, and related asset or delivery domains, but the exact set varies by application, region, and CDN path. Do not assume that every Disney-owned domain should use the same policy. A domain may host general corporate content, help pages, advertising, or regional services that are better left direct. Route only what your playback test demonstrates to be relevant.
Use the browser developer tools as a second source of evidence. In the Network panel, filter by Doc, Fetch/XHR, and media-related requests, then inspect the host column after pressing Play. You do not need to copy every analytics request. Focus on domains whose route changes cause a visible result: login completion, catalog loading, playback initialization, subtitle loading, or stable bitrate.
- Record the client: Chrome, Firefox, Clash Verge Rev, Android, or another application may produce different requests.
- Record the action: login, homepage, playback, subtitle selection, and account management can use different hosts.
- Record the matched rule: a broad
GEOIPor final proxy rule may be catching traffic before your intended streaming rule. - Record the selected node: node location, congestion, UDP support, and IP reputation can affect playback independently of YAML syntax.
Keep this inventory in a small dated note. When a service changes its CDN or redirects authentication, you can compare a new capture with the old one instead of replacing the entire configuration blindly. This also prevents an overly broad rule such as DOMAIN-SUFFIX,googleapis.com or a vendor-wide suffix from sending unrelated traffic through your streaming group.
4. Design policy groups for selection and fallback
A practical configuration normally separates the streaming policy from the ordinary proxy policy. The streaming group can contain a manual selector, a low-latency fallback group, and a direct option for testing. The direct option is useful because it gives you a control experiment: if Netflix or Disney+ fails through the relay but works directly, the issue may be the relay’s IP reputation or route rather than the domain rules.
A simple Mihomo-style structure could look like this. Replace the proxy names with entries that exist in your own profile; the names below are examples only.
proxy-groups:
- name: STREAMING
type: select
proxies:
- Streaming Relay 1
- Streaming Relay 2
- Auto Select
- DIRECT
- name: Auto Select
type: url-test
url: https://www.gstatic.com/generate_204
interval: 300
tolerance: 80
proxies:
- Streaming Relay 1
- Streaming Relay 2
- name: PROXY
type: select
proxies:
- Auto Select
- DIRECT
rules:
- DOMAIN-SUFFIX,netflix.com,STREAMING
- DOMAIN-SUFFIX,netflix.net,STREAMING
- DOMAIN-SUFFIX,nflxvideo.net,STREAMING
- DOMAIN-SUFFIX,nflximg.net,STREAMING
- DOMAIN-SUFFIX,disneyplus.com,STREAMING
- DOMAIN-SUFFIX,disney-plus.net,STREAMING
- MATCH,PROXY
The important design decision is not the exact group name. It is the separation of responsibilities. STREAMING should be easy to change while you are watching a title. PROXY should remain your general fallback and should not silently override the explicit streaming lines. Put specific service rules before broad rules, because Clash evaluates rules from top to bottom and uses the first match.
If your subscription already provides proxy groups, do not duplicate them without checking the generated configuration. Some clients use anchors, templates, or remote providers that regenerate the file when you update a subscription. A hand-edited rule can disappear after the next update. Prefer a supported override mechanism, or keep a small patch file that you can reapply and verify.
| Traffic type | Recommended policy | Reason |
|---|---|---|
| Netflix and Disney+ playback domains | STREAMING | Allows quick relay testing without changing all traffic. |
| Local banking, government, and work portals | DIRECT | Reduces unnecessary latency and avoids regional security checks. |
| General international websites | PROXY or a separate selector | Keeps everyday routing independent from streaming experiments. |
| Unknown destinations | Explicit fallback | Prevents accidental routing through a surprising group. |
5. Configure DNS, client mode, and platform behavior
Rules can be correct while traffic still behaves unexpectedly if the client mode is inconsistent. For browser testing, system proxy mode is often sufficient. It lets ordinary applications that respect the operating system proxy follow the same policy without forcing every packet through a virtual interface. TUN mode is more appropriate when you need to capture applications that ignore system proxy settings, including some desktop launchers, games, media apps, and command-line programs.
TUN mode should be enabled deliberately. It can capture traffic that you did not intend to proxy, and it may interact with another VPN, endpoint security product, virtual machine, or corporate network agent. Before enabling it, write down how to disable it and confirm that your Clash client has a working DIRECT path. If a streaming application still bypasses Clash in system mode, test TUN with only one network interception tool active.
For DNS, keep the configuration understandable while diagnosing. Make sure the resolver mode matches the expectations of the profile and that DNS queries are not escaping through a different network path from the connections they support. If you use fake-IP mode, check that the relevant domains are not placed in an unsuitable fake-IP filter. If you use redir-host mode, confirm that the returned addresses and rule matching behavior are consistent with your client version.
Mobile applications require extra care. Android apps may use certificate pinning, QUIC, IPv6, or background services that do not behave like a browser tab. Start the title while observing the Clash connection list, then switch the selected STREAMING node and repeat the test. On iOS or TV platforms, configuration delivery and traffic capture can be more restricted; use only the client modes and profiles supported by that platform rather than assuming desktop instructions transfer directly.
Do not judge the setup from the first page load alone. Test this sequence: open the application, sign in, load the catalog, start a title, seek forward, change subtitles, pause for several minutes, resume, and then start a second title. A rule set that handles the homepage but fails during resume or subtitle retrieval is not complete. Check whether the failing connection is a new hostname, a DNS resolution issue, or a node that loses the long-lived session.
6. Test performance, privacy, and maintenance
Choose nodes by sustained behavior rather than by location labels alone. During playback, observe start time, buffering, bitrate changes, and whether the player drops to a low quality after several minutes. Test at least two nodes in the STREAMING group at different times of day. A relay that works at noon may become congested during the evening, and an automatic selector based only on a small URL test may choose a node that is fast for tiny requests but weak for continuous video delivery.
When playback fails, isolate one variable at a time. First confirm that the domain appears in Clash connections. Next confirm the matched rule and group. Then try another node in the same group. After that, compare STREAMING with DIRECT if the service is normally available directly in your region. Only after those tests should you change DNS mode, TUN settings, or large portions of the rule set. This order avoids turning a simple bad node into a long, confusing configuration exercise.
Typical symptoms point to different causes. A catalog that never loads often suggests authentication or core web domains are missing. A title that opens but never starts may indicate a media, license, or CDN hostname using the wrong policy. Subtitles that fail while video plays can involve a separate asset request. Frequent buffering usually points to relay congestion, packet loss, or an unsuitable transport rather than a missing domain rule. A login loop can be caused by DNS inconsistency, cached cookies, system time errors, or an exit IP that the service distrusts.
Keep privacy in mind when selecting a relay. A proxy provider can observe connection metadata, and a compromised node could create risks beyond poor video quality. Do not reuse streaming credentials for unrelated services, enable multi-factor authentication where available, and avoid importing unknown YAML snippets that contain remote providers, scripts, or forced DNS settings. Review changes after every subscription update, because a provider may reorder rules or replace a policy group without preserving your local assumptions.
Compared with single-purpose VPN apps that expose only one global switch, Clash provides a more practical control surface for this scenario: separate Netflix and Disney+ traffic into a selectable group, retain DIRECT routes for local services, inspect the exact matched rule, and change a relay without disrupting the rest of your day. If you want that level of control across Windows, macOS, Linux, Android, and other Mihomo-compatible environments, the next step is to choose a supported client and import your profile carefully.
Ready to get started?
Install a compatible Clash client, import your profile, and test Netflix and Disney+ routes with a dedicated streaming policy group.
Build separate streaming, local, and general proxy routes across desktop and mobile clients.
Free Download