Clash Notion Figma Miro Workflow: Best Setup Tips 2026

Why Notion, Figma, and Miro Need Coherent Clash Routing

A remote-work workspace rarely talks to one hostname. Notion may load its application shell from one collection of domains, fetch images and file attachments from another, and maintain long-lived synchronization connections in the background. Figma adds collaborative document data, comments, fonts, thumbnails, prototype assets, and real-time editing sessions. Miro combines board content, embedded media, presence indicators, invitations, and export jobs. To a user, each product looks like one web application; to Clash, each page load may be a sequence of DNS lookups, HTTPS requests, WebSocket sessions, and object-storage downloads.

That distinction explains why “the homepage opens” is a weak connectivity test. You can open Notion but fail to save a block, open Figma but lose multiplayer cursors, or load a Miro board while images remain blank. A browser tab may also keep a cached service worker or previously resolved address, making the first visual impression look healthy while a new collaboration session is already failing. The useful question is not whether an app opens, but whether its complete workflow—login, workspace discovery, editing, synchronization, asset loading, sharing, and export—uses a predictable route.

The best Clash Notion Figma Miro workflow therefore starts with policy design rather than a giant list of guessed domains. Decide which services need a selected proxy group, which local resources must remain DIRECT, and which general web traffic should follow your normal fallback policy. A narrow rule that clearly expresses intent is easier to audit than a broad rule that accidentally moves banking, printers, internal dashboards, or local file servers through a remote exit.

Build a Service Policy Before Writing YAML

Start by drawing four traffic categories. The first is collaboration traffic: application domains, API endpoints, synchronization hosts, and WebSocket destinations required by Notion, Figma, or Miro. The second is supporting delivery traffic: content delivery networks, image hosts, font resources, update endpoints, and object storage used by the applications. The third is local and organizational traffic: your router, NAS, office printer, company intranet, self-hosted Git service, and private DNS suffixes. The fourth is everything else, which should use a general policy group rather than being silently classified as part of a productivity application.

This separation prevents a common mistake: adding one obvious hostname and assuming the entire product follows it. For example, a Figma document can render its interface while a separate asset host serves blank thumbnails. A Miro board can display its title while a persistent socket fails to deliver new cursors. Notion can show cached pages while newly edited content waits indefinitely for synchronization. When rules are organized by workflow, you can test each function independently instead of reacting to vague complaints such as “Figma is slow.”

Choose a dedicated group such as WORK-COLLAB for the collaboration category. A selector is a good default when you want deliberate control over geography or provider. A url-test group is useful when several permitted nodes offer similar access and you want Clash to prefer the healthiest measured path. A fallback group is more conservative: it keeps an ordered primary and backup list and moves only when health checks fail. Do not confuse a url-test winner with a universal performance guarantee; a short probe may be fast while a long WebSocket session experiences congestion, packet loss, or idle timeouts.

Keep the group name stable across desktop clients. Clash Verge, Clash Verge Rev, Mihomo-based clients, and other maintained interfaces may display the same YAML differently, but consistent policy names make shared configuration easier to understand. If one device calls the group REMOTE-APPS and another calls it FigmaProxy, troubleshooting becomes a translation exercise. Stable names also help you compare logs when a laptop and a home-office desktop appear to experience different failures.

Separate Local Destinations from Cloud Services

Remote collaboration often touches local infrastructure. A Figma designer may upload a file from a local folder, a Notion user may paste a link to an internal document, and a Miro workshop may run beside a video call, printer, or meeting-room display. These destinations should not inherit a broad proxy rule merely because the browser is in a “work” session. Keep private address ranges, loopback traffic, local hostnames, and known corporate suffixes on DIRECT when your network policy requires it.

Be careful with DNS behavior. In redir-host mode, the application may receive a real address and then connect normally; in fake-IP mode, the client may see a synthetic address that Clash maps back to the queried hostname. Both designs can work, but inconsistent exclusions create confusing symptoms. A local NAS may become unreachable when its hostname is captured by a fake-IP rule, while a cloud service may resolve locally and bypass the intended policy. Record whether the failure is DNS resolution, TCP connection, TLS negotiation, or application synchronization before changing several settings at once.

proxy-groups:
  - name: WORK-COLLAB
    type: select
    proxies:
      - Preferred-Work-Node
      - Backup-Work-Node
      - DIRECT

rules:
  - DOMAIN-SUFFIX,internal.example,DIRECT
  - IP-CIDR,192.168.0.0/16,DIRECT,no-resolve
  - DOMAIN-SUFFIX,notion.site,WORK-COLLAB
  - DOMAIN-SUFFIX,figma.com,WORK-COLLAB
  - DOMAIN-SUFFIX,miro.com,WORK-COLLAB
  - MATCH,Default

This sample is intentionally illustrative rather than a complete vendor rule set. Service domains change, and a copied list can become stale or overly broad. Treat every domain as a hypothesis that must be confirmed through the client’s connection log, browser developer tools, or the product’s documented network requirements. Avoid adding an entire cloud provider suffix simply because one image request used it; that can route unrelated services and substantially enlarge the blast radius of a typo.

Route Notion for Editing, Sync, and Attachments

Notion users usually notice routing problems through delayed saves rather than a completely blank page. A cached workspace can remain visible while new edits wait for synchronization. Pages with uploaded images or files expose a second path: the document itself may be reachable, but attachments can come from a storage or CDN destination that follows a different rule. A reliable Notion setup should therefore test three actions separately: sign in and open a workspace, create and reload a small edit, and upload or download a representative attachment.

Begin with a dedicated policy for the Notion domains confirmed in your logs and official documentation. Do not assume every hostname containing “notion” belongs to the same function, and do not place generic domains such as an entire shared CDN behind the work group without checking their purpose. If a page opens but edits remain pending, inspect whether the browser maintains a WebSocket or another long-lived connection. A proxy group that performs well for short HTTPS requests may still be a poor choice for persistent sessions because of unstable latency or aggressive connection reaping.

Browser cache can hide a routing change. Test in a private window only as a diagnostic, not as a permanent workflow, because private mode changes cookies, extensions, and service-worker behavior. After selecting another Clash group, reload the workspace, make a clearly identifiable test edit, wait for the save indicator, and reopen the page. If the result differs between a normal window and a private window, investigate cached application state before rewriting all rules.

Attachments deserve their own test. Upload a small image, open it in a new tab, download it again, and watch the Clash connection log during each action. If the page works but the attachment fails, the missing destination is probably not the main application hostname. Add the narrowest confirmed rule, then repeat the test from a second device. This method is safer than routing every storage or CDN hostname through the same group, especially on a shared home-office connection where other users depend on direct local services.

Keep Figma and Miro Real-Time Sessions Stable

Figma and Miro are especially sensitive to session consistency because collaboration is not limited to downloading HTML and JavaScript. A board or design file may maintain a long-lived connection for presence, comments, cursors, object changes, and conflict resolution. If the initial page uses one exit and the real-time channel later uses another, the service may reconnect repeatedly or degrade to a less capable mode. The user sees disappearing cursors, stale comments, duplicated updates, or a warning that the connection is unstable.

Test Figma with a small two-person scenario. Open the same file on two authorized accounts or two browser profiles, move an object, add a comment, and observe whether the change appears promptly on the other session. Then open the file’s assets, inspect a prototype link, and export a small frame. Each action exercises a different combination of application requests and media delivery. If editing works but export is slow, the issue may be an asset or export endpoint rather than the document channel itself.

For Miro, test board loading, object movement, comments, invitation links, and export separately. A board with many images and embedded videos can be much heavier than a blank board, so use a realistic workspace rather than a minimal demo. If the board opens but new objects appear only after a manual refresh, focus on the persistent connection and proxy timeout behavior. Check whether the selected node changes during the session, whether Clash logs repeated reconnects, and whether another VPN or browser extension is intercepting the same traffic.

Avoid switching a selector group while actively editing a shared file. A deliberate change is reasonable during diagnosis, but changing exits mid-session can produce confusing authentication prompts and duplicated reconnects. Save work, close the document, switch the group, clear only the relevant connection state if necessary, and reopen. Stability is usually more valuable than chasing a one-time latency score, particularly for workshops where several participants depend on the same board.

Operational rule: Measure the complete collaboration action, not only the first page-load time. A node that wins a 204 probe but drops long-lived sockets is not the fastest node for Figma or Miro.

Maintain the Same Workflow Across Desktop and Home Office Devices

A shared setup should distinguish between configuration portability and device-specific permissions. The YAML can share policy names, rule ordering, DNS intent, and proxy-group structure, but TUN mode, system proxy integration, firewall permissions, and network extensions vary by operating system. Clash Verge Rev on Windows may require administrator approval for a service or virtual adapter, while a macOS client may request Network Extension permission. Copying a profile does not copy those operating-system approvals.

Use one canonical profile source and give it a clear revision label. Keep provider URLs private, avoid placing subscription tokens in screenshots, and document which rules are local overrides. If family members or colleagues use the same home-office device, avoid changing a global selector without telling them; a choice that improves a Figma workshop can make an unrelated local service feel broken. Prefer a predictable default group and record temporary experiments in a change note.

Compare devices with a small verification matrix:

• Confirm the active profile and Clash core version.
• Check that WORK-COLLAB points to the intended group on each device.
• Verify local printers, NAS shares, and private sites remain direct.
• Open a Notion page and confirm a test edit synchronizes.
• Open a Figma file and verify a second session receives a change.
• Open a Miro board and test comments or live object movement.
• Review logs for unexpected fallback, DNS errors, or repeated reconnects.

When one device fails, compare observable facts before editing the shared profile. Check whether the failure follows the account, network, device, browser profile, or selected node. Windows system proxy settings may affect some applications but not others; macOS network extensions may capture traffic that a browser-only proxy does not; mobile or tablet clients may expose only a subset of the same controls. A clean comparison prevents a local permission problem from becoming a risky global rule change.

Keep rule providers and remote configuration sources under review. A provider update can reorder a domain rule or introduce a broad match that changes the result without any local edit. After a provider refresh, repeat the three-product smoke test and inspect the active rule for one Notion, one Figma, and one Miro connection. If the service is business-critical, pin a known-good revision or maintain a rollback copy according to your organization’s change-control policy.

Clash for Windows-era guides often encourage broad system-proxy toggles without explaining application boundaries, while browser extensions may cover only the tab and miss desktop helpers or TUN-captured traffic. Generic VPN clients can be simpler but usually expose less rule-level visibility for separating Notion, Figma, Miro, and local devices. By contrast, Clash V.CORE gives this workflow explicit policy groups, connection logs, local-network exclusions, and portable rule structure, making it easier to reproduce a stable collaboration setup across desktop and shared home-office devices; when you are ready to validate your profile on the right platform, download Clash V.CORE and apply the same narrow, observable routing plan.