When a network decides that Telegram should not load, the app rarely tells you why. The spinner keeps turning, messages sit with a clock icon, and photos never resolve. In that moment, Telegram proxy settings are the only lever the app itself gives you — a client-side configuration that reroutes your traffic through an intermediary server instead of connecting directly to Telegram's data centers. Understanding what that lever actually moves, and what it leaves untouched, is the difference between a two-minute fix and an afternoon of guessing.
This guide is structured the way the problem usually unfolds: first define what “blocked” means at the transport level, then take the shortest reachable path on each platform, then examine the exceptions and side effects that nobody mentions in a one-line tutorial, and finally verify the result and know how to roll back cleanly. Operational paths are described by menu name rather than by build number, because exact labels shift between client releases and platform skins. Where a conclusion is based on observation rather than documentation, it is marked as such and paired with a reproducible check you can run yourself — so the reasoning stays auditable even as the menus move.
What “Blocked” Actually Means for a Telegram Client
Telegram does not depend on a single hostname or a single port. The official clients maintain connections to Telegram's data centers over an encrypted protocol called MTProto, and they are built to try alternative endpoints and transports when one path fails. That resilience explains a common confusion: two people on the same Wi-Fi network can report different results, because one client may have found a working fallback path while the other is still pinned to a dead one. A block is rarely a single switch; it is usually one of three distinct failure modes, and each one responds differently to a proxy.
- Name-level filtering. DNS answers for Telegram domains are altered or dropped, so the client cannot resolve an address at all. The symptom is a fast, permanent failure rather than a slow one.
- Address-level blocking. Known IP ranges are rejected at the network edge. The client connects, then stalls, then retries — producing the familiar “Connecting…” loop.
- Protocol fingerprinting or throttling. Traffic is allowed but shaped, delayed, or reset once its pattern is recognized. This is the most frustrating case because the app sometimes half-works: text may arrive while media never loads.
Distinguishing these three matters because the symptom narrows the fix before you touch a single field. A name-level failure usually resolves the instant any working tunnel is in place, whereas a fingerprinting network can reject a perfectly healthy proxy simply because the handshake looks like what it is hunting for. Example: if the app fails within a second of opening, suspect resolution; if it hangs on “Connecting…” for thirty seconds, suspect address-level rejection; if it connects but media never loads, suspect shaping.
A proxy addresses the first two cleanly and the third only if the proxy itself disguises the traffic. That is the entire technical justification for having two proxy types in the client rather than one. Note what a proxy does not do: it does not change your account identity, it does not encrypt anything Telegram has not already encrypted, and it cannot help if the proxy endpoint you chose is itself blocked by the same network. Those three boundaries are where most failed attempts originate.
Where Proxy Settings Sit in Telegram's Feature Map
Proxy configuration is a connection-layer setting, not an account setting. It sits apart from privacy controls, two-step verification, and active session management, and it governs only how this particular installation reaches Telegram's servers. The practical consequence, based on empirical observation across clients, is that proxy entries are stored locally on each device rather than synchronized through your account: add a proxy on the desktop client and your phone will typically still be on a direct connection. You can confirm this in under a minute — add an entry on one device, open the proxy screen on another, and compare the lists.
This local-storage behavior is worth internalizing before you start, because it reframes the task. You are not configuring “your Telegram”; you are configuring “Telegram on this phone, on this network, today.” A traveler who sets up a proxy before departure and then changes SIM cards may find the entry still present but useless, since the new network may block the proxy endpoint instead of Telegram itself. Treat proxy entries as network-specific artifacts that need periodic review, not as a one-time account preference.
There is also an accounting relationship worth knowing: traffic routed through a proxy is still counted in the app's network usage statistics, and it still consumes your mobile data plan. A proxy changes the route, not the volume. If you are debugging a data-usage complaint, disabling the proxy will not reduce consumption — it will only change which hop the packets take.
MTProto Proxy vs SOCKS5: Which One Solves Which Problem
The client offers two protocol families, and choosing between them is the single most consequential decision in this whole process. An MTProto proxy is a relay purpose-built for Telegram: it speaks the same protocol the app already uses and forwards encrypted packets to Telegram's data centers without being able to read them. Because it can be configured to pad or disguise its handshake, it is the type that tends to survive protocol fingerprinting. A SOCKS5 proxy is a generic, decades-old standard for tunneling TCP connections; it knows nothing about Telegram, supports optional username and password authentication, and performs no obfuscation of its own.
| Dimension | MTProto proxy | SOCKS5 proxy |
|---|---|---|
| Designed for | Telegram traffic specifically | Any TCP application |
| Traffic disguise | Configurable padding or TLS-like handshake | None by default |
| Authentication | A shared secret embedded in the link | Optional username and password |
| Share format | In-app link carrying server, port, secret | In-app link carrying server, port, credentials |
| Typical use | Networks that fingerprint Telegram's protocol | You already control a tunnel endpoint |
| Main weakness | Depends on volunteers running relays | Easier to detect and block by pattern |
The decision rule that follows is simple. If the network merely refuses Telegram's addresses but does not inspect traffic, either type works and SOCKS5 is often the lower-latency choice because the path is more direct. If the network resets connections once it recognizes the protocol, an MTProto proxy with disguise enabled is the only realistic option. A useful small scenario: a five-person research group shares one SOCKS5 endpoint provided by their institution's IT department for outbound access; it works fine on a network that blocks by IP range, but the same endpoint fails on a domestic connection that fingerprints traffic, and the group has to fall back to an MTProto relay shared by a colleague.
Advanced users should also note that the two types are not mutually exclusive. The client lets you keep several entries saved and switch the active one, which is exactly the right pattern for someone who moves between home, office, and mobile data. Save one MTProto relay and one SOCKS5 endpoint, order them by observed reliability, and switch rather than delete when a network changes.
Version Evolution: How Proxy Configuration Moved Through the Client
Proxy support has not always looked the way it does now. In earlier generations of the clients, the feature was a single pair of fields: one server, one port, one toggle. There was no list, no share link, and no way to keep a working configuration around while testing another. That shape made sense when proxies were an edge case for a small number of users, and it became untenable once entire networks started filtering the app. Specific build numbers are deliberately omitted here, because menu labels and placements have been reshuffled repeatedly and a version string copied from an old tutorial is more likely to mislead than to help. Check your own installed version if you need to cite one.
The evolution that matters for daily use happened along three axes, and each has a migration implication:
- From one proxy to a managed list. Clients now let you save multiple entries and choose the active one. Migration advice: if you have an old single-proxy setup that still works, add it to the list form and keep the original notes, because re-entering a long secret from memory is error-prone.
- From manual typing to link-driven setup. Tapping a shared link now pre-fills server, port, and credentials behind a confirmation sheet. Migration advice: prefer links over transcription. A single transposed character in a secret produces a proxy that “connects” and then silently fails.
- From a top-level settings row to a nested one. Placement has drifted between the privacy section and the data-and-storage section across platforms and releases. Migration advice: stop memorizing the path and start using the settings search field, which is stable even when the hierarchy is not.
Read together, those three shifts describe a feature that moved from a power-user workaround to a routine connection setting without ever being redesigned in a single step. That is why old instructions can be partially right — the concept survives even when the location does not.
A hypothetical illustration of why this matters: a user follows a three-year-old screenshot, looks for a proxy row that has since moved, concludes the feature was removed, and installs an unverified third-party client instead. The feature was never removed — only relocated. Whenever a path in this article does not match what you see, treat that as a labeling difference and look for the same concept one level up or down before assuming a functional change.
The Shortest Reachable Path on Android
On Android, the fastest route to the configuration screen is the settings search field rather than manual navigation. Open Telegram, go to Settings, and use the search icon at the top of the settings list to type “proxy”. Depending on your build, the proxy row lives under Privacy and Security or under Data and Storage; both entries open the same Proxy Settings screen, so either will do. If you prefer to navigate manually, check Privacy and Security first, then Data and Storage.
- Open Settings → Privacy and Security → Proxy (or Data and Storage → Proxy on builds where the row sits there).
- Turn on the master switch that enables proxy usage. Nothing routes through a proxy until this is on, even if entries exist.
- Tap Add Proxy, then choose the protocol type — MTProto or SOCKS5.
- Enter the server address and port. For MTProto, paste the secret exactly as provided. For SOCKS5, add the username and password if the endpoint requires authentication.
- Save, then confirm the status line reports a successful connection rather than a spinner.
The order of those steps is not arbitrary: the master switch comes before the entry so that a saved-but-inactive configuration cannot be mistaken for a working one. If you save an entry and the app still behaves as though nothing changed, check the switch first — it is the cheapest possible explanation and, based on empirical observation, a common one.
The fallback path is usually faster than the form: if someone has shared a proxy link in a chat, tapping it opens a confirmation sheet with the server details already filled in. Accepting adds the entry and enables it. This is the path to recommend to non-technical contacts, because it removes the transcription step where most failures happen.
Tip: Keep at least two working entries saved. When one relay is rate-limited or goes offline, switching is a single tap; rebuilding a configuration from scratch on a throttled connection is not.
The Shortest Reachable Path on iOS
The iOS client exposes the same concepts with slightly different choreography. Open Telegram, tap Settings, and use the search field at the top of the settings screen to look for “proxy” — on iOS this is consistently faster than hunting through sections, because the row's parent section has changed across releases. Manual navigation typically leads through Privacy and Security or Data and Storage, mirroring Android's split.
- Open Settings and locate the proxy entry via search or the sections named above.
- Enable the proxy switch.
- Tap Add Proxy and select MTProto or SOCKS5.
- Fill in server and port; paste the secret for MTProto, or enter credentials for SOCKS5.
- Tap Done and return to the chat list to watch the connection indicator settle.
The behavioral difference that catches people out is modal versus in-place confirmation. On iOS, proxy links often arrive as an in-app sheet that must be explicitly accepted before anything is saved; dismissing the sheet by swiping leaves no trace, and the user may believe the proxy was added when it was not. If a shared link “did nothing,” re-tap it and look carefully for the confirmation control. As with Android, a second saved entry is cheap insurance.
Desktop and Web: The Path Most People Miss
On the desktop client, proxy configuration lives in a different neighborhood entirely, which is why people who successfully configured their phone often cannot find it on their laptop. The route is Settings → Advanced → Connection type. There you will find the choice between a direct connection and a custom proxy, plus an Add proxy control that opens a dialog with protocol selection and the appropriate fields — server, port, and secret for MTProto; server, port, and credentials for SOCKS5.
- Open Settings and select Advanced.
- Open Connection type.
- Select the custom proxy option, then choose Add proxy.
- Pick MTProto or SOCKS5, complete the fields, and save.
- Watch the connection status in the same panel; it reports whether the client has established a session through the proxy.
The phrase “Connection type” is the part worth remembering, because it explains the mental model: on desktop, a proxy is treated as one of several ways to connect, not as a separate feature bolted onto privacy settings. Look for the connection concept and the proxy controls follow.
Web clients are a separate story. As of this writing, the browser-based version does not expose the same in-app proxy list and instead inherits whatever proxy configuration the browser or operating system provides. If your only access is a browser, the practical sequence is to configure a system-level or browser-level proxy first, then open the web client. This is also why “the proxy works on my phone but not in the browser” is a coherent report rather than a contradiction: two different layers are being configured.
Warning: Switching the active proxy on desktop while a call is in progress will drop that call. Change connection settings before joining meetings, not during them.
Share Links vs Manual Entry: Reading a Proxy Link Correctly
Proxy links are compact, and that compactness hides the information you need to judge whether an entry is legitimate. The two formats in circulation follow a predictable pattern — one for MTProto relays, one for SOCKS5 endpoints:
t.me/proxy?server=HOST&port=PORT&secret=SECRET t.me/socks?server=HOST&port=PORT&user=USERNAME&pass=PASSWORD
Reading the query string tells you three things before you accept anything. First, the server value shows where your traffic will actually go — a domain you have never heard of is not automatically malicious, but it is a party you are now trusting with your connection metadata. Second, the port is a clue about the configuration; unusual ports are normal for relays, so do not read meaning into them. Third, the secret for MTProto is a long hexadecimal string that, in configurations documented in Telegram's own MTProxy tooling, may carry a short prefix indicating padding or a TLS-like handshake mode. You do not need to decode it, but you should never hand-edit it: a truncated or retyped secret yields a proxy that appears to connect and then fails silently.
- Do not retype secrets. Copy the full link, or use the tap-to-add flow. This single rule eliminates the majority of “the proxy doesn't work” reports.
- Check for truncation. Messaging apps sometimes wrap long links; if the sheet shows a server but no secret, the link was cut.
- Prefer links from a source you can name. “Forwarded many times” is not provenance.
Those three habits cost nothing and replace a whole category of guesswork. They also make troubleshooting honest: when a link is intact and the source is known, a failure almost always belongs to the network or the relay rather than to your typing.
A reproducible verification: after accepting a link, open the proxy list and compare the saved server and port against the original link text. If they differ, the link was mangled in transit and the entry should be deleted rather than debugged.
What Each Toggle Does (and What It Doesn't)
The proxy screen contains fewer controls than people assume, and misreading one of them wastes time. The master switch decides whether the client is allowed to use a proxy at all; with it off, saved entries are inert. The list below it selects which entry is active — only one is used at a time, and reordering changes preference rather than enabling multiples. Removing an entry is not the same as disabling the switch: deletion is permanent, while the switch is reversible in one tap.
A note on calls deserves its own paragraph, because it is the most common source of confusion. Call routing and messaging routing do not necessarily behave identically, and some builds expose additional options around how calls are handled when a proxy is active. If your client shows such an option, the conservative default is to leave call routing on the direct path unless messaging itself is impossible, since voice traffic through a distant relay tends to produce one-way audio and dropped connections. If your build shows no such option, do not go looking for one; there is nothing to configure.
Finally, the toggle does nothing to your account. It does not change your phone number, your sessions list, or your privacy settings, and turning it on will not log you out or migrate your chats. Anyone who tells you to reinstall the app before trying a proxy is skipping a step that costs nothing.
Choosing a Proxy You Can Actually Trust
Telegram does not operate a public fleet of free proxies, and no official client ships with a preloaded relay. Every entry you add was run by someone — a volunteer, an organization, a colleague, or you. That framing should govern your risk assessment. The good news is that an MTProto relay forwards packets it cannot decrypt, so the operator does not read your messages. The bad news is that the operator still sees metadata: which addresses you connect to, when, and roughly how much data moves. For most people that is an acceptable trade; for a journalist working a sensitive story, it is a decision worth making deliberately rather than by convenience.
SOCKS5 endpoints carry a different profile. They are generic tunnels with no Telegram-specific design, and if authentication is required, the credentials you enter are part of the handshake with that endpoint. That is normal for the protocol, but it means the entity running the tunnel is an infrastructure provider in a meaningful sense. The strongest option, when it is available, is a SOCKS5 endpoint you or your organization controls, because the trust question collapses into something you can already answer.
Tip: Test throughput before committing. Send yourself a large photo or a short video file through the proxy and observe whether it completes in a reasonable time. Empirical observation across public relays suggests quality varies enormously — some are fast for days and then collapse under load, which is a property of volunteer capacity, not of your device.
A practical middle path for teams: designate one person to maintain a short list of two or three verified relays, share them as links in a private group, and review the list when the network environment changes. This avoids the pattern where every team member independently adopts a different unvetted relay from a public channel.
Verifying That Traffic Is Really Going Through the Proxy
A proxy that has been added is not necessarily a proxy that is working, and the client's status indicators are subtler than a green dot. Verification should be a deliberate two-minute routine, not a glance.
- Read the status label. The proxy screen reports whether the client has established a connection through the active entry. A persistent spinner or a retry message means the handshake is failing, not that the network is slow.
- Send a message to Saved Messages. A message that moves from clock icon to a single check mark confirms an outbound round trip. A message that stays on the clock is queued, not sent.
- Load media that is not cached. Open a channel you have never visited and scroll back. Text arriving while images stall is the signature of a partially working path, often a sign of aggressive throttling rather than a dead proxy.
- Repeat on a second network. Toggle Wi-Fi off and retry on mobile data. If the proxy works on one and not the other, the problem is network-specific filtering of the proxy endpoint, not your configuration.
Together these four checks separate three different states that all look alike from the chat list: the proxy is not being used at all, the proxy is used but failing, and the proxy is working while the network still throttles media. Only the middle state calls for a configuration change.
Be careful with a false positive: a message that appears to send instantly may simply be an optimistic UI state, and cached media loads even with no connection at all. The reliable signal is a fresh, uncached round trip on a network where direct connections demonstrably fail. If you want a repeatable experiment, record the behavior with the proxy off first, then on, on the same network within a few minutes of each other — the contrast is the evidence.
Exceptions and Side Effects: When a Proxy Makes Things Worse
Every hop adds latency, and a relay run by a volunteer on a residential connection can add a great deal of it. The most common side effect is not failure but sluggishness: messages send, media eventually arrives, and the app feels wrong in a way that is hard to articulate. Battery consumption typically rises as well, because the client spends more time in reconnect loops when the relay is unstable. These are working hypotheses grounded in how the protocol behaves, and they are easy to observe qualitatively on your own device by comparing a session with the proxy on and off.
- Voice and video calls degrade first. Real-time traffic tolerates added latency worst. If calls matter more than messaging on a given day, consider switching the proxy off for the call.
- The relay itself can be blocked. Popular public relays are discovered and filtered quickly. A proxy that worked last month failing today is expected behavior, not a malfunction.
- Media downloads may time out. Large files over a slow relay often fail partway. Downloading in smaller batches, or switching to a faster entry first, is the usual workaround.
- Reconnect storms. On flaky networks the client may retry constantly, producing heat and battery drain. Disabling the proxy until you are on a stable network is a legitimate choice.
Notice that none of these require a new configuration. They are symptoms of the path you chose, and the remedy is usually a different entry or a temporarily disabled switch rather than a fresh setup.
There is a subtler exception worth flagging: a proxy can make a partially blocked network feel worse than no proxy at all. If the network throttles all encrypted traffic indiscriminately, routing through a relay adds a hop without removing the throttle, and the user concludes the proxy is broken when the network is the problem. The diagnostic is to try a completely different network — a different carrier, or a public hotspot — before discarding a configuration that may be fine.
Rollback: Disabling or Removing a Proxy Without Losing Messages
Rollback is the step people skip, and it is the one that matters when something goes wrong at an inconvenient time. The procedure is deliberately boring, which is the point.
- Open the proxy screen using the same path you used to configure it.
- Turn the master switch off. This immediately returns the client to a direct connection and preserves every saved entry for later reuse.
- If you want to remove the entry permanently, select it and delete. Do this only after you are sure you do not need the configuration.
- Watch queued messages resolve. Messages held while the proxy was failing should send once a working connection is restored; nothing is lost by toggling the switch.
Two caveats complete the picture. Deleting an entry means you will need the original link or secret again, so if the source is a chat message that may scroll away, save the link somewhere durable first. And because proxy settings are local, rolling back on one device has no effect on your other devices — check each one if you were troubleshooting across platforms.
Scenario Checklist: When Proxy Settings Are Worth It
Proxy configuration is not a universal good, and treating it as one produces unnecessary complexity. The following boundaries are the ones that come up most often in practice.
| Worth configuring | Not worth configuring |
|---|---|
| Telegram fails only on specific networks you must use | Telegram fails everywhere, including on networks known to be open |
| Text works but media consistently stalls | The account itself is restricted or limited |
| You have a named, trusted source for an endpoint | Your only option is an unvetted relay from a public list |
| You need continuity for a small team or project group | One-off access you can defer to another network |
| Local rules and your organization's policy permit it | Using it would breach a network policy or local law |
The second column deserves emphasis. A proxy fixes routing; it does not fix an account-level restriction, a device clock that is badly wrong, or corrupted local storage. If the app misbehaves identically on three different networks, the proxy is not the variable you should be changing.
Troubleshooting: Symptom, Cause, Verification, Fix
The following table is organized as a diagnostic sequence rather than a list of tips. Work from the symptom you can actually observe, confirm the cause with the verification step, and only then apply the fix.
| Symptom | Likely cause | Verification | Fix |
|---|---|---|---|
| Stuck on “Connecting…” after enabling | Secret or port mistyped; endpoint unreachable | Compare saved server and port with the original link | Delete the entry and re-add via the link |
| Works on Wi-Fi, fails on mobile data | Carrier blocks the relay endpoint | Toggle Wi-Fi off and retry the same entry | Switch to a different saved entry |
| Text sends, images stall | Throttling or very low relay throughput | Open an uncached channel and scroll back | Try another relay; retry media later |
| Calls connect but audio is one-way | Latency and NAT behavior through the relay | Disable the proxy and repeat the call | Route calls directly where the client allows it |
| Worked last week, dead now | Relay filtered or taken offline | Test a second saved entry on the same network | Replace the entry; keep two saved as routine |
| Slow, warm device, battery drains | Reconnect loop against an unstable relay | Compare behavior with the master switch off | Disable until on a stable network |
Read the table top to bottom and you will notice a pattern: the verification column almost always asks you to compare two states, with the proxy on and off, or on one network and another. That comparison is what converts a vague complaint into a cause, and it is the part most troubleshooting shortcuts leave out.
If none of these resolve the issue and the app misbehaves on every network with the proxy disabled, stop tuning connection settings and look at the account and the installation instead. That boundary is the most useful thing a troubleshooting table can tell you.
Compliance and the Boundary of What a Proxy Can Do
A proxy is a routing tool, and routing tools sit inside legal and organizational frameworks that vary by jurisdiction and by network. Some networks prohibit circumventing their filters as a matter of policy; some employers restrict external relays on corporate devices. Nothing in this article changes those constraints, and configuring a client is not a legal opinion. If you are acting on behalf of an organization, the question to answer first is whether the connection method is permitted, not whether it is technically possible.
It is also worth separating circumvention from enforcement evasion. A proxy changes how your client reaches Telegram's servers; it does not change how Telegram's anti-abuse systems evaluate your account, and it will not resolve a restriction applied to the account itself. Users occasionally expect a proxy to bypass a spam limitation, and it does not — the limitation follows the account, not the route. Similarly, a proxy does not anonymize you from Telegram, because your account is still your account regardless of the path taken to reach it.
The responsible default for teams is documentation: record which endpoint is approved, who maintains it, when it was last verified, and what to do when it fails. This converts an ad-hoc workaround into an operational procedure with an owner, which is the difference between resilience and a recurring fire drill.
Best Practices Checklist
The rules below are decision rules rather than steps. They compress everything above into something you can apply in under a minute.
- Use the settings search field instead of memorized paths; hierarchy changes, search does not.
- Add proxies by tapping links, never by retyping secrets.
- Keep two verified entries saved and switch rather than rebuild.
- Prefer MTProto relays when the network fingerprints the protocol; prefer SOCKS5 when you control the endpoint.
- Verify with an uncached round trip on a network where direct connections fail.
- Turn the master switch off — not the entry — when you need to restore direct routing quickly.
- Leave call routing on the direct path unless messaging is otherwise impossible.
- Expect public relays to disappear; treat replacement as routine maintenance.
- Confirm that the connection method is permitted on the network and device you are using.
Applied consistently, these rules reduce the whole topic to a small set of habits. The technical configuration is genuinely simple; what makes it feel complicated is the absence of a verification routine and the tendency to treat a proxy as a permanent account setting rather than a network-specific choice.
Frequently Asked Questions
Does enabling a proxy change my account, sessions, or chats?
No. A proxy only changes the route your client uses to reach Telegram's servers. Your account, active sessions, and chat history are untouched, and turning the proxy off restores the previous behavior immediately. Because proxy entries are stored locally per device rather than synced through your account, you may need to configure each device separately.
Should I choose MTProto or SOCKS5?
Choose MTProto when the network appears to recognize and reset Telegram's traffic, because it can disguise its handshake. Choose SOCKS5 when you already have a tunnel endpoint you trust and the network blocks by address rather than by protocol pattern. If you are unsure, try the MTProto entry first and compare behavior with a SOCKS5 entry on the same network.
Why does a proxy work on Wi-Fi but not on mobile data?
The two networks filter differently. A relay that is reachable from one may be blocked by the other, especially if it is a widely shared public endpoint. Verify by toggling Wi-Fi off and retrying the same entry; if it fails, switch to a second saved proxy rather than rebuilding the first.
Can the proxy operator read my messages?
An MTProto relay forwards encrypted packets it cannot decrypt, so it does not read message contents, but it can observe connection metadata such as timing and data volume. A SOCKS5 endpoint is a generic tunnel with no Telegram-specific design, so the trust question extends to whoever operates that infrastructure. When the content is sensitive, use an endpoint you or your organization controls.
What to Do Next
The core conclusion is narrow and practical: Telegram proxy settings solve routing problems, and routing problems are the only ones they solve. Start by confirming which failure mode you actually have — an unreachable address, a reset connection, or throttled traffic — because that diagnosis determines whether an MTProto relay or a SOCKS5 endpoint is the right instrument. Then take the shortest path: search for the proxy row in settings rather than memorizing a hierarchy, add the entry by tapping a link rather than retyping a secret, and switch the master toggle rather than deleting configurations you may need again.
If you want a concrete next step, do this today: save one working MTProto entry and one working SOCKS5 entry, verify both with an uncached message and a media load, and write down where the links came from. That small piece of housekeeping converts a recurring emergency into a one-tap switch — and when the network changes again, which it will, you will already be prepared.
On version expectations, hold the plan loosely at the interface level and firmly at the conceptual one. Menu labels, section parents, and dialog wording have changed repeatedly across client releases and continue to do so; the underlying model — a locally stored list of entries, a master switch, a protocol choice between MTProto and SOCKS5, and a status indicator — has been far more stable. Re-read your own build whenever a path here does not match, and treat any future relocation of the proxy row as a labeling change to be re-learned rather than a feature to be re-installed.
