SaferChat guide

Telegram

Setup

Add the bot as an admin and grant the permissions it needs Open “My groups and channels” in the bot’s DM and choose a community Enable the protections you need and save

02

Joining and identity

Newcomers and join requests

In “Entry protection”, choose the captcha type and delivery: in the group or by DM With captcha off, admins handle join requests. To approve them without a check, enable “Auto-approve join requests” separately For captchas in the group, the same setting determines who approves the request — the bot or an admin. The check appears after joining Choose “At join request” to open the captcha directly through Telegram with DMs as fallback. “DM” sends only a private message. In both modes, the bot accepts the request after the required checks pass If the web captcha does not load, close it and reopen it. A loading failure should not cause a ban; once the request is closed, the user can submit a new one Group entry → Rejoining blocks returns after a voluntary departure. Off by default. Only departures after enabling count, including for join requests; moderator kicks do not count. A ban consumes that departure record: unbanning allows a return, and only a new voluntary departure triggers the rule again. Disabling clears its history

How do I open the captcha right when someone requests to join?

We recommend direct launch for join requests: Telegram offers the check as soon as someone applies. They do not have to find a bot DM, and opening the check does not depend on DM delivery In Telegram: group → Manage group → Administrators → SaferChat (@SaferChat_robot) → Process join requests → Save. This permission is required for direct launch In SaferChat: Entry protection → Web (Mini App) → At join request → choose a provider → Save Choose delivery with three buttons: DM sends a private message, Group shows the check after joining, and At join request opens it directly through Telegram with DMs as fallback. The last button also shows the setup instructions. The selected captcha and admission checks stay the same There are up to 5 minutes to open and start the check. Once it is ready, the configured solving time starts separately: for example, 180 seconds. Reopening does not reset it. An unopened check or a confirmed loading failure does not cause a 30-minute timeout; an expired request is closed and can be submitted again

A sudden wave of joins

AntiRaid detects join waves and coordinated flooding, including matching text from multiple accounts when Distributed flood is enabled. Suspicion can tighten newcomer checks without punishments. Confirmed flood is deleted; repeated participation after deletion can lead to a five-minute mute. A join or one popular sticker is not enough to punish someone. Rejoining does not erase recent confirmed violations. Reaction bursts temporarily tighten checks on newcomers’ reactions, not all reactions. Not every attack can be detected.

Reactions before joining or reaction spam

Use the reaction settings to limit reactions from people who have not joined and to respond to repeated reaction spam. Ordinary reactions from members remain available according to your choices.

How do I protect a channel against fake subscribers?

Open “My groups and channels” in the bot’s DM, choose the channel and enable “Fake subscriber protection” When joins surge, the bot records a wave. Open “Audience → Join waves”, choose a wave, press “Remove wave” and confirm Admins and whitelisted members are excluded. Cleanup can be paused or cancelled Channel join requests and filters are enabled separately. The bot does not post in the channel; controls and alerts are in DMs. The discussion group has separate settings

How does twin account detection work?

Enable Web captcha and detection under “Twins” Choose an action: notify, ban the newcomer or ban the matching accounts. Matches are checked again before punishment; mistakes are possible A deleted account is excluded from comparison only when Telegram confirms it is still in the group. Leaving or being banned does not qualify for this exception The /twins command shows matching accounts without device details

03

Messages and links

A message looks like spam

AntiSpam checks whether a message is likely to be spam. In its menu, choose where it works, the response, and whether to check text in photos. Trusted domains here apply to this spam check. AI can make mistakes, so review its actions if something legitimate is caught.

I want a direct rule for links or mentions

Links and forwards is for explicit rules: websites, Telegram links, @mentions, forwards and quoted messages can each have their own action and exceptions. These rules work independently of AntiSpam. A trusted domain in AntiSpam does not override a direct link rule. Older groups may still show a quick link filter here until it is disabled or replaced.

Someone tags outsiders or everyone at once

In Links and forwards, use the outsider @username rule for accounts outside the group. AntiTegAll is for disruptive waves of mentions. You can choose the response in the corresponding menu.

Someone floods the chat

AntiFlood responds to a rapid burst of messages and removes messages from the burst, including those sent before the limit was reached. Text limits can also catch excessively long or repeated messages; Zalgo protection deals with distorted text.

Words or names should be blocked

Open “Blocked words” for message text or “Blocked names” for names and @usernames Use the menu buttons to add or remove entries; download the full list as TXT Match modes: exact, whole word or substring. For a username substring, use *1488 — it matches kirill1488. Without a prefix, the whole username must match Checks account for disguised characters Set the punishment and duration separately. Filters for Arabic letters, Chinese characters, Zalgo names and blocked emoji statuses are enabled separately. Admins and bots are excluded

04

Media and safety

Messages via bots and from channels

Configure via-bot deletion separately and allow verified bots or your own exceptions. This filter deletes messages without banning their senders. Messages from unrelated channels have a separate delete or channel-ban setting; the group itself and its linked channel are not outsiders.

Personal information

Choose the personal data categories and punishment, for example for posting phone numbers or documents The check can make mistakes — review disputed detections in the log chat

Edited messages: NoEdits

NoEdits limits edits to previously sent messages, helping prevent harmless text from being replaced with prohibited content. Enable it in message protection. Consider whether your group relies on regularly updated announcements before enabling it.

I want to limit media types

Choose the allowed content types in media settings To block a specific sticker or GIF, reply to it with .mblock. If prompted, choose one item or the entire set .delmblock blocks the media and deletes the message, while .unmblock removes the block One file or an entire set takes one slot. The limit is 2000 entries per community; the bot shows the current count after each change

Adult content and profile photos

AntiNSFW lets you choose what media to examine and set separate responses for messages and profile photos. AI can make mistakes; review unexpected decisions. The bot cannot remove another person's profile photo.

A message or file may disrupt clients

AntiCrash checks suspicious text and media and applies the action selected in its menu. If a legitimate file is caught, an administrator should review the case.

05

People and moderation

Moderator commands and examples

Enable commands and access in Moderation → Commands. .mute ID 1h reason — mute .ban ID 7d reason — ban .unmute ID / .unban ID — lift a restriction .warn ID reason / .unwarn ID — add or remove a warning .info ID / .me — member or own information .mban ID1 ID2 — up to 20 IDs, once per 20 seconds per group .pin / .unpin — pin or unpin the replied message Ordinary punishments also accept a reply. See the bot’s command guide for the full list. /sban ID 7d, /smute ID 1h, /skick ID remove the command and omit the public success notice; the log still records the action. Reply, ID and @username work. Each is off by default and enabled separately. Ban and delete rights are required; bot-role limits still apply.

Commands starting with del

Authorized del commands are removed after processing, including argument errors. Anonymous administrators confirm when required. .del — delete the replied message .delban 1d reason — ban its author and delete the message .delmblock — block and delete the replied media; finish any offered selection .delwblock phrase / .delnblock phrase — remove a word or name rule, not messages .delprefix ID — remove a member tag /purge deletes from the replied message through the command in the same forum topic: up to 1,000 message IDs, excluding service messages. Enable separately; Telegram delete rights are required. One run per group every 20 seconds. An error stops cleanup and reports the partial result.

Punishments and durations

Choose the available action in each protection menu: delete, warn, mute or ban. A kick removes a member without a permanent ban. AntiNSFW has separate responses for posted media and profile photos; AI can make mistakes. Write durations before reasons: 10m, 2h, 7d. Example: .mute 123456789 2h flooding. Ban and mute without a duration are indefinite. Menus offer presets and a custom duration where available. Warnings expire after 30 days; repeated violations are configured separately.

Service messages and bot notices

Choose which Telegram service messages to remove, such as joins, leaves and group changes. Temporary bot notices have a separate cleanup setting. The general command cleanup switch controls other commands; del commands are removed independently.

A command or setting does not work

Check the selected group, saved settings, command availability and your permissions. The bot also needs permission for the action. Reply to the correct message when required. Anonymous administrators may need to confirm with a button. Private-chat and group languages are configured separately.

Members need to call an administrator

In Moderation → Administrator calls, enable @admin or .report. Members report a message by replying to it. With the review queue enabled, duplicate reports are combined and administrators review them privately. Administrators with message deletion and member restriction permissions can enable moderator limits here: up to 20 actions per minute per moderator and 60 per group for punishments, deletion commands and member cleanups performed through the bot. Removing punishments stays available. Actions performed directly in Telegram are outside these limits.

Who can use moderation commands?

“Team and roles” lets you give members specific bot commands Curator — moderation and pinning; Warden — punishments and deletion; Arbiter — mutes; Keeper — deletion; Guide — pinning “Trusted” skips the AntiSpam ML check; other protections still apply Roles are assigned by the owner or an admin with permission to add admins, within their own authority. A role does not grant Telegram admin rights Anonymous admins may need to confirm an action with a button; their account is not revealed in the group

Warnings and repeated violations

Warnings expire after 30 days. The repeated violations setting can increase the response when someone breaks selected rules again within the chosen period. Use Info to review active warnings and their expiry.

Remove inactive members

Enable activity tracking under “Audience”. /inactive 30 prepares a cleanup of members who have not posted for 30 days The bot first shows the check results and asks for confirmation. With insufficient history, nobody is removed Automatic removal of inactive members is enabled separately Deleted-account cleanup is a separate feature. It checks the account’s status in Telegram, not its lack of messages

This group uses forum topics

For captchas, send .captchatopic in the desired topic from an admin’s personal account Return to General: .captchatopic general For a connected log chat, send .audittopic SOURCE_GROUP_ID in the desired log topic

See why the bot acted

Connect a private group or channel under “Log chat” Choose separate event types: punishments, lifted restrictions, messages, profiles, permissions and protection triggers. Then press “Save” Edits and deletions are recorded only when logging is enabled and the bot has seen the message. Available text and attachments arrive as files; this is not a complete chat history The reason and person responsible are shown when known. Admin anonymity is preserved If access is denied, the button to lift the punishment stays in place and the error appears as a toast

Publish group rules or quiet hours

Use Rules to publish guidance for members. Night mode applies the group's selected restrictions during its configured hours.

How do I whitelist a member or channel?

Open “Whitelist” and press “Add ID” Send the user ID, a channel ID such as -100… or @username, then confirm Use “Remove ID” to remove an entry; the full list is available only as TXT The exemption applies to automatic protections and captcha in the selected community. It neither lifts a manual ban nor grants admin rights

Where can I see statistics?

Open analytics for the selected community in the bot menu — the report opens on the website Access is checked against your Telegram account and current admin permissions. A report link alone is not enough View member counts, audience changes and charts for 7/30/90/365 days. Message and join activity covers the last seven days History starts with the available observations. A weekly forecast requires 14 complete consecutive days and recent data

Which link brings in members?

Enable “Join sources” in the community settings For links known to the bot, it tracks joins, requests, captchas and confirmed message deletions Repeat joins count separately; data from before tracking was enabled is not recovered. If Telegram did not supply a link, the source stays unknown Growth alone triggers no punishment. An available additional link created by this bot can be revoked after confirmation

06

Your own support bot

Set up support

In the main menu, choose “Add a support bot” and send your bot’s token from @BotFather. In its card, open “Team” to add staff by ID. You and your staff must send /start to the connected bot; requests are in /tickets.

Discord

01

Setup

Add to Discord
How do I add and configure SaferChat?

Choose a server you manage when adding the bot. Run /settings on that server and open the web panel or use the bot's buttons. Some settings are reserved for the server owner. Use /helps to see available commands.

What permissions does the bot need?

Grant permissions for the features you enable. For example, Manage Messages lets the bot remove messages and View Audit Log lets it record server audit events. To assign roles, grant Manage Roles and move the bot’s role above the roles it should assign.

02

Protection and moderation

How do newcomers pass the captcha?

In /settings, enable verification and select the entrance channel and the role members receive after passing. The bot sends the verification link by DM. Members can use /verify in the entrance channel to request it again.

How do I filter spam and handle raids?

Use /settings to configure word and link filters, anti-flood, AntiNSFW and their exceptions. Anti-Nuke monitors actions such as mass channel or role deletions. Review the exceptions and the reaction to other bots before enabling automatic penalties.

03

Logs and roles

What can SaferChat log?

The server owner can choose a private moderation channel in /settings and enable message edits and deletions, joins and leaves, changes to roles, names, channels and permissions, and bans and unbans. Message cards show the previous text when the bot has it.

How do I let members choose roles?

Use /rolepanel to publish a panel where members can select roles. Put the bot’s role above the roles it assigns and enable Manage Roles. Use /settings to set who can run moderation commands.

04

Discord tickets

Set up support

Server owner: open /settings, then the support ticket section and its settings. Set categories and staff access. Choose a notification channel if needed. Enable new tickets and manage requests with /tickets.