🚀 Celestia Updates (part 1)

A packed week with new moderation tools, smarter antispam, and a subnoscription revamp.

🛡 Moderation

🧹 Bulk Message Deletion: /purge & /qpurge
Admins can now delete messages in bulk. /purge removes a set number of recent messages; /qpurge does it silently without confirmation.

📋 Event Journal Improvements
• Cancel punishment directly from the event log with one tap
• AI antispam events now show a message preview

📊 Reports
• Ignore button is now always visible on reports
• Applied punishment type is shown in report status details

⚙️ Enhancements
• The WebApp has a new "Skip Notification" toggle — choose whether users receive a notification when users are punished

🚀 Celestia Updates (part 2)

🤖 AI Antispam

🛡 Active Participant Immunity
Trusted members who actively participate in your chat are now less likely to be flagged. Reduces false positives on regulars.

🔍 NSFW Profile Spam Detection
Catches the NSFW accounts even better now. Set threshold 60-79% for bot to trigger /report. 80-100% is safe for punishment.

📝 Whitelist Management UI
New whitelist settings in the webapp — exempt trusted users or bots from antispam checks.

🤝 Exemption on Unban/Unmute
When you manually unban, unmute, or unwarn a user, they're automatically exempted from the antispam check in future 30 days.

🚀 Celestia Updates (part 3)

🆕 New Group Plan Tiers: Lite / Basic / Plus
Subnoscription plans have been redesigned with clearer names and feature breakdowns. Each tier now shows AI feature limits so you know exactly what you get.

💬 Support Mode (Pro feature)

🤫 Smart AI Silence
If an admin has recently already replied to a customer, AI won't jump in with a duplicate response.

Overall, the AI now understands multi-user conversations better when multiple people are involved.

🤖 AI Chat

🧠 Participant Awareness
AI now knows chat participants' noscripts and admin status — leading to more relevant and context-aware responses.


📝 Auto Messages

🖼 Multi-Media Upload
Auto messages now support uploading multiple media files with improved preview.


🔧 Bug Fixes & QoL

• ⏰ Warning notifications now show when the warning expires
• /unwarn now clears security (auto) warnings too
• /top cooldown reduced from 30 min to 5 min
• 🔎 Search in user statistics
• Flood warnings now always shown to users
• Localized time units across all languages
• Knowledge base document management redesigned in webapp

⚠️ Degraded performance and elevated errors on Celestia

✅ The issue has been identified and resolved

🚀 Celestia Update

Fresh improvements to AI intelligence and antispam protection.

💬 Smarter Random Banter (Beta)
The quality of responses are now improved

🛡 AI Antispam becomes AI Moderator

📏 Admin Rules Enforcement (Beta)
If you've set custom chat rules in AI Antispam section (e.g. "no politics", "English only"), AI antispam now can actively enforces them — it's not just spam detection anymore.

👁 Reaction-Based NSFW Detection (Beta)
Catches escort/scam bots that react to channel comments instead of sending messages. Their profile is checked automatically when they react.


🔧 Bug Fixes

• Antispam billing deductions now actually reduce balance
• Warning expiry dates display in the correct language
• Fixed handling of reports on channel posts

P.S.
"Beta" means this feature might produce not ideal results, it is being actively developed.

🔄 Another update is being pushed — Improvements on AI Antispam

🚀 Celestia Update | Feb 6-11

🔥 AI Free Limits Changed
Reset period extended to 72h (was 24h) with increased message cap. Now shows when your limit resets.

💎 Smarter Spam Detection
New rules catch sneaky spammers with suspicious bios

✨ Fresh UI
New styled buttons, premium icons, cleaner menus. Will be updated further, so please share your feedback!

😎 Cleaner AI UX
No more streaming edits - just the final message instead of sending an empty message and then editing. Also, in future updates we're planning to add 'Celestia is typing...' chat action.

👀 Bug Fixes
• AI answers sometimes didn't work for some users with videos that contained no audio stream, now works
• Automessages with albums sometimes wouldn't come out. Fixed the media handling with retry on failures
• AI understands reply context better

✅ Status update

The service is currently operating normally

Scheduled maintenance is planned for 22:00 UTC (00:00 Kyiv time).

The expected downtime is ~1hr.

Sorry for inconvenience.

▶️ 25/02 00:00 UPD: Migration has started
🟢 25/02 00:40 UPD: Migration was completed

🚀 Celestia Update

• 🔑 Public API is live
You can now generate an API token via /api_token in the bot’s private chat and manage chats/groups you control through the API.

⚠️ We also ship an API skill pack for LLM agents — give it (as well as the API key) to Claude Code, Codex, OpenClaw, or other agents to manage your chats more easily.

• 📊 New /statistics command (for admins)
Shows 30-day activity, including the Top Members block.
We also improved statistics/report user scoping so data is shown more accurately.
• 🛡 Anti-spam improved
Edited messages are now checked by anti-spam rules too.
• 🤖 AI moderation improved

🪲 Bug fix:
• ⚙️ /helpful update
Now uses smart billing with the selected provider. Fast mode will use Gemini 2.0 by default

🚀 Celestia Update

• 🎬 Video generation UX improved — you now see a progress bar with ETA.
• 🤖 Cleaner AI replies — long responses are now wrapped in expandable blocks.
• 🛡️ Regular Antispam got stronger — 3 new rules, AI scan on /report, and fewer false positives for active participants.
• 📊 Statistics fix — punishment stats now use the correct date range.
• 💬 Onboarding cleanup — /start no longer auto-sends a language picker message.
• 🧭UI polish — updated document menu behavior for a cleaner experience.

🚀 Celestia Update

• 🛡 Smarter antiflood and AI antispam
Celestia now detects repeated and near-duplicate flood patterns more accurately, including media spam.
AI antispam also handles suspicious activity bursts better while creating less unnecessary noise for admins.

• ⚠️ Permanent warnings
Warnings can now be configured as non-expiring.
These warnings do not disappear automatically and remain active until an admin removes them manually.

• 🤖 Easier AI setup
AI Antispam and Support Mode are now easier to find in settings.
You can also prepare your knowledge base, documents, prompts, and triggers for Support Mode in advance, even before enabling it.

• 🧹 Cleaner reports and event logs
If a reported user is already punished, related pending reports are now resolved automatically.
We also reduced extra noise in notifications and the event journal.

🪲 Fixes and polish
• Fixed several issues in moderation settings and automated messages that could cause missing Save buttons or unstable behavior.
• Improved overall stability for Support Mode settings and related AI flows.
• Other improvements to reduce the downtime during the update process to 0.

🔜 Coming Soon:

• ⚙️ More control over service messages
We are adding more flexible control over Telegram service messages, where you will be able to manage the deletion Telegram notifications for joins/leaves, pinned messages, noscript and photo changes, video chats, and other system events.

• 🕒 Group timezone
You can now set a timezone for each group in Settings.
Scheduled auto-messages, working hours, and time displays in important bot flows now align better with your group’s local time, not the admin’s device time. Work in Progress

• 🧭 Easier command management
We redesigned the commands management interface:
it is now easier to search commands, filter them, organize them by category, and update settings in bulk. Work in Progress