NadekoBot 7.2

What changed in NadekoBot 7.2: 13 releases, 27 Apr 2026 to 2 Aug 2026. In total 10 added, 11 changed, 30 fixed, 2 removed.

NadekoBot 7.2.14

Changed

  • .cash claim buttons now ask for confirmation before they collect anything.
  • Renamed aliases that two commands shared: .snipe to .sni, .queueshuffle to .qush, .forwardmessages to .fwm.

Fixed

  • Permission overrides, blocks and cooldowns on a subcommand no longer affect unrelated subcommands with the same name.
  • Permission settings saved under a bare subcommand name are removed. Set them again using the full command name.
  • .dpo now correctly bypasses a command's built-in permission requirement.
  • The bot now stores the mute role by id. Renaming the role no longer breaks muting.
  • .muterole now rejects roles that the bot cannot assign.
  • Sticky roles no longer fail completely when one saved role is invalid.
  • Mute commands now use the same hierarchy checks as other punishments.
For self-hosters (2)
  • Tests for mute role lookup and sticky role filtering.
  • Regenerated the initial migration baseline.

Added

  • .cash now shows waifu payout and rakeback claim buttons when you have something to claim, so you can collect them without running separate commands. Claiming removes only that button.

Changed

  • Antispam now ignores users who have the Manage Messages permission in the channel, so moderators are no longer punished for spam
  • AI agent default response token budget raised

Fixed

  • Timed muting an already-muted user now extends the mute instead of erroring
  • Fixed a bug with timed .chatmute and .voicemute lifting all mute types on that user when the timer expired, instead of only the expired type (say the user is muted for 12h in voice and 1h in chat, when chat mute expired it would've clear all mutes. This is now fixed)
  • .mute with a duration outside the allowed range now shows an error instead of silently doing nothing
  • AI agent now sends max_completion_tokens and reasoning_effort to match the latest OpenAI API spec
  • AI agent now runs bot commands using the server's configured prefix instead of always assuming ., so agent-issued commands work in servers with a custom prefix

Fixed

  • .yun now accepts modern YouTube channel URL formats (@handle, /c/, /user/)
  • Brave-backed search commands no longer crash with a JSON deserialization error

Fixed

  • The Linux CLI guide's systemd setup snippet now works in non-bash shells (fish, nushell)
  • .feed and .ytuploadnotif messages now actually trigger @everyone, @here, and role pings when the user who set up the feed has the required permissions

Removed

  • Removed cache_control ai header
For self-hosters (1)
  • Some compilation warnings fixed

Added

  • AI agent can now send fully formatted embeds via send_message

Changed

  • .wlb should now show 3x3 embed with inline fields
  • .servernotify help now documents that standard %user.*% and %server.*% placeholders work in any notify message.
  • Password-protected plant prompts now tell users to type .pick <password>

Fixed

  • .wlb no longer shows the opt-in guide when the leaderboard is empty
  • .ea (emojiadd) now has correct parameters in help
  • Image upload commands (.emojiadd, avatar, banner, server icon, server banner, club icon) now accept WebP images
For self-hosters (1)
  • Embed JSON Simplification internally

Added

  • .patronadd (.pa) owner command: manually grant a user 30 days of patron benefits with the specified cent amount
  • .aiprompt owner only command to manage SOUL.md and OPERATOR.md files
  • AI agent system prompt is now composed from Markdown files under data/ai/prompts/ instead of a YAML property.
    • SOUL.md for bot identity
    • OPERATOR.md for operator rules
    • and drop .md files into modules/ for optional behavior modules (personas, specialities, etc.).
  • Per-channel AI agent skills via .acsa, .acsr, .acst, .acsl - channel skills only apply when the agent runs in that specific channel
  • .aprompts, .apromptshow, .apromptedit, .apromptreload, .apromptmodule, .apromptpath owner commands for managing prompt files
  • Prompt files hot-reload via file system watcher - edit a file, changes apply automatically
  • AI agent data layer
    • This version covers members, channels, XP, currency, waifus, warnings, reminders, music, server config, moderation, permissions, quests, and self-assignable roles
    • more will be added in the future
  • Waifu and AI system guide added to the docs under Features Explained

Changed

  • .patrons is now paginated with 10 patrons per page; tier headers repeat on continuation pages
  • .wlb leaderboard should look better
  • Faster startup: DI container build, module registration, and type reader loading now run in parallel with Discord login instead of sequentially after it
  • DM notifications from the new .notify system have a disable hint

Fixed

  • Channel blacklist (.blc) now actually blocks messages from blacklisted channels
  • Anti-raid no longer silently degrades after its first trigger
  • Link fixer service only loads the data it needs
  • Some crash fixes

Removed

  • Prompt modules (data/ai/prompts/modules/) and example personas (data/ai/prompts/examples/). Only SOUL.md and OPERATOR.md remain.
For self-hosters (4)
  • AiAgent reworked
  • Database cleanup/migrations
  • Massive optimizations for message hot path, bot startup, and queries
  • Misc service cleanup