-
OAuth-based Twitch login flow with auto-closing browser tab
-
Channel selector dropdown for quick channel switching
-
Manual chat input field to send messages directly from the app
-
Auto-saves all settings to
Documents\MiniBot\settings.json -
Enable or disable each script individually using toggles (works even when offline)
-
Each script has its own ⚙️ settings popup for configuration:
- AskAI: Server address, port, model dropdown (auto-fetched), max tokens (1–255), optional system message, configurable command trigger
- Weather: Format preset dropdown with 4 options plus custom format support with help reference
- Translate: Choose target language from dropdown
- ButtsBot: Adjustable reply frequency (syllable replacements are fixed at 5%)
- ClapThatBot: Adjustable reply frequency
- MarkovChain: Brain file manager, banned words list with case-insensitive filtering, clean database function
- Sound Alerts: Add trigger-to-sound mappings, adjust volume, and stop/clear playback queue
- WalkOnScript: Assign personalized walk-on sounds per user, plays once per stream unless reset by reconnect or stream restart
-
Ignore messages from specific users using the Ignore List popup
-
Stylish dark mode UI with centralized theming, rounded corners and tooltips
-
Most scripts operate offline. Weather and Translate use public web APIs and require an internet connection, but no user setup is needed.
- Windows 10 or 11
- .NET 8 Runtime (unless using self-contained build)
- Twitch account
Optional:
- Local GPT4All server with compatible models such as
llama3-8b-instructorphi-3-mini-instruct
ButtsBot and ClapThatBot use embedded offline models, no external services required.
- Download the latest release from GitHub
- Extract the contents
- Run
MiniBotLauncher.exe
Settings are stored at:
C:\Users\<YourName>\Documents\MiniBot\settings.json
Markov brain data is stored at:
C:\Users\<YourName>\Documents\MiniBot\markov_brain_<channel>.json
-
Enter your Twitch Bot Username
-
Enter your Client ID from https://dev.twitch.tv/console/apps
- Set redirect URL:
http://localhost:8750/callback/
- Set redirect URL:
-
Select or enter a Channel Name from the dropdown (previously joined channels are remembered)
-
Click Login and authorize access in the browser (tab closes automatically after success)
-
Click Connect
-
Toggle the scripts you want active (toggles work even when not connected)
-
Click each script's ⚙️ to configure its behavior
-
Use the 📄 button in the top-right to open the Ignore List
-
Use the chat input field at the bottom to send messages directly from the app
AskAI requires GPT4All running locally. Now includes server/port configuration, automatic model listing, and configurable command trigger (default:
!askai). Responses with trailing tags liketutor:orresponse:are automatically cleaned unless clearly part of the content. ButtsBot replaces syllables at random, one syllable is always replaced if none were hit by chance. MarkovChain will respond every 35 messages. Features banned words filtering (input and output), brain file manager, and database cleaning. Sound Alerts maps chat commands to sound files and queues them for playback with a short delay. It supports.mp3and.wav, adjustable volume, and a button to stop and clear the current queue. WalkOnScript plays a specific sound the first time a user speaks during a stream. Triggers reset on reconnect or when stream start time changes.
| Script | Description |
|---|---|
| AskAI | Answers natural language queries using a local AI model (configurable command trigger) |
| Weather | Provides current weather using wttr.in with 4 preset formats or custom % notation |
| Translate | Detects non-English and translates into your chosen target language |
| ButtsBot | Replaces ∼5% of syllables in messages with "butt" — reply rate adjustable in settings |
| ClapThatBot | Detects adjective+noun phrases and responds with "I'd clap that" — reply rate adjustable |
| MarkovChain | Learns from chat to generate phrases — includes banned words, brain manager, and DB cleaning |
| Sound Alerts | Plays sound files on command triggers with volume and queue control |
| WalkOnScript | Plays a unique walk-on sound per user the first time they speak after the stream starts |
- Open
MiniBotLauncher.slnin Visual Studio 2022+ - Build in
Releasemode - Right-click project → Publish
- Use self-contained, single-file publish option for portability
MIT License — free for personal or commercial use
Third-party libraries:
- OpenNLP — POS tagging (Apache 2.0)
- CMUdict — Syllable parsing (BSD-style license)
- TwitchLib — Twitch connection
- GPT4All — Local AI inference engine
- Newtonsoft.Json — JSON parsing for settings and Markov brain
- NAudio — Audio playback for .wav and .mp3 support
Thanks to:
- TwitchLib
- GPT4All
- OpenNLP
- CMUdict
- Newtonsoft.Json
- NAudio
Built with ❤️ by Ixitxachitl
