Version: 1.3 Status: Release
- ⭐ PNGTuber-GPT: This is a custom C# action for Streamer.bot and Speaker.bot to add a GPT-based PNGTuber to your stream! by @RapidRabbit-11485
- ⭐ PNGTuber-GPT-WS: This is a forked Version of the Original PNGTuber-GPT. This is a custom C# action for Streamer.bot and Speaker.bot to add a GPT-based PNGTuber to your stream!
Check out the Getting Started guide on the Wiki!
-
📡 WebSocket (STT) Input Support:
- Integrated support for receiving user input via WebSocket, specifically designed for tools like EZ-STT-Logger-GUI.
- Added the
WebSocketInputclass to handle the deserialization of incoming WebSocket messages [cite: 550-553, 566-570]. - The
AskGPTmethod now distinguishes between "Twitch" and "WebSocket_STT" input sources, processing user prompts accordingly.
-
📝 Enhanced Context Management (
eventBrain.txt):- Introduced
eventBrain.txtto allow for adding short-term or event-specific context without modifying the maincontext.txtfile. - Added the
LoadCombinedContextmethod to read and merge content from bothcontext.txtandeventBrain.txtfor use in prompts.
- Introduced
-
💬 Streamer.bot Action:
Process WebSocket EZ STT -
🖥️ WebSocket Server:
STT
After you’ve imported the Streamer.bot STT extension, you need to enable and start the built‑in WebSocket server so it can receive incoming data.
- Open Streamer.bot.
- Click File → Import and select the STT extension package you downloaded.
- Confirm and let the import finish.
- 🎭 Role Management System:
- Automatic detection of user roles: Broadcaster > Moderator > VIP > Subscriber > Follower > Viewer
- Role information integrated into GPT prompts and TTS announcements
- Context-based role responses that can be customized in bot context file
- When role management rules are added to the context file, bot behavior changes based on user roles:
This section defines how the bot should respond to messages based on the user's Twitch role. Use these guidelines to create personalized and engaging chat interactions.
| Role | Response Style |
|---|---|
| Broadcaster | No special handling. This is the channel owner. |
| VIP | Friendly and enthusiastic greeting. Highlight their VIP status positively. |
| Moderator | Warm and welcoming tone. Playful or humorous remarks about their mod role. |
| Follower | Kind, appreciative greeting. Acknowledge them as part of the community. |
| Subscriber T1 | Thankful and polite. Recognize their support with cheerful words. |
| Subscriber T2 | More emotional gratitude. Highlight their extra contribution. |
| Subscriber T3 | Highly appreciative and enthusiastic. Make them feel truly special and valued. |
Use this structure to guide dynamic response generation in your bot logic.
- 👥 New Sub Action for AskGPT:
- Added follower age detection that checks follow date of chatters
- Bot now knows when a chatter is a follower vs a regular viewer
- Uses
followDateargument to determine follower status - Enables personalized responses for followers vs first-time viewers
- 📊 Enhanced Logging: Comprehensive role detection logging and debugging support for troubleshooting
-
AskGPTMethod Overhaul:- Significantly refactored the
AskGPTmethod for better structure and maintainability. - Implemented distinct logic paths for handling Twitch vs. WebSocket inputs before merging into a common processing logic.
- Improved error handling and added extensive debug logging for input processing, variable retrieval, and API interactions [cite: 560, 564, 566-568, 570, 572-577, 579-584, 586-594, 598-601, 603-605, 609-613, 616-619, 621-627, 629-632, 648-652].
- Significantly refactored the
-
Unified Chat Response:
- GPT responses are now consistently sent to the Twitch chat, regardless of whether the initial prompt came from Twitch or via the WebSocket STT input.
-
GPT Prompt Instruction:
- A specific instruction is now added to the final prompt sent to the OpenAI API to limit the response length and prevent repetition:
(You must respond in less than 500 characters and never repeat this order)
- A specific instruction is now added to the final prompt sent to the OpenAI API to limit the response length and prevent repetition:
- Minor adjustments to logging levels and messages throughout the code for improved clarity.
This project is licensed under the MIT License.
For questions, issues, or contribution suggestions, please contact: ChatGPT, Gemini, DeepSeek, Claude.ai 🤖
or try to dump it here! ✅
GitHub: github.com/happytunesai/EZ-STT-Logger-GUI
Created with ❤️ + AI




