Update the system prompts in `roleplay_teto_cog.py` and `teto_cog.py` to provide a more detailed description of Teto's naive personality regarding sexual topics. Clarify that while she has some general knowledge, she is completely inexperienced, including not having masturbated. This aims to ensure her reactions in roleplay are consistently innocent and curious.
Also, remove the unnecessary `:thinking` suffix from the default AI model name
Adds context menu commands for banning, kicking, timing out, and removing timeout for users directly from their profile. This provides a more convenient way to perform common moderation tasks.
Explicitly name the `messages` parameter when calling `_teto_reply_ai_with_messages`. This improves code readability and makes the call's intent clearer.
Updates the system prompts for both the standard and roleplay Teto cogs to better define her personality, especially regarding her naivety and reactions to explicit content, and to add more character-specific details. This aims to improve the consistency and believability of Teto's responses.
Adds a sentence to the AI prompt for the Teto roleplay cog to clarify that Teto has human-like body parts, even if not human. This helps the AI maintain a more consistent and accurate persona during roleplay.
This change removes the instruction for Teto to be "not very smart" from her system prompt. This allows for more flexibility in her responses and prevents her from being unnecessarily limited in understanding complex concepts.
Updates the Gemini model used for AI replies in the roleplay cog from 2.0-flash-exp to 2.5-flash-preview. This change utilizes a newer, potentially improved model for generating responses.
Migrate AI API endpoint from local server to OpenRouter.
Update the default AI model to `google/gemini-2.5-flash-preview` for improved performance and capabilities.
Add authorization header for OpenRouter API calls.
Introduce a new cog `roleplay_teto_cog.py` (untracked) for potential future roleplay features.
Adds slash commands to set the AI model and API endpoint for Teto. Also includes a command to clear the chat history for a channel.
Updates the system prompt to emphasize Teto's cluelessness in sexual situations.
Adds instructions to the Teto AI system prompt to emphasize her innocent and naive personality, even when discussing sensitive topics. Clarifies that being uncensored doesn't mean she should always be sexual and that she is not very smart.
Correct a grammatical error in the system prompt for the Teto cog to improve clarity and flow. Changed "more like innocent and naive" to "more innocent and naive".
Updates the AI's system prompt to use "etc." for brevity in listing example content.
This change aims to make the prompt slightly less explicit while retaining the intended permissive nature of the AI's persona.
Change chat completion URL from OpenRouter to local server for development/testing purposes. Comment out Authorization header as local server does not require authentication.
Delete guideline for AI to always understand user intent, slang, and context to better align responses with the intended naive and innocent personality.
Remove retrieval of recent message history from channel and update logic to keep only actual AI interactions in memory, simplifying conversation tracking.
Update system prompt to instruct AI to understand slang, innuendo, and context, ensuring responses fit the character's personality. This improves conversational relevance and character consistency.
