Uses the image provider configured by the admin. Manual upload remains available.
Assign values to stats: 15, 14, 13, 12, 10, 8
Create accounts for each player. Players only see characters and sessions they have been assigned to.
Manage multiple API providers (OpenAI, DeepSeek, Local LLMs, etc.)
Choose models independently for shared narration, player POV prose, and backend agent work. Agents handle rules resolution, state bookkeeping, summaries, suggested actions, DJ choices, and utility prompts. Leaving a role on the fallback preserves the previous single-model behavior.
Used by OpenAI-compatible models that support reasoning_effort.
DM Instructions are built-in and automatically updated with new features (XP tracking, dice rolling, etc.)
Picks a new shared soundtrack after every narrated turn. Players receive the same track in real time; each player keeps their own volume.
Players can generate character avatars and illustrate their current POV using this image provider. POV scenes appear for everyone in real time.
Keeps the latest 3 illustrated turns in each campaign and removes older or orphaned image files.
AI-powered narration for DM messages. Click the speaker icon on any DM message to hear it read aloud.
TTS requires an OpenAI API key. Set OPENAI_TTS_API_KEY in environment, or use an OpenAI configuration above.
Send hidden instructions to the AI. Players won't see these messages, but the AI will use them to guide the story.
The AI uses a summary of past events to maintain story continuity. You can view, edit, or force-regenerate this summary.
Let the AI generate a short action/response for a character and send it automatically. Useful for NPCs or absent players.