fix: switch to gemini-2.5 stable & improve audio generation

fix: take request timeout from config

fix: await coroutine

chore: move llm init to generate prompt step

fix: add retry/trycatch logic

feat: add util for LLM retries

feat: add timeout with retries for the llm requests

feat: implement Google API key management and refactor client usage in audio and image generation

fix: parallel game generation

fix: make get_user_state calls asynchronous in image_agent and music_agent

fix: update get_user_state and set_user_state calls to be asynchronous

fix: update import statements to use redis_state for user state management

feat: implement Redis-backed user state management and update app structure

fix: imports for music prompt

fix: runtime error with user choice

fix: runtime error for image and music generation

feat: use more precise prompts for music and image generation

feat: adjust scene plot

feat: adjust the language usage prompt and the length of scenes descriptions

feat: enhance game interaction with custom choice input and update scene handling

feat: update image handling in node_player_step for improved scene transitions

Merge pull request #3 from DeltaZN/llm_agent_integration

fix: correct typo in STORY_FRAME_PROMPT and enhance SCENE_PROMPT clarity

Merge branch 'main' of https://github.com/DeltaZN/gradio-mcp-hackaton into feat/improve-image-generation

feat: improve image generation

feat: implement LLM game graph and user state management

fix: music generation hangs the server

A new start