refactor: migrate weather provider to Open-Meteo, improve location extraction, and inject server time into persona context

feat: add support for immediate camera trigger and bypass LLM generation in think loop

feat: implement camera capture tool to enable vision capabilities via ffmpeg

refactor: require essential environment variables by raising RuntimeErrors instead of using default values

feat: add crypto price tracking tool and update persona architecture documentation

refactor: remove redundant LLAMA_MODEL configuration from memorize and think modules

chore: update default USER_ID from OppaAI to Guest

refactor: enable multi-user session support by mapping system prompts and histories to user IDs

refactor: scope memory management and system prompts to individual user_id sessions throughout the chat pipeline

refactor: update LLM warmup, optimize audio handling by removing conversion, and refine VRM animation API

Add user_id via login

Fix wrong llm url

Update tool call

Add tools call

Remove obsolete stuffs

Refactor build search query again

Refactor build search query

Refactor build search query

Change websearch mechanism

Fix over search

Combine web search and web fetch in one function

Fix missing except bug

Add web fetch call

Add web search system prompt

Fix web search not triggered issue

Fix the search string leak

fix: comment out orphaned Groq closing brace

Switch to Modal

Fix web search leak

Remove LLM response streaming

Fix web search feature

Fix search loop

Add websearch feature

Fix the Groq model API connection

Add Groq API endpoint

Switch back to Groq for testing first

Switch back to Modal API

Fix Gemini API format

Fix Gemini format

Switch to Google Gemini API

Switch to llama LLM format

Switch to llama3.1 model

Switch to Groq API

Update core/think.py

Update core/think.py

Switch to OpenRouter token

Update core/think.py

Update core/think.py

Add more debug print

Add debug print