Add lifecycle, divorce, elections, memory query, baby agents, hospital births

NN training: realistic persona data, graphs, hot-reload endpoint

Fix StockEx order endpoint: use shared database helper

Add StockEx trading integration for player agents

Add life events, goals, pregnancy system; fix profile editor bugs

Fix Gemini circuit breaker: distinguish per-minute rate limit from daily quota

Fix Groq circuit breaker: don't treat per-minute 429 as daily quota exhaustion

Fix LLM switcher: quota status detection, NN→100%, slider sync, label cleanup

Fix quota runtime estimates: use RPM as bottleneck, sync top probability slider

Fix quota endpoint: per-provider tick rates for accurate runtime estimates

Add quota-aware LLM switching: budget calculator, probability slider, Gemini/Groq support

Diversify agent activities, add local training script, scheduled Gemini cycle, architecture diagram

Add NN provider as default, remove HF inference (7B/0.5B)

Stop persisting LLM probability to DB and default endpoint to 0.10

Default LLM probability to 10% for cloud providers, ignore stale DB value

Fix Gemini conversations: bump LLM budget to 2 and remove double probability gate

Reset Gemini circuit breaker after failed startup probe

Add logging config so server output is visible in HF container logs

Fix HF provider switch: don't raise at init if HF_TOKEN missing

Show soci-agent HF models unconditionally (don't require HF_TOKEN)

Clean up LLM selector: only deployed HF models, remove Ollama soci variants

Add all soci-agent model variants to LLM selector

Set soci-agent-7b as default across all providers and UI

Set soci-agent-7b as default LLM for Ollama and HF providers

Set soci-agent-q4 as default LLM for all providers

Add Soci fine-tuned model and training data pipeline

Fix NameError: import PROVIDER_OLLAMA and PROVIDER_CLAUDE in server.py

Set default LLM call probability to 10% for all providers

Auto-fallback provider on startup if configured provider is quota-exhausted

Fix Gemini default: use gemini-2.0-flash-lite (confirmed free-tier working)

Add _last_error tracking to GeminiClient and GroqClient for diagnosis

Fix HF provider: require token, hide from menu if absent, fix stuck provider

Fix Gemini model: default to gemini-1.5-flash, robust fallback detection

Fix Gemini fallback: add 'serverless' to model-not-found keyword list

Persist LLM probability across restarts; fix Gemini model fallback

Add LLM probability slider: controls AI usage per call site

Fix Gemini circuit-breaker: expire at midnight Pacific, not fixed 8h

Fix Gemini RPM (4 not 14) and add daily quota warnings at 50/70/90/99%

Fix Gemini quota circuit-breaker: detect 'quota' in 429 body

Fix Gemini 429 log truncation: escape braces in response body

Log Gemini 429 body in complete_json too

Log Gemini 429 response body for diagnostics

Clean up token env vars; add Gemini to free provider menu

Remove paid providers from menu; circuit-break on 402 (no credits)

Fix HF LLM: use free SmolLM3 as default, strip thinking tokens

Fix HF Inference: migrate to router.huggingface.co/v1, use Qwen as default

Fix HF router URL: model ID must be in the path

Fix HF endpoint: migrate to router.huggingface.co

Silence JSON parse warnings for empty LLM responses

Fix HF serverless URL: use per-model endpoint path