Harden Plato voice prompt output

Add voice Langfuse diagnostics

Add ElevenLabs custom LLM endpoint

Polish voice QA entry screen

Add ElevenLabs voice QA widget

Use first user message as trace input

Increase chat output token limits

Fix Langfuse trace previews

Restore Langfuse trace inputs

Stabilize Langfuse turn traces

Tune Plato prompts per provider

Sync prompts with Langfuse

Update provider model defaults

Restrict HF demo to Plato model set

Fix HF chat session isolation and model fallback

Stabilize HF Space startup for Claude deployment

Fix Claude Haiku defaults and HF resilience

Restrict Space to Plato Claude Haiku and isolate Langfuse traces

Fix Langfuse usage tracking and Plato prompt

update: claude models

update: claude api key variable&huggingface pw

Update prompts/configs

HF Space clean code-only snapshot