Fix context + add psychoeducation + hide text during selection

Mobile-friendly: single column, ultra-concise

Streamlined UX: sidebar I-statement builder, 5 stages

Add feelings selector UI at Stage 2

Remove label from input box - cleaner interface

Hide needs selector until Stage 5 - clean opening page

Strengthen feelings and needs prompts - more interactive and connected

Remove buggy Chatbot component - use Markdown display instead

Fix Chatbot: use type='messages' format to avoid gradio_client schema bug

Improve API error handling and add setup instructions

Try Gradio 5.9.1 with default Chatbot type

Switch to Gradio 5.6.0 with messages format

Downgrade to Gradio 4.44.0 to fix compatibility bug

Fix Gradio 5.x compatibility: use messages format for Chatbot

Add startup logging for API key debugging

Show debug info in chat when API key not found

Fix Gradio compatibility - pin to stable 4.44.0, add chatbot type

Add debug info for API key troubleshooting

Try multiple API key secret names for compatibility

Fix API key name back to anthropic_key

Fix API key secret name to match user's setting

Update UI with custom color palette (warm gold, sage green, deep forest)

Simplify header text, link to CNVC

Remove resource links from sidebar

Rebuild as NVC Deep Practice learning space

Remove 'Safely' from main heading

Add expandable practice goal examples

Fix broken Sarah Peyton attachment quiz link

Move simulation setup from sidebar to main page

Add prominent mobile menu instruction at top of page

Fix |aI| pronoun: only for AI self-reference, not user scripts

Add |aI| pronoun requirement to debrief prompt

Fix debrief bug: capture conversation before clearing state

Update sidebar welcome text with ARI framing

Fix 404 error: Update model from deprecated claude-3-opus to claude-sonnet-4

Update setup instructions to be clear for both mobile and desktop users

Update welcome message to be more mobile-friendly

Change arrow direction back to pointing left

Change arrow direction to point to sidebar

Add scroll instruction to welcome text

Clear conversation state when entering debrief mode

Fix debrief message format and remove duplicate bubbles

Fix debrief initialization order to prevent NameError

Fix duplicate chat bubble in debrief initialization

Update debrief functionality with new therapeutic reflection prompt

Remove debug messages from API initialization

fix: Simplify Anthropic client initialization and downgrade package version

fix: Update Anthropic client initialization to fix proxies parameter error

Update to use Anthropic API instead of OpenAI

ported over new wip app. note - its using gpt4o not claude