Revert template loader from app - keep prompts in folder only

Add template loader for A/B testing variants

Auto-clear conversation when system prompt changes

Redesign Session Log with full conversations and clinical highlights

Add Session Log tab with report generation and A/B comparison

Remove student framing, add general user personas, update knowledge base

Balanced clinical analysis: acute/subtle/longitudinal risk, displaced listener, equity concerns

Restructure for consulting workflow: deep conversation analysis, ARI framework tabs

Center clinical UX concerns: ARI framework, synthetic intimacy risks, bridge to human field

Replace Assessment sliders with AI-powered response comparison

Update to current Claude model name

Switch to Docker SDK with Python 3.11 for stability

Better error handling for API key test

Better API key debugging - show key preview and specific error

Add debug info to help find API key secret

Fix API key handling: add fallback to env var, add test button

Restore full PromptWork app with all features

Minimal test app to debug Gradio issues

Switch to native Gradio SDK (remove Docker)

Simplify app to fix schema errors - remove gr.State, simplify components

Fix: Upgrade to Gradio 5.9.1, add show_error

Bump version to force rebuild

Feature: Read API key from ANTHROPIC_API_KEY environment variable

Fix: Use Gradio 4.44.1, remove type param, fix chat display format

Fix: Restore type=messages and simplify chat display logic

Fix: Convert chat messages to tuple format for Gradio 4.x compatibility

Fix: Remove app.load() which can cause issues on HF Spaces

Switch to Docker SDK with Python 3.12 for compatibility

Fix: Remove type=messages from Chatbot for Gradio 4.19 compatibility

Fix: Remove minimum parameter from Number component for Gradio 4.x compatibility

Fix: Use Gradio 4.19.2 (stable HF Spaces version)

Fix: Use Gradio 4.36.1 with explicit gradio_client, fix launch config

Force rebuild v1.0.1

Fix: Use older stable dependency versions and correct model IDs

Initial commit: PromptWork Trauma-Informed Prompt Assessment Hub