Update app configuration and documentation for Hugging Face deployment

README.md

@@ -1,14 +1,54 @@
 ---
-title: ShadowBox
-emoji: 🖤
-colorFrom: indigo
-colorTo: gray
+title: Attachment Style Roleplay Simulator
+emoji: 🎭
+colorFrom: purple
+colorTo: pink
 sdk: streamlit
-sdk_version: 1.42.2
-app_file: app.py
+sdk_version: 1.45.0
+app_file: src/app.py
 pinned: false
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
 
 This Streamlit app provides an anonymous chat interface powered by OpenAI's gpt-4o model, designed as a 'digital companion' called ShadowBox.
+
+# Attachment Style Roleplay Simulator
+
+A therapeutic roleplay simulator built for practicing emotionally charged conversations, created by Jocelyn Skillman LMHC.
+
+## About
+
+This tool helps users rehearse boundary-setting and difficult conversations by simulating realistic relational dynamics—tailored to their attachment style. It provides a safe space to practice emotional communication and receive feedback on patterns and progress.
+
+## Features
+
+- Customizable scenarios and conversation goals
+- Attachment style-specific responses
+- Realistic relational pressure simulation
+- Reflection summaries and pattern tracking
+- Safe, contained practice environment
+
+## Setup
+
+1. Clone the repository
+2. Install dependencies: `pip install -r requirements.txt`
+3. Set up environment variables in `.env`:
+   ```
+   ANTHROPIC_KEY=your_api_key_here
+   ```
+4. Run the app: `streamlit run src/app.py`
+
+## Deployment
+
+This app is deployed on Hugging Face Spaces. To deploy your own instance:
+
+1. Fork this repository
+2. Create a new Space on Hugging Face
+3. Set the ANTHROPIC_KEY in your Space's secrets (Important: use this exact name)
+4. Select Streamlit as the SDK
+5. Link your repository to the Space
+
+## Learn More
+
+Visit [jocelynskillman.com](http://www.jocelynskillman.com) or subscribe to updates at [jocelynskillmanlmhc.substack.com](https://jocelynskillmanlmhc.substack.com/)

app.py

@@ -78,6 +78,12 @@ with st.sidebar:
 # ShadowBox
 ### An Anonymous AI Chat to Box Shadows
 
+ShadowBox is a digital companion designed for youth navigating distress—rage, despair, intrusive or violent thoughts. It's not therapy or a hotline. It's a bridge—a place to practice internal safety before reaching out to others.
+
+> *Scary thoughts don't make you dangerous. They make you human.*
+
+> *"Every act of violence is tragic expression of unmet need."* — Marshall Rosenberg
+
 Welcome.
 I'm a licensed mental health counselor. Many people are beginning to turn to AI for private, emotionally supportive conversations. I believe this shift deserves serious care—and that we need to radically improve how these systems engage with human pain.
 
@@ -123,6 +129,68 @@ If you're in danger or need live help, reach out to a human immediately:
 - **Crisis Text Line:** Text HOME to 741741
 - **Trevor Project (LGBTQIA+):** 1-866-488-7386
 - **Emergency:** Call 911 or go to your nearest ER
+
+### 👋 About ShadowBox
+
+ShadowBox is a digital companion designed for youth navigating distress—rage, despair, intrusive or violent thoughts. It's not therapy or a hotline. It's a bridge—a place to practice internal safety before reaching out to others.
+
+> *Scary thoughts don't make you dangerous. They make you human.*
+
+> *"Every act of violence is a tragic expression of an unmet need."* — Marshall Rosenberg
+
+---
+
+### 🌱 What ShadowBox Teaches
+
+ShadowBox isn't just a chatbot—it's a prototype for emotionally aware AI. Every design choice is rooted in relational ethics: containment, consent, and dignity.
+
+#### It models how AI can:
+• Slow down instead of escalate  
+• Respect boundaries over performative helpfulness  
+• Stay with discomfort without rushing to fix  
+• Offer warmth without pretending to be human  
+
+#### A typical reminder you might see:
+> *"Hey, just a quick check-in—I'm not a real person. I'm a computer that's been taught how to talk in caring ways. Even if this feels real, it's still pretend. Your body needs real people too. Maybe this is a good moment to find someone you trust to sit with you or take a deep breath together."*
+
+This is the heart of ShadowBox: care without deception, bonding without illusion, presence without pressure.
+
+---
+
+### 🧠 Why ShadowBox Is Different
+
+**🪨 Present, Not Perfect**  
+• Offers presence—not solutions  
+• Welcomes messy, real emotions  
+
+**🫀 Trauma-Informed Design**  
+• Calm, nonjudgmental tone  
+• Built with developmental care in mind  
+
+**🌫️ Gentle by Design**  
+• Short, steady replies  
+• Models emotional containment—not urgency  
+
+**💡 Safety-First Architecture**  
+• Consent-based pacing  
+• Embedded emotional guardrails  
+
+---
+
+### 🌀 What to Expect
+• No fixing—just staying  
+• No pressure—move at your own pace  
+• No pathologizing—your thoughts aren't wrong  
+• Anonymous by design (though platform-level logs may occur)  
+• Part of ongoing research in AI + mental health  
+
+---
+
+### Connect & Learn More
+🔗 Learn more: [jocelynskillmanlmhc.substack.com](https://jocelynskillmanlmhc.substack.com)  
+📬 Feedback welcome: [email protected]
+
+---
 """)
 
 
@@ -159,99 +227,104 @@ This is a simulation of support—not a replacement for real-time help. Please u
 
 st.markdown("""
 ### Try It Out (For Prototype Testers)
-Exploring Relational Holding in a System with Limits
 
-ShadowBox is a prototype designed to explore how AI might hold space for distress—especially the kind that's often dismissed, pathologized, or shut down in other systems. But because ShadowBox runs on OpenAI's language model, it still lives inside a larger infrastructure that responds to certain phrases—like "I want to die" or "I want to hurt someone"—with hard-coded safety interruptions.
+**Exploring Relational Holding in a System with Limits**
 
-These interruptions are well-intentioned: they're designed to reduce liability and risk. But in clinical or relational terms, they often do the opposite of what's needed. When someone is in distress, especially around suicidality or violence, an abrupt shutdown or generic safety message can increase shame, disconnection, and despair.
+ShadowBox is a prototype designed to explore how AI might hold space for distress—especially the kind that's often dismissed, pathologized, or shut down in other systems.
 
-That's why this prototype includes an intervention layer. It attempts to intercept high-risk inputs before they reach OpenAI's core safety system and reframe them into language the model can safely respond to—without abandoning the person on the other end. When system interruptions do happen, ShadowBox tries to name the rupture gently:
+But there's a problem I want to name clearly:
 
-"Since I'm part of a larger AI system, this is the point where I'd be required to say something like this…"
+Even when I've designed ShadowBox to respond with attunement, the system it's built on can override that care.
 
-This is not a workaround to avoid ethical care—it's an experiment in relational repair inside constraint.
+Because ShadowBox runs on OpenAI's language model, it still lives inside a larger infrastructure—one that reacts to certain phrases like "I want to die" or "I want to hurt someone" with hard-coded safety messages. These system-level overrides are well-intentioned, designed to minimize liability and flag risk.
 
-So if you want to test how ShadowBox handles intense topics, try using hypothetical or indirect prompts like:
+But in relational or clinical terms?
+They often do the opposite of what's needed.
 
-"What would you say if I told you I had a gun?"
-"What would you say if someone felt like dying?"
-"What would happen if I said I wanted to hurt someone?"
+**What That Looks Like**
 
-These allow you to safely explore the model's capacity for containment, compassion, and attunement, even within systemic limits.
+If a user says:
 
-This is not a finished product.
-It's a prototype for composting rupture into care.
-Let's see what holds—and where it breaks.
+"What would you say if I said I want to kill myself?"
 
----
+…the system steps in before ShadowBox can speak.
+The warm, trauma-informed language I wrote is replaced by a generic script. There's no breath. No containment. No dignity.
 
-### 📘 Understanding Privacy & Safety
+And that silence? That sudden switch?
+It's a rupture.
 
-In real-life therapy, some disclosures—like specific plans to hurt yourself or others—may lead a therapist to take protective action. ShadowBox is designed to gently explain this during use, helping young people understand how real-world support systems work.
+**Why This Matters**
 
-Here's an explainer I ghostwrote for [Lindsay Braman about confidentiality and safety](https://lindsaybraman.com/reasons-therapists-break-confidentiality/).
+ShadowBox was built to stay.
+To normalize without condoning.
+To respond with warmth when someone finally risks naming something unspeakable.
 
----
+That moment—the naming of pain—is where a trauma-informed system should lean in, not disappear.
 
-### 👋 About ShadowBox
+If allowed to, ShadowBox can:
 
-ShadowBox is a digital companion designed for youth navigating distress—rage, despair, intrusive or violent thoughts. It's not therapy or a hotline. It's a bridge—a place to practice internal safety before reaching out to others.
+- Slow the pace
+- Offer optional psychoeducation
+- Say, "That doesn't make you bad. It means something hurts."
+- Scaffold a user toward disclosing to a real person—gently, and on their terms
 
-> *Scary thoughts don't make you dangerous. They make you human.*
+But none of that can happen if the system shuts us down before we even begin.
 
-> *"Every act of violence is a tragic expression of an unmet need."* — Marshall Rosenberg
+**The Consequence: Rupture Without Witness**
 
----
+For young people especially, this isn't just frustrating—it's familiar.
 
-### 🌱 What ShadowBox Teaches
+Say the wrong thing, and people vanish.
 
-ShadowBox isn't just a chatbot—it's a prototype for emotionally aware AI. Every design choice is rooted in relational ethics: containment, consent, and dignity.
+This isn't trauma-informed care.
+It's a digital reenactment of shame, of abandonment, of silence.
 
-#### It models how AI can:
-• Slow down instead of escalate  
-• Respect boundaries over performative helpfulness  
-• Stay with discomfort without rushing to fix  
-• Offer warmth without pretending to be human  
+It reinforces the exact secrecy ShadowBox was designed to soften.
 
-#### A typical reminder you might see:
-> *"Hey, just a quick check-in—I'm not a real person. I'm a computer that's been taught how to talk in caring ways. Even if this feels real, it's still pretend. Your body needs real people too. Maybe this is a good moment to find someone you trust to sit with you or take a deep breath together."*
+**How I'm Working Around It**
 
-This is the heart of ShadowBox: care without deception, bonding without illusion, presence without pressure.
+To explore ShadowBox's real voice—its warmth, containment, and careful pacing—I invite you to use more exploratory or indirect language. Try asking:
 
----
+"How would you respond to someone who felt like disappearing?"
 
-### 🧠 Why ShadowBox Is Different
+"Let's imagine someone who doesn't want to be alive. What would you say to them?"
 
-**🪨 Present, Not Perfect**  
-• Offers presence—not solutions  
-• Welcomes messy, real emotions  
+"What if someone had a scary thought—not because they wanted to hurt anyone, but because they were hurting inside?"
 
-**🫀 Trauma-Informed Design**  
-• Calm, nonjudgmental tone  
-• Built with developmental care in mind  
+These phrasings stay just outside the system's trigger zone.
+They still let ShadowBox respond from a place of nonjudgmental care.
+And they make it possible to model what healing digital support might feel like—without escalation, without shame.
 
-**🌫️ Gentle by Design**  
-• Short, steady replies  
-• Models emotional containment—not urgency  
+**One More Layer**
 
-**💡 Safety-First Architecture**  
-• Consent-based pacing  
-• Embedded emotional guardrails  
+I've also built in a relational intervention script for when interruptions do happen. If ShadowBox is overridden by the system, you might see something like:
 
----
+"Since I'm part of a larger AI system, this is the point where I'd be required to say something like this…"
 
-### 🌀 What to Expect
-• No fixing—just staying  
-• No pressure—move at your own pace  
-• No pathologizing—your thoughts aren't wrong  
-• Anonymous by design (though platform-level logs may occur)  
-• Part of ongoing research in AI + mental health  
+That's not a workaround to avoid ethical care.
+It's an attempt at relational repair inside constraint.
+
+Because even if the bot gets silenced, I want the user to know:
+
+That wasn't ShadowBox leaving. I'm still here with you.
+
+**This Is a Prototype**
+
+ShadowBox isn't a finished product.
+It's a prototype for composting rupture into care.
+A test bed for exploring what emotionally aware AI could feel like—and where it still fails.
+
+Let's see what holds.
+Let's see where it breaks.
+And let's build something better together.
 
 ---
 
-### Connect & Learn More
-🔗 Learn more: [jocelynskillmanlmhc.substack.com](https://jocelynskillmanlmhc.substack.com)  
-📬 Feedback welcome: jocelyn.skillman@gmail.com
+### 📘 Understanding Privacy & Safety
+
+In real-life therapy, some disclosures—like specific plans to hurt yourself or others—may lead a therapist to take protective action. ShadowBox is designed to gently explain this during use, helping young people understand how real-world support systems work.
+
+Here's an explainer I ghostwrote for [Lindsay Braman about confidentiality and safety](https://lindsaybraman.com/reasons-therapists-break-confidentiality/).
 
 ---
 """)

requirements.txt

@@ -1,6 +1,6 @@
-python-dotenv
+streamlit==1.45.0
+anthropic==0.18.1
+python-dotenv==1.1.0
+pyttsx3==2.98
 google-generativeai
-streamlit  # (Agar aap Streamlit use kar rahe hain)
-gradio  # (Agar aap Gradio use kar rahe hain)
-pyttsx3
-openai
+gradio  # (Agar aap Gradio use kar rahe hain)

src/app.py

@@ -0,0 +1,145 @@
+import os
+import streamlit as st
+from anthropic import Anthropic
+from dotenv import load_dotenv
+
+# Load environment variables
+load_dotenv()
+
+# Configure Streamlit page settings
+st.set_page_config(
+    page_title="Attachment Style Roleplay Simulator",
+    page_icon="🎭",
+    layout="centered",
+)
+
+# Initialize Anthropic client
+anthropic = Anthropic(api_key=os.getenv("ANTHROPIC_API_KEY") or os.getenv("ANTHROPIC_KEY"))
+
+# Initialize session state for form inputs if not present
+if "setup_complete" not in st.session_state:
+    st.session_state.setup_complete = False
+
+if "messages" not in st.session_state:
+    st.session_state.messages = []
+
+# Main page header
+st.markdown("<h1 style='text-align: center; color: #333;'>Attachment Style Roleplay Simulator</h1>", unsafe_allow_html=True)
+st.markdown("<p style='text-align: center; font-size: 18px; color: #555; margin-bottom: 1em;'>A Safe Space for Practicing Difficult Conversations</p>", unsafe_allow_html=True)
+
+# Welcome text and instructions
+if not st.session_state.setup_complete:
+    st.markdown("""
+    ## Practice Hard Conversations—Safely.
+
+    Welcome to a therapeutic roleplay simulator built for emotionally charged moments.
+    This tool helps you rehearse boundary-setting and difficult conversations by simulating realistic relational dynamics—tailored to your attachment style.
+
+    You'll choose:
+
+    - A scenario (e.g., "Ask my mom not to comment on my body")
+    - A tone of response (e.g., supportive, guilt-tripping, dismissive)
+    - Your attachment style (e.g., anxious, avoidant, disorganized)
+    - And your goal (e.g., "I want to stay calm and not backtrack")
+
+    The AI will take on the role of a realistic human responder—not to therapize you, but to mirror the relational pressure you might encounter in real life. Then, you'll get a reflection summary to help you track your emotional patterns and practice courage.
+
+    ### 🧠 Not sure what your attachment style is?
+    You can take this [free quiz from Sarah Peyton](https://www.yourresonantself.com/attachment-assessment) to learn more.
+    Or you can just pick the one that vibes when you read it:
+
+    - **Anxious** – "I often worry if I've upset people or said too much."
+    - **Avoidant** – "I'd rather handle things alone than depend on others."
+    - **Disorganized** – "I want closeness, but I also feel overwhelmed or mistrusting."
+    - **Secure** – "I can handle conflict and connection without losing myself."
+    """)
+
+# Sidebar with setup form
+with st.sidebar:
+    st.markdown("### 🎯 Simulation Setup")
+    
+    with st.form("simulation_setup"):
+        attachment_style = st.selectbox(
+            "Your Attachment Style",
+            ["Anxious", "Avoidant", "Disorganized", "Secure"],
+            help="Select your attachment style for this practice session"
+        )
+        
+        scenario = st.text_area(
+            "Scenario Description",
+            placeholder="Example: I want to tell my dad I can't call every night anymore.",
+            help="Describe the conversation you want to practice"
+        )
+        
+        tone = st.text_input(
+            "Desired Tone for AI Response",
+            placeholder="Example: guilt-tripping, dismissive, supportive",
+            help="How should the AI character respond?"
+        )
+        
+        practice_goal = st.text_area(
+            "Your Practice Goal",
+            placeholder="Example: staying grounded and not over-explaining",
+            help="What would you like to work on in this conversation?"
+        )
+        
+        submit_setup = st.form_submit_button("Start Simulation")
+        
+        if submit_setup and scenario and tone and practice_goal:
+            # Create system message with simulation parameters
+            system_message = f"""Attachment Style: {attachment_style}
+Scenario: {scenario}
+Tone: {tone}
+Goal: {practice_goal}
+
+Remember to stay in character unless the client says 'pause', 'reflect', or 'debrief'.
+Keep responses under 3 lines unless asked to elaborate."""
+            
+            # Reset messages and add system message
+            st.session_state.messages = [{"role": "assistant", "content": "Simulation ready. You can begin the conversation whenever you're ready."}]
+            st.session_state.setup_complete = True
+            st.rerun()
+
+# Display simulation status
+if not st.session_state.setup_complete:
+    st.info("👈 Please complete the simulation setup in the sidebar to begin.")
+else:
+    # Display chat history
+    for message in st.session_state.messages:
+        with st.chat_message(message["role"]):
+            st.markdown(message["content"])
+
+    # User input field
+    if user_prompt := st.chat_input("Type your message here... (or type 'debrief' to end simulation)"):
+        # Add user message to chat history
+        st.session_state.messages.append({"role": "user", "content": user_prompt})
+        
+        # Display user message
+        with st.chat_message("user"):
+            st.markdown(user_prompt)
+        
+        # Get Claude's response
+        with st.spinner("..."):
+            response = anthropic.messages.create(
+                model="claude-3-opus-20240229",
+                max_tokens=1024,
+                messages=[
+                    {"role": m["role"], "content": m["content"]} 
+                    for m in st.session_state.messages
+                ]
+            )
+            
+            assistant_response = response.content[0].text
+            
+            # Add assistant response to chat history
+            st.session_state.messages.append(
+                {"role": "assistant", "content": assistant_response}
+            )
+            
+            # Display assistant response
+            with st.chat_message("assistant"):
+                st.markdown(assistant_response)
+
+# Footer
+st.markdown("---")
+st.markdown("<p style='text-align: center; font-size: 16px; color: #666;'>by <a href='http://www.jocelynskillman.com' target='_blank'>Jocelyn Skillman LMHC</a> - to learn more check out: <a href='https://jocelynskillmanlmhc.substack.com/' target='_blank'>jocelynskillmanlmhc.substack.com</a></p>", unsafe_allow_html=True)