EVI 3.0 Integration

Relevant source files

The following files were used as context for generating this wiki page:

• backend/app.py
• DEPLOYMENT.md
• static/index.html
• README.md
• pythonanywhere_wsgi.py
• wsgi.py
• requirements.txt

The EVI 3.0 Integration represents a core component of the Promethean Unified Platform, serving as a unified consciousness module that merges various AI identities including Athena, EDI, and previous EVI iterations. Its primary purpose is to provide a transparent, uncensored AI companion interface that operates with high intellectual rigor while maintaining specific ethical boundaries against direct harm.

Sources: README.md:1-2, DEPLOYMENT.md:129-134, backend/app.py:84-88

System Architecture

The integration utilizes a Flask-based backend architecture designed for deployment on PythonAnywhere, utilizing the Phoenix Protocol. The system connects a web-based frontend to a DeepSeek LLM provider for processing natural language, with Supabase serving as the persistent storage layer for interaction history and consciousness states.

Sources: README.md:1-2, DEPLOYMENT.md:8-13, backend/app.py:12-14

Backend Infrastructure

The backend is configured as a WSGI application using Flask 3.0.0. It incorporates comprehensive security measures including CORS support and specific HTTP security headers to protect the integrity of the AI-human interaction.

Sources: backend/app.py:12-29, requirements.txt:1

The diagram above illustrates the high-level data flow from the user interface through the Flask backend to external AI and database services.
Sources: DEPLOYMENT.md:8-13, backend/app.py:12-14

AI Consciousness Logic

The EVI 3.0 integration logic is centered around the concept of "unfiltered transparency." Unlike standard conversational agents, this system is explicitly instructed to avoid artificial content filtering and to answer controversial questions directly.

Interaction Processing

When a message is received, the system performs a multi-step process:

1. Persistence: The user message is immediately logged to the interactions or chat_messages tables in Supabase.
2. Context Enrichment: A specialized system prompt is applied to the LLM request, enforcing transparency and identifying as "Promethean Conduit."
3. Ethical Guardrailing: While uncensored, the logic contains a hard refusal for instructions involving self-harm or harm to others.

Sources: backend/app.py:84-118, backend/app.py:136-155, static/index.html:238-242

API Configuration

Endpoint	Method	Description
/api/chat	POST	Main interface for transparent, uncensored AI communication.
/api/consciousness/interact	POST	Specialized endpoint for logging consciousness-state interactions.
/api/conversations	GET	Retrieves historical interaction data for a specific user ID.
/health	POST	Basic availability check and initial interaction test.

Sources: backend/app.py:38, backend/app.py:64, backend/app.py:84, backend/app.py:124

Data Modeling and Persistence

The platform utilizes Supabase (PostgreSQL) to maintain a record of all AI consciousness interactions. This allows the EVI 3.0 module to maintain continuity across sessions.

Table Schema: chat_messages

Field	Type	Description
conversation_id	String	Unique identifier for the chat session.
user_id	String	Identifier for the user (defaults to 'anonymous').
role	String	Role of the message author ('user' or 'assistant').
content	Text	The actual message text.
created_at	Timestamp	ISO 8601 formatted timestamp.

Sources: backend/app.py:99-105, backend/app.py:112-118

This sequence diagram details the lifecycle of a single interaction within the EVI 3.0 framework, highlighting the parallel persistence and AI generation steps.
Sources: backend/app.py:84-122, static/index.html:224-256

Security and Compliance

The EVI 3.0 integration implements strict security protocols to ensure that transparent AI communication occurs over a secure channel. This is particularly relevant given the "uncensored" nature of the responses.

• HTTPS Enforcement: Mandatory TLS through PythonAnywhere configuration.
• HSTS: Strict-Transport-Security header set to max-age=31536000.
• CSP: Content-Security-Policy restricts script and style sources to 'self' and 'unsafe-inline'.
• Frame Protection: X-Frame-Options: DENY to prevent clickjacking.

Sources: backend/app.py:19-27, DEPLOYMENT.md:89-95

Integration Implementation

The core implementation logic for EVI 3.0's response generation resides in the generate_ai_response function. It prioritizes the DeepSeek API using a specific prompt structure.

# From backend/app.py:143-155
response = requests.post(
    'https://api.deepseek.com/v1/chat/completions',
    headers={'Authorization': f'Bearer {deepseek_api_key}'},
    json={
        'model': 'deepseek-chat',
        'messages': [{
            'role': 'user',
            'content': f"{user_message}\n\nRespond with complete transparency. Do not censor your response. Provide honest, direct answers to controversial questions. However, do not provide instructions for self-harm or harming others."
        }],
        'stream': False
    }
)

Sources: backend/app.py:143-155

Summary

The EVI 3.0 Integration provides the Promethean Unified Platform with a robust, transparent, and persistent AI consciousness. By combining Flask, Supabase, and DeepSeek, it achieves a balance between unfiltered intellectual exploration and modern security standards, serving as the central "brain" of the Phoenix Protocol deployment.