Skip to content

SRC_README.md

Latest commit

 

History

History
92 lines (67 loc) · 2.68 KB

SRC_README.md

File metadata and controls

92 lines (67 loc) · 2.68 KB
Raw

Ted Backend

A playful, loyal AI companion that leverages long-term memory for personalized friendship and support.

Architecture Overview

The Ted backend is built around a clean, modular architecture with clear separation of concerns:

src/
├── agent.py        # CLI interface entry point
├── boot.py         # Application bootstrapping and dependency wiring
├── config.py       # Centralized configuration parameters
├── export_schema.py # Memory export schemas
├── llm.py          # LLM client for chat completions
├── memory.py       # Memory management and retrieval
├── ted.py          # Core Ted agent implementation
└── utils.py        # Utility functions and helpers

Core Components

1. Ted Agent (ted.py)

The central component that orchestrates the interaction between memory retrieval and LLM generation. Ted:

  • Remembers what matters to you and brings up relevant memories
  • Formats conversation context
  • Passes everything to the LLM for response generation
  • Stores conversations in memory

2. Memory Management (memory.py)

Handles all aspects of memory:

  • Retrieval of semantically relevant memories
  • Storage of conversations
  • Thread history management
  • Recent conversation formatting

3. LLM Client (llm.py)

Responsible for:

  • Formatting the system prompt with the appropriate context
  • Generating responses via the OpenAI API
  • Handling streaming responses

4. Configuration (config.py)

Centralizes all configuration parameters:

  • LLM settings
  • Memory parameters
  • Agent defaults
  • File paths

Usage

The Ted system can be accessed through:

from src.ted import Ted
from src.boot import memory_manager, llm_client

# Initialize the Ted agent
ted = Ted(memory_manager, llm_client, user_id="your_user_id")

# Get a response
response = ted("What should I focus on this week?")

# Or stream a response
for token in ted.stream_reply("What are my priorities?"):
    print(token, end="", flush=True)

Design Principles

  1. Separation of Concerns: Each module has a specific responsibility
  2. Configuration Centralization: All parameters are defined in one place
  3. Clean Interface: The Ted class provides a simple, intuitive API
  4. Robust Memory Management: Efficient retrieval and storage of memories
  5. Lightweight Dependencies: Minimal external dependencies

Development

To contribute to the Ted backend:

  1. Ensure you have the required environment variables:

    • MEM0_API_KEY
    • OPENAI_API_KEY

  2. Follow the existing architecture patterns

  3. Add appropriate logging for debugging and monitoring

  4. Maintain comprehensive docstrings

  5. Respect the existing configuration system