Basic

A prompt-driven therapist with optional chain-of-thought reasoning, configurable for different therapeutic approaches.

Overview

Property	Value
Key	basic
Type	LLM-based
Focus	Configurable via prompt template

Key Features

• Prompt-driven: Therapeutic approach is defined entirely by the prompt template
• Chain-of-thought (CoT): Optional reasoning step before generating a response
• Configurable: Swap prompt files to target different therapeutic modalities (CBT, MI, etc.)

How It Works

1. Prompt Loading: Loads a system prompt from the configured prompt_path.
2. Session Init: Builds the conversation with the system prompt as the first message.
3. Response Generation: On each turn, appends the user message, calls the LLM, and appends the assistant reply to history.
4. CoT Mode (optional): When use_cot=True, the model returns structured reasoning alongside the response content.

Usage

CLI

uv run python -m examples.simulate therapist=basic

Python

from patienthub.therapists import get_therapist

therapist = get_therapist(agent_name="basic", lang='en')

response = therapist.generate_response("Hello, I've been feeling anxious lately.")
print(response.content)

Configuration

Option	Type	Default	Description
prompt_path	string	data/prompts/therapist/CBT.yaml	Path to prompt file
use_cot	bool	False	Enable chain-of-thought structured output
model_type	string	"OPENAI"	Model provider key
model_name	string	"gpt-4o"	The LLM model to use
temperature	float	0.7	Controls response randomness
max_tokens	int	8192	Max response tokens
max_retries	int	3	API retry attempts

Response Format

Without CoT, returns a plain string response. With use_cot=True, returns a structured object:

class ResponseWithCOT(BaseModel):
    reasoning: str  # Internal chain-of-thought (not shown to client)
    content: str    # Therapist's actual response

Use Cases

• General-purpose therapist baseline
• CBT training simulations (default prompt)
• Research on different therapeutic prompt designs
• Educational demonstrations