Skip to main content
The Config module loads and exports all environment variables used throughout Junkie.

​
Environment Loading

Configuration is loaded from .env file using python-dotenv: 
from dotenv import load_dotenv
load_dotenv()

​
Database Configuration

​
POSTGRES_URL

​
POSTGRES_URL
str
default:""
PostgreSQL connection URL for database operations
Example: 
POSTGRES_URL=postgresql://user:pass@localhost:5432/junkie

​
Model and Provider Configuration

​
PROVIDER

​
PROVIDER
str
default:"groq"
AI model provider (“groq” or custom provider URL)

​
MODEL_NAME

​
MODEL_NAME
str
default:"openai/gpt-oss-120b"
Model identifier to use

​
CUSTOM_PROVIDER_API_KEY

​
CUSTOM_PROVIDER_API_KEY
str
default:"None"
API key for custom provider

​
GROQ_API_KEY

​
GROQ_API_KEY
str
default:""
Groq API key for Groq models

​
FIRECRAWL_API_KEY

​
FIRECRAWL_API_KEY
str
default:""
Firecrawl API key for web scraping capabilities

​
Agent Configuration

​
MODEL_TEMPERATURE

​
MODEL_TEMPERATURE
float
default:"0.3"
Model temperature for response randomness (0.0-1.0)

​
MODEL_TOP_P

​
MODEL_TOP_P
float
default:"0.9"
Top-p sampling parameter (0.0-1.0)

​
AGENT_HISTORY_RUNS

​
AGENT_HISTORY_RUNS
int
default:"1"
Number of previous conversation runs to include in context

​
AGENT_RETRIES

​
AGENT_RETRIES
int
default:"2"
Number of retry attempts for failed agent calls

​
DEBUG_MODE

​
DEBUG_MODE
bool
default:"false"
Enable debug mode for agents

​
DEBUG_LEVEL

​
DEBUG_LEVEL
int
default:"1"
Debug verbosity level (1-3)

​
MAX_AGENTS

​
MAX_AGENTS
int
default:"100"
Maximum number of agent teams to cache in memory

​
Tracing Configuration

​
TRACING_ENABLED

​
TRACING_ENABLED
bool
default:"false"
Enable Phoenix tracing for observability

​
PHOENIX_API_KEY

​
PHOENIX_API_KEY
str
default:"None"
Phoenix API key for tracing

​
PHOENIX_ENDPOINT

​
PHOENIX_ENDPOINT
str
Phoenix tracing endpoint URL

​
PHOENIX_PROJECT_NAME

​
PHOENIX_PROJECT_NAME
str
default:"junkie"
Phoenix project name for trace organization

​
MCP Configuration

​
MCP_URLS

​
MCP_URLS
str
default:""
Comma-separated list of MCP server URLs
Example: 
MCP_URLS=https://mcp1.example.com,https://mcp2.example.com

​
Chat Context Configuration

​
CONTEXT_AGENT_MODEL

​
CONTEXT_AGENT_MODEL
str
default:"gemini-2.5-flash-lite"
Model to use for context/history analysis

​
CONTEXT_AGENT_MAX_MESSAGES

​
CONTEXT_AGENT_MAX_MESSAGES
int
default:"50000"
Maximum messages to analyze for context

​
TEAM_LEADER_CONTEXT_LIMIT

​
TEAM_LEADER_CONTEXT_LIMIT
int
default:"100"
Number of recent messages to include in team leader context

​
Usage Example

from core.config import (
    PROVIDER,
    MODEL_NAME,
    GROQ_API_KEY,
    MAX_AGENTS,
    POSTGRES_URL
)

print(f"Using provider: {PROVIDER}")
print(f"Model: {MODEL_NAME}")
print(f"Max agents: {MAX_AGENTS}")

​
Environment File Template

# Database
POSTGRES_URL=postgresql://user:pass@localhost:5432/junkie

# Model Provider
CUSTOM_PROVIDER=https://api.example.com/v1
CUSTOM_PROVIDER_API_KEY=your_api_key
CUSTOM_MODEL=gpt-5
GROQ_API_KEY=your_groq_key

# Agent Settings
MODEL_TEMPERATURE=0.3
MODEL_TOP_P=0.9
AGENT_HISTORY_RUNS=1
AGENT_RETRIES=2
MAX_AGENTS=100

# Debug
DEBUG_MODE=false
DEBUG_LEVEL=1

# Tracing
TRACING=false
PHOENIX_API_KEY=your_phoenix_key
PHOENIX_PROJECT_NAME=junkie

# MCP
MCP_URLS=https://mcp.example.com

# Context
CONTEXT_AGENT_MODEL=gemini-2.5-flash-lite
CONTEXT_AGENT_MAX_MESSAGES=50000
TEAM_LEADER_CONTEXT_LIMIT=100

# Optional
FIRECRAWL_API_KEY=your_firecrawl_key

Build docs developers (and LLMs) love

Get started for freeTalk to us