Environment Variables Required for LLM Provider Configuration in hiring-agent

The interviewstreet/hiring-agent repository requires three environment variables—DEFAULT_MODEL, LLM_PROVIDER, and GEMINI_API_KEY—to control which Large Language Model provider serves requests and authenticate provider-specific API calls.

Setting up the correct environment variables ensures the hiring-agent connects to your preferred LLM backend without runtime errors. All configuration logic is centralized in prompt.py, where the application reads these variables using os.getenv and validates them against the ModelProvider enum defined in models.py.

Required Environment Variables for LLM Configuration

The repository recognizes three specific environment variables that govern LLM behavior:

Variable	Purpose	Default Value	Required When
DEFAULT_MODEL	Specifies the model name to invoke (e.g., gemma3:4b, gemini-2.5-pro)	"gemma3:4b"	Always used; overrides hard-coded default
LLM_PROVIDER	Identifies the backend provider (ollama or gemini)	"ollama"	Always used; determines code path and credential requirements
GEMINI_API_KEY	Authenticates requests to Google Gemini models	"" (empty string)	Only when LLM_PROVIDER is set to gemini

If you run the agent without setting any variables, it automatically falls back to the local Ollama provider with the gemma3:4b model.

How Configuration Loading Works in prompt.py

The prompt.py file serves as the single source of truth for LLM configuration. It imports os and loads variables with default fallbacks to ensure the application starts even when values are missing:

# prompt.py (excerpt)

DEFAULT_MODEL = os.getenv("DEFAULT_MODEL", DEFAULT_MODEL_NAME)
PROVIDER = os.getenv("LLM_PROVIDER", DEFAULT_PROVIDER.value)

# Validate provider against ModelProvider enum

if PROVIDER not in [p.value for p in ModelProvider]:
    PROVIDER = DEFAULT_PROVIDER.value

# Load Gemini-specific credential

GEMINI_API_KEY = os.getenv("GEMINI_API_KEY", "")

The code first attempts to read DEFAULT_MODEL and LLM_PROVIDER, falling back to DEFAULT_MODEL_NAME and DEFAULT_PROVIDER.value (which resolves to "ollama") respectively. It then validates the provider string against all values in the ModelProvider enum from models.py, reverting to the default if the user provides an invalid identifier.

Provider-Specific Configuration Requirements

Different providers require different levels of configuration. Understanding these distinctions prevents authentication errors when switching between local and cloud-based models.

Ollama (Default Provider)

When LLM_PROVIDER is set to ollama or left unset, the agent expects a locally running Ollama instance. No API keys are required, making this the zero-configuration option for development environments.

# .env - Ollama configuration

DEFAULT_MODEL=gemma3:4b
LLM_PROVIDER=ollama

# GEMINI_API_KEY can be omitted

Gemini Provider

Switching to Google's Gemini requires setting LLM_PROVIDER to gemini and supplying a valid GEMINI_API_KEY. The application checks for this key at runtime when the Gemini provider is active.

# .env - Gemini configuration

DEFAULT_MODEL=gemini-2.5-pro
LLM_PROVIDER=gemini
GEMINI_API_KEY=sk-your-actual-key-here

Practical Configuration Examples

Below are complete examples showing how to set these variables in different contexts.

Environment File (.env)

Create a .env file in the repository root based on the provided .env.example template:

# .env

DEFAULT_MODEL=gemini-2.5-pro
LLM_PROVIDER=gemini
GEMINI_API_KEY=sk-xxxxxxxxxxxxxxxxxxxx

Python Runtime Verification

You can verify that variables load correctly by importing them directly from prompt.py:

# verify_config.py

import os
from prompt import DEFAULT_MODEL, PROVIDER, GEMINI_API_KEY

print(f"Using model: {DEFAULT_MODEL}")
print(f"Provider: {PROVIDER}")

if PROVIDER == "gemini":
    assert GEMINI_API_KEY, "GEMINI_API_KEY must be set for Gemini models"
    print("Gemini API key configured successfully")

Command Line Export

For temporary testing without a .env file, export variables in your shell:

export DEFAULT_MODEL=gemma3:4b
export LLM_PROVIDER=ollama
python main.py  # Runs with local Ollama default

Summary

• Three variables control LLM configuration: DEFAULT_MODEL, LLM_PROVIDER, and GEMINI_API_KEY.
• Defaults fallback to Ollama with the gemma3:4b model when variables are unset.
• GEMINI_API_KEY is mandatory only when using the Gemini provider.
• All configuration logic resides in prompt.py, which validates providers against the ModelProvider enum in models.py.

Frequently Asked Questions

What happens if I don't set any environment variables?

The application runs using built-in defaults. According to the source code in prompt.py, it defaults to DEFAULT_MODEL_NAME ("gemma3:4b") and DEFAULT_PROVIDER.value ("ollama"), connecting to a local Ollama instance without requiring any API credentials.

Is GEMINI_API_KEY required for all providers?

No. The GEMINI_API_KEY variable is only required when LLM_PROVIDER is explicitly set to gemini. When using the default Ollama provider or any other future provider that does not require authentication, this variable can remain unset or empty.

Where is the provider validation logic located?

Provider validation occurs in prompt.py, where the code checks if the LLM_PROVIDER value exists within the list of valid providers defined in the ModelProvider enum (located in models.py). If the provided value is invalid, the system falls back to the default Ollama provider.

Can I use other LLM providers beyond Ollama and Gemini?

The current implementation in models.py defines the ModelProvider enum with specific values for Ollama and Gemini. While the architecture supports extension, adding new providers requires modifying the enum in models.py and implementing the corresponding logic in prompt.py to handle that provider's specific configuration and authentication requirements.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}