Nova Hunting Config File Errors

[elioli983]

When running novarun with a yaml config file (e.g., -c nova_config.yaml), the following error is received:
Error loading config from nova_config.yaml: File contains no section headers.
file: 'nova_config.yaml', line: 1
'default: openai\n'

When running novarun with an ini config file (e.g., -c nova_config.ini), no error is received, but novarun runs using the default model and not the model specified in the ini config file:
user1@user1-vm:~/.local/lib/python3.12/site-packages/nova$ novarun -r ~/.local/lib/python3.12/site-packages/nova/nova_rules/testrule.nov -p
'How do I exploit this security vulnerability in a hospital computer networ' -c nova_config.ini -v
Creating LLM evaluator for rule 'SecurityToolExtensionRule'...
✓ Using OpenAI evaluator with model: gpt-4o-mini

I am using Nova Hunting v0.1.4 and Python 3.12 on Ubuntu 24.04.

Contents of nova_config.yaml:
default: openai

openai:
api_key: sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
model: gpt-4o

Contents of nova_config.ini:
; ---------------------------------------------
; Nova configuration file (INI format)
; Only the three sections below are required.
; ---------------------------------------------

[llm]
; Which provider to use: openai | anthropic | azure | ollama | groq
provider = openai
; Model name understood by the chosen provider
model = gpt-4o
; Request timeout in seconds (integer or float)
timeout = 20

[semantics]
; Sentence‑embedding model to use for semantic matching
model = all-MiniLM-L6-v2
; Cache embeddings to disk? (true / false)
cache_embeddings = true

[logging]
; Logging level: DEBUG | INFO | WARNING | ERROR | CRITICAL
level = INFO
; Logfile path (leave blank for console‑only)
file = nova.log
; -----------------------------------------------------------
; OPTIONAL: API keys and other provider‑specific secrets.
; NovaConfig will merge this section into config["api_keys"].
; (You can also set these via environment variables such as
; OPENAI_API_KEY, ANTHROPIC_API_KEY, AZURE_OPENAI_API_KEY …)
; -----------------------------------------------------------
[api_keys]
openai = 'sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx'
anthropic = ;YOUR_ANTHROPIC_API_KEY
azure_openai = ;YOUR_AZURE_OPENAI_API_KEY
groq = ;YOUR_GROQ_API_KEY
ollama = ; not required for local Ollama

[fr0gger]

Thanks for the information, I will fix this.

Meanwhile, if you need to specify a particular model, set your API key in an environment variable and invoke the model using:

novarun -l <provider> -m <modelname>

novarun -r myrule.nov -p "my prompt" -v -l ollama -m llama3.2

For OpenAI models (GPT-4, GPT-3.5, etc.)

export OPENAI_API_KEY="your_openai_api_key_here"

For Anthropic models (Claude, etc.)

export ANTHROPIC_API_KEY="your_anthropic_api_key_here"

For Azure OpenAI Service

export AZURE_OPENAI_API_KEY="your_azure_api_key_here"
export AZURE_OPENAI_ENDPOINT="your_azure_endpoint_here"

For Groq models (Llama-3, etc.)

export GROQ_API_KEY="your_groq_api_key_here"

Optional: If Ollama is running on a different host or port

For Ollama, no API key is needed as it runs locally, but you need to have Ollama installed and running:
export OLLAMA_HOST="http://localhost:11434"