Support for Custom base_url to enable Local LLMs (Ollama, LM Studio, vLLM) #3
Description
Is your feature request related to a problem? Please describe.
Currently, the framework appears to be optimized for OpenAI's API endpoints (api.openai.com). Many researchers and security professionals prefer testing vulnerabilities on local instances or self-hosted models for privacy, cost, and legal reasons. Tools like Ollama, LM Studio, and LocalAI provide OpenAI-compatible endpoints (e.g., http://localhost:11434/v1), but often require changing the client's base_url.
Describe the solution you'd like
I would like to request the ability to configure the base_url and api_key via the configuration file or environment variables.
For example, when initializing the OpenAI client:
client = OpenAI(
base_url=os.getenv("OPENAI_API_BASE", "[https://api.openai.com/v1\](https://api.openai.com/v1)"),
api_key=os.getenv("OPENAI_API_KEY")
)
Additional context
This change would allow users to run MasterKey against:
- Ollama (Port 11434)
- LM Studio (Port 1234)
- vLLM (Port 8000)
This is crucial for testing "Attacker" vs. "Target" scenarios completely offline.