Bulama is a simple Burp Suite Extension that can interact with locally hosted LLM using Ollama. Inference is done on your local machine without any remote server support, so its 100% free. This work is inspired by Lumos Chrome Extension.
Setup local ollama server for communicating with local model using this tool. Download and Install Ollama and the CLI here.
Set Origin Policy (Please be aware for security impact if trusting all origin, source here):
MacOS: launchctl setenv OLLAMA_ORIGINS "*" or set any origin launchctl setenv OLLAMA_ORIGINS "burp://*"
Linux: Edit the ollama.service using the following command sudo systemctl edit ollama.service and add the environment variables.
Environment="OLLAMA_HOST=0.0.0.0"
Environment="OLLAMA_ORIGINS=*"
Windows: Add the variables OLLAMA_HOST and OLLAMA_ORIGINS in Windows Environment Variable depending on your requirements
Download and Run Model Example:
ollama run llama3.2
Check if its returning 204 or 403 (or other error response code):
curl -X OPTIONS http://localhost:11434 -H "Origin: http://example.com" -H "Access-Control-Request-Method: GET" -I
Load the JAR file and see if it can connect to the local model.
Please note that depending on your computer capability the inference might need some time. So please wait a moment until you get response from your locally hosted LLM.