https://github.com/drake9098/vulnerabilitybot

A client-server structure to make queries and send it to an AI model
https://github.com/drake9098/vulnerabilitybot

cybersecurity llm llm-inference

Last synced: about 2 months ago
JSON representation

A client-server structure to make queries and send it to an AI model

• Host: GitHub
• URL: https://github.com/drake9098/vulnerabilitybot
• Owner: Drake9098
• Created: 2024-08-30T17:35:38.000Z (5 months ago)
• Default Branch: main
• Last Pushed: 2024-08-30T17:56:43.000Z (5 months ago)
• Last Synced: 2024-11-17T21:36:56.558Z (2 months ago)
• Topics: cybersecurity, llm, llm-inference
• Language: Python
• Homepage:
• Size: 13.7 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# VulnerabilityBot

Made in collaboration with Giuseppe Bellamacina: https://github.com/GiuseppeBellamacina

A project made of a client, a server, an AI component and a database, where queries and responses are stored. You can use the client to set the details of your queries, specifically aimed at fixing or mitigating software vulnerabilities, then send them to the server, who will create a record in the db table and give the query to the AI. The model will generate a reponse and send it back to the server, that will then update the record and set the request to completed.

The AI model that was used is Dolphin. It will process queries in batches of three, but you can change this if you'd like to by editing the config file. For the client, we chose to use the Streamlit library.

Other requirments are:

- quart

- uvicorn

- httpx

- pandas

- sqlite3

- all the needed langchain packages

Launching the start.bat file will automatically start all three separate processes, you'll then be able to upload queries by writing them down manually or uploading a json/excel file.

Note that this project is still a work in progress, we didn't test for bugs or issues estensively. Any contribution is welcome!