FW-GPT Troubleshooting Assistant is a web app that helps network administrators troubleshoot common networking issues using natural language processing. This app uses the FW-Gate API and SSH to gather debug commands and sends them to the ChatGPT API for analysis. The app then provides troubleshooting advice based on the analysis results. Additionally, there is a chatbot available that enables you to use a LLM combined with the documentation.
To use FW-GPT Troubleshooting Assistant, simply select a problem category (e.g. VPN issues) from the sidebar menu, and choose the problem type. The app will gather the debug output from your FW-Gate firewall and send it to the ChatGPT API for analysis. The app will then provide you with troubleshooting advice based on the analysis results. If you are not happy with the response from chatGPT you can modify the prompts used in the server folder and rebuild the docker image.
- System - FW-guard, High Memory, High CPU
- Connectivity - Packet Flow
- Routing - BGP Down
- VPN - VPN Down
- Network - Interfaces
- ChatMode
If the community expresses interest in this tool, I will incorporate additional categories and problem types.
FW-GPT Troubleshooting Assistant requires no dependencies to be installed. Instead, it can be run as a Docker container using docker compose.
To run FW-GPT Troubleshooting Assistant, simply follow the steps below:
Copy the docker-compose file from the repo and run docker compose up. Make sure to modify the env with your openai api key. If you want to build the images locally, clone the repo and use the docker-compose-dev file.
docker compose up
Connect to the FW-gpt-nginx reverse proxy container on port 3050 and you are all set.
Contributions are welcome! To contribute to FW-GPT Troubleshooting Assistant, simply fork the GitHub repository and submit a pull request with your changes.
FW-GPT Troubleshooting Assistant is licensed under the MIT license. See LICENSE for more information.