[Enhancement]: Openvas api intergration

[s-b-repo]

Target Component

Security Tools Integration

Enhancement Description

will help find cves and with pentesting

Technical Details

No response

Designs and Mockups

No response

Alternative Solutions

No response

Verification

• I have checked that this enhancement hasn't been already proposed
• This enhancement aligns with PentAGI's goal of autonomous penetration testing
• I have considered the security implications of this enhancement
• I have provided clear use cases and benefits

[asdek]

Hello @s-b-repo,

OpenVAS requires systemd to run out of the box. Before implementing this for all users, I'd suggest you experiment to see if this functionality is actually useful for AI agents and whether they can use it correctly.

Setup Instructions:

You can use the base image vxcontrol/kali-linux:systemd and install OpenVAS on top of it.

Option 1 - Package Installation:

FROM vxcontrol/kali-linux:systemd

RUN apt update && apt install openvas gvm -y

# Copy modified entrypoint that starts both systemd and gvm
COPY container-entrypoint.sh /usr/local/bin/container-entrypoint
RUN chmod +x /usr/local/bin/container-entrypoint

Option 2 - Build from Source:
Use the build approach from: https://github.com/greenbone/openvas-scanner/blob/main/.docker/railguards/debian_stable.Dockerfile

Important Notes:

• Base image files: https://github.com/vxcontrol/kali-linux-image
• The image has a custom entrypoint /usr/local/bin/container-entrypoint (see here)
• You need to modify this script during your image build to start both systemd and gvm

Integration:

1. Set DOCKER_DEFAULT_IMAGE_FOR_PENTEST=myorg/kali-linux:openvas in your .env file
2. Restart PentAGI - new pentest tasks will use your custom image
3. Modify agent system prompts in the web UI or add OpenVAS usage to your custom flow prompts

I'd like to see results from this testing before making changes for all users. Can you share how the agents perform with OpenVAS integration?

[s-b-repo]

Hello @s-b-repo,

OpenVAS requires systemd to run out of the box. Before implementing this for all users, I'd suggest you experiment to see if this functionality is actually useful for AI agents and whether they can use it correctly.

Setup Instructions:

You can use the base image vxcontrol/kali-linux:systemd and install OpenVAS on top of it.

Option 1 - Package Installation:

FROM vxcontrol/kali-linux:systemd

RUN apt update && apt install openvas gvm -y

Copy modified entrypoint that starts both systemd and gvm

COPY container-entrypoint.sh /usr/local/bin/container-entrypoint
RUN chmod +x /usr/local/bin/container-entrypoint

Option 2 - Build from Source: Use the build approach from: https://github.com/greenbone/openvas-scanner/blob/main/.docker/railguards/debian_stable.Dockerfile

Important Notes:

* Base image files: https://github.com/vxcontrol/kali-linux-image

* The image has a custom entrypoint `/usr/local/bin/container-entrypoint` (see [here](https://github.com/vxcontrol/kali-linux-image/blob/master/container-entrypoint.sh))

* You need to modify this script during your image build to start both systemd and gvm

Integration:

1. Set `DOCKER_DEFAULT_IMAGE_FOR_PENTEST=myorg/kali-linux:openvas` in your `.env` file

2. Restart PentAGI - new pentest tasks will use your custom image

3. Modify agent system prompts in the web UI or add OpenVAS usage to your custom flow prompts

I'd like to see results from this testing before making changes for all users. Can you share how the agents perform with OpenVAS integration?

if we train them its possible how it works and just make a api for openvas comptiablity to relay command based on what the ai want to do