I like contributing to this project.
I just though about implementing some more features. I share here, in case someone wants to contribute or when I have free time:
- Implement MCP server on the project. It is interesting, because, we can expose a pentestagent as a MCP server for focused pentesting tasks. Could also be used to create several pentesagent instances and connect them together via MCP allowing to have a pentestagent crew, but with a more standardized modularized approach.
- For /interact mode, sometimes, the model ask to execute several tools at once. They are executed sequentially, one after the other, improve it to execute them concurrently.
- Add a command to save and restore conversation memory to / from json file.
- For some commands that may take long to execute, get real time feedback (such as a long nmap), instead of waiting for the tool to finish to print the result.
- Separate in different workspaces, different pentest sessions (there is already some code related to /workspaces, but I don't know how it works yet).
- Allow the possibility to spawn tasks that can take long time and let them executing in background. Implement a way for the LLM to check the status and the output at the end.
I like contributing to this project.
I just though about implementing some more features. I share here, in case someone wants to contribute or when I have free time: