Muahaha, interactive Live Response inside of PowerShell!
@NathanMcNulty2万 位关注者
31,165 次观看
24 条评论
评论来自 X (Twitter) 上的原帖

@HackingLZ
Do you do offense full time now?
6 个赞

@NathanMcNulty
🤣
1 个赞

@NathanMcNulty
I read a lot of what the AI reasons about in the terminal, and I've learned a ton about techniques I never would have thought about It's crazy how short that trip is from "I just want this to be transparent to the user" to "I just built an OST" 😂
3 个赞

@fabian_bader
Very cool. One more conference and then I will have a bit more time to add stuff to XDRInternals as well. 😂
3 个赞

@NathanMcNulty
I apologize for the flood of stuff I'm about to push to the branch... I need to figure out the best naming for all this stuff, like the best verbs and heirarchy. I'll publish the few branches, we can discuss whenever :P
1 个赞

@BertJanCyber
👀
1 个赞

@NathanMcNulty
👀
1 个赞

@Cyb3rMonk
@BertJanCyber how about sending the same command to multiple sessions at the same time?
2 个赞

@NathanMcNulty
@BertJanCyber I should create a separate cmdlet for that, but it would be very doable The more difficult part of the logic will be tracking API limits or handling the quotas when they are inside jobs and not visible to the user, but I think I can work around that
1 个赞

@Cyb3rMonk
@BertJanCyber You mean Claude can work around? 🤪
1 个赞

@NathanMcNulty
@BertJanCyber Haha, yeah, it'll do most of the fixing for me :P My current workflow, though could be improved, is to capture via browser using our helper extension, and let AI scaffold the cmdlet. Then I edit and test until I hit issues or big changes, then pass back to AI, then I finish it.
2 个赞

@NathanMcNulty
@BertJanCyber I would say 90% is written by AI, but I delete probably 20% of what it writes by refining things or fixing issues it didn't know about There is no way I could develop this in my free time without AI, the time commitment would be too much For tracking:
2 个赞

@m19o__
Red teamer in heart
1 个赞

@SecurityJosh
This is cool, can you execute arbitrary PS? (Maybe by uploading the command as a .ps1 file to the live response library and then executing the run command against it in the background)
1 个赞

@NathanMcNulty
The reason they give us a limited set of command instead of a full shell is to ensure commands are pre-meditated, pre-authorized, and consistent But if we have a second shell (or way to bg the current task), then we can definitely do that I could add a special command in shell
1 个赞

@AlfredBernhardN
Nice. How to replicate it? Is there a full guidance? Github repo to understand?
1 个赞

@NathanMcNulty
I implemented as part of our XDRInternals PowerShell module, currently waiting for review but should be in soon :)

@AlfredBernhardN
Thank you
1 个赞

@mrmichaeljstew
This is the kind of quality-of-life improvement that actually changes incident response workflows. The ability to stay in PowerShell without context-switching out to the portal is underrated — especially when you're working at 2am and muscle memory matters most.
1 个赞

@NathanMcNulty
Humans can only click on things so fast, and we are further slowed down by the unnecessary content in the browser Beyond a human being able to grab data, grep, pipe, etc., parts of this module opens an opportunity for application and agent development that we never had before :)

