Headed to Lavaltrie to meet the team! Quick stop in Burlington, VT for a weekend with my wife and in laws.

🙏 if you're on HackerNews, I would appreciate an upvote on the blog post: news.ycombinator.com/item?id=4560...

PowerShell Universal joins Devolutions ✨

Together with @adamrdriscoll.com, we're building momentum for innovation and creating new exciting integration opportunities 💡

Read more in my latest blog post 👇
blog.devolutions.net/2025/10/powe...

PowerShell Universal joins Devolutions — a big step for IT automation.

Flexible orchestration today, deeper integrations with RDM (Avalonia) and Devolutions Gateway for secure just-in-time automation tomorrow.

Read the blog & share your ideas: devo-inc.net/4oifZIb

#PowerSh#PowerShellt#Automationhref="/hashtag/RDM" class="hover:underline text-blue-600 dark:text-sky-400 no-card-link">#RDM

I'm beyond excited to announce our acquisition of PowerShell Universal 🚀 — a big step toward simplifying IT automation and orchestration 🤖💡

Read the press release now 👉 devolutions.net/company/pres...

And look out for my blog post later today! ✍️

I'm very excited to continue to drive PowerShell Universal at Devolutions! By joining Devolutions, we're going to be able to scale and enhance PowerShell Universal in new and awesome ways. Look out for my blog post later today.

Big news: Devolutions has acquired PowerShell Universal—and its creator Adam Driscoll has joined our team.

SMBs/MSPs: stronger IT automation (centralized scripts, orchestration, web dashboards) + AI-in-context integrations across our stack, including RDM.

Read the news: devo-inc.net/46WDTn2

It's documented as a limitation, the current GitHub Copilot integration in SSMS doesn't have an agent mode: learn.microsoft.com/en-ca/ssms/g...

🚨 It's another two-for-Tuesday over in the SSMS team!

Announcing the release of SSMS 22 Preview 3 and SSMS 21.6!

SSMS 22 Preview 3 includes:

- GitHub Copilot in SSMS

- ARM64 support for core SSMS scenarios

Start with Makena's post: techcommunity.microsoft.com/blog/sqlserv...

Enjoy! 👯‍♀️

But in other cases, it's more than just copy/pasting. I used the Atlassian MCP server in Claude last week to hunt JIRA tickets I could close. Instead of getting links to open 20 URLs one by one to close them, I just asked Claude to close them for me. Much faster, and I approved the list of tickets

You can vet by looking at it the same way in agent mode, unless you start allow listing tools and commands. It's confirm-by-default unless you change it not to be. The ask mode just forces manual steps in between, where you have to copy/paste commands to run instead of approving commands to run

MCP clients require you to confirm before running tools by default - you can allow list them, or in the case of generic tools that run commands, you can filter commands to be automatically allowed. It's a work in progress but it's so much easier than copy/pasting all the time

We have an MCP server in RDM that can already do better than this with SQL Server, here's a demo from just two weeks ago: m.youtube.com/watch?v=LEXb...

I just installed SSMS 22 Preview 3 with the GitHub Copilot extension to try it out: it looks like a simplified version of the one from Visual Studio, and unless I'm mistaken, it's in "ask" mode only, with no agent mode? I was expecting it to be able to run the queries for me 😐

Oh... It's been bought by Cursor? cursor.com/blog/superma...

Also, this is the first time I hear about Supermaven, I'll have to take a look. There are so many services these days it's hard to keep up: supermaven.com

I tried M365 Copilot and definitely wasn't impressed. You'd be better off with ChatGPT + the Outlook connector

I doubt you wouldn't know about it, as it's not cheap

For PowerShell, the experience has been the same for me: ChatGPT definitely isn't the best, it frequently has issues with character escaping for me. Claude, on the other hand, is vastly superior to ChatGPT when it comes to generating PowerShell scripts

Trying to limit the amount of data fed into the LLM makes sense given the cost and the context window. I just wonder if it might be worth looking into custom hosting of open source models where it might be easier to just aggressively load all the file contents for a decent price?

The main limitation I've hit is that once VSCode finds code of interest, it appears to be capping the chunk size it will read from the file to 2000 lines. I don't notice it unless I need to explicitly feed it a very large file. Is there a way to tweak this to ingest larger chunks of data at once?

🧩 I work with a very large codebase and I’m exploring which AI coding assistants truly scale — especially for code search, context length, and chunking.
So far, VS Code + GitHub Copilot + Claude Sonnet 4.5 has performed best for me ⚙️
👉 What setups work best for you?

You can also use -ErrorAction SilentlyContinue to avoid getting the error if it's not found, and then just test the Get-Command result: gist.github.com/awakecoding/...