Augmenting agents with cyberware C# is not like the other languages Dev, founder, dropout (neovim + 3x Arch btw)
Joined November 2025
- Tweets294
- Following476
- Followers72
- Likes3.2K
Adding Microsoft Entra ID to my "Minecraft bloomberg terminal" just so i can burn all these tokens xD
Claude draws now.
Ask anything with a shape in it and the picture is in the reply, while it's still typing. A fork in a decision, how your services hang together, a screen before it's built.
Under it: Claude writes Graphviz or SVG, drawer renders it in the terminal, in place. One binary, kitty or ghostty.
/plugin marketplace add Sureffi/cyberware.sh
/plugin install drawer@cyberware
github.com/Sureffi/drawer
Even more Arch machines lol Starting to see the vision with cloud agents ngl
github.com/Sureffi/hetzner.s…
This is the future. .NET and nuget, local package management
New on our Engineering blog: Creating in-house source control.
In May, we merged around 2,500 code changes per day. In September, we merged more than 30,000 per day. As a result, we became bottlenecked by developer infrastructure built for human (rather than agent) scale.
Here, we share how we built a new version control system—and fully migrated our monorepo from git—in five months: anthropic.com/engineering/re…
Sureffi retweeted
The false narrative around AI is that you don't need to learn anything since the AI agent will do it for you.
The reality is that having domain expertise:
1- Makes the results of the AI much more precise and reliable and aligned with what you want
2- Reduces unknown errors (in software engineering in particular but in other fields as well)
3- Reduces LLM token spend because you need fewer back-and-forths with the agent
If anything, domain expertise is more important than ever.
A doc's meaning is relative to its interpreter, so there's no true "good" documentation. There's only documentation that's efficient relative to *this* reader.
Sureffi retweeted
We deleted 800k lines of unit tests from the @sazabi monorepo earlier this week.
Three hypotheses:
1- Unit tests "lock-in" slop code by making it hard to change or remove
2- Unit tests slow down dev cycles by making your agent do more work locally and by increasing CI times
3- Unit tests waste money by making your agent generate more tokens and by adding useless CI runs
Will report back soon with the results!
Yk those docs that frontier models like to write for your codebases? It's making them worse btw, measurably so
