Meza (@vsbmeza) header
Meza (@vsbmeza) profile picture

Fostering highly efficient software teams focusing on quality output, culture and sustainability. Neurodiversity Advocate

London, UK
Joined May 2009
Photo by Meza (@vsbmeza)Photo by Meza (@vsbmeza)Photo by Meza (@vsbmeza)Photo by Meza (@vsbmeza)Photo by Meza (@vsbmeza)Photo by Meza (@vsbmeza)Photo by Meza (@vsbmeza)Photo by Meza (@vsbmeza)Photo by Meza (@vsbmeza)Photo by Meza (@vsbmeza)
The MacOS support for RunnerFoundry has landed github.com/stateshifters/run… If you're a self-hoster, you can now have all 3 operating systems provisioned for yourself. Next up: forgejo support soon...
4
94
What a coding agent reads before it builds a screen, and the moment it stops to ask.
1
1
1
53
I've been working on, and enjoying the benefits of this for the past good few months now and I'm ready to share it with the world. Allowing for a fully self hosted, dynamic @github runner pool even for your personal repositories! github.com/stateshifters/run…
1
1
59
- It does not use ARC - It does not use Kubernetes - It can provision any vm via libvirt - It can provision docker runners - You have to maintain your runners, this purely orchestrates - As a result of the above, you're 100% in control (and responsible for) what you do with it
1
17
Oh, and if it weren't clear, steady state is 0 running anything. Runners get provisioned and assigned based on demand. AND you can cluster them if you have multiple hosts that you want to enroll in the pools
5
Replying to @thsottiaux
@thsottiaux I get that dot needs compute but so does 6.1sol! It's the slowest it's ever been!
1
1
115
4.5 minutes... for a single readme commit
19
"That was my mistake" should come with token refunds
1
65
They're completely unaware that I am fixing a plethora of @AikidoSecurity dependency violations right now with @OpenAI codex remote on multiple threads.
2
109
I needed a @charmcli agent skill. I couldn't find one after 2 minutes of searching, so I made my own to keep my agents on track. Maybe it's useful for anyone else? github.com/meza/skills/tree/…
1
86
I'm pursuing a state where my agents understand at least the basics of what I'm looking for as their output. A new skill I've created is a catalog of "concerns" that can be attributed to a clean codebase. github.com/meza/skills/tree/…
2
2
93
@unclebobmartin you might get some mileage out of this one
13
I use the same set of concerns with the `/review-swarm` (that is also available as a plugin there) which dispatches agents with a single lens at a time. It's been really great at finding the gaps in the output of the implementer agents.
1
31
Seems like you truly must separate the implementer from the reviewer. The narrower lens they have, the better they perform that one thing.
23
what in the claude...
1
60
let me put the actual delta on the table
12
This is an informational post about informing you that a post has been issued for your information. the hell @Microsoft ???
56
This is another that is turning out to be immensely useful. This helps an agent assess code-review feedback before presenting or acting on it. It treats reviewer comments as evidence to verify, not as an automatic work queue or permission to change code github.com/meza/skills/tree/…
Your coding agent "fixed" the lint error by suppressing it again? Mine too, so I made a skill for that. Fix the cause. Don't game the linter. Been using it for a few weeks now, this is my most precious skill in my arsenal. github.com/meza/skills/tree/…
1
117
Your coding agent "fixed" the lint error by suppressing it again? Mine too, so I made a skill for that. Fix the cause. Don't game the linter. Been using it for a few weeks now, this is my most precious skill in my arsenal. github.com/meza/skills/tree/…
1
1
2
203