I want the opposite of the new iPhone Duo. Give me a blackberry that runs @omarchy
I jumped on @subquadratic Snap early on for Mandrel and now we have a new version Snap 1.1.
We use Snap for memory and to determine superseded context. Replaced a smaller reasoning model and much much cheaper.
Not all memory is the same!
If you don't get to experience this magical time every year, you should put this on your bucket list.
Blue Ridge Parkway!
I never worry about starting a new session, we work, I `/clear` and then `run boot up ref:resume`
Every project is the same
I can run multiple projects this way
We start right back up where we left off
Mandrel
This happens every time a new model is getting ready to come out. It's a pattern. I don't see this with my work but I can always count on X to inform me a new model is imminet.
My uneducated guess is they have to change the harness and tune it. Which changes how current model performs.
I also think this is severely over blown.
How to have a full-time job and still get work done.
Mandrel Punch Loop:
Scope the work->file punches with required 'proof'->Picker selects based on dependency order->Foreman builds->Critic inspects->Merge on pass or send back to be re-selected
Mandrel OSS - making good progress. Just a few more things and I will put Mandrel up against any other AI project management/memory tool.


















