Founder at FirePress: we host websites + memberships + payments + mailing list thanks to Ghost. 🤖 CiCD everything. ☁️ Internet's opinion isn't my own.
Montréal, Canada
Joined March 2009
- Tweets7.2K
- Following327
- Followers702
- Likes4.8K
Been chewing on this, and I think @poteto nailed it.
Formal verification lives in pillar #2. On one end, verification loop checks. On the other, formal verification checks every possible inputs.
New on our Frontier Red Team blog: using Claude to formally verify software at large scale.
Formal verification can prove that code is free of certain bugs, but historically it has only been feasible for small codebases (or required a lot of expert time).
With AI models, however, this is beginning to change. In collaboration with @theoremlabs, we used Claude to prove security properties in over one hundred real-world open-source libraries over the span of a few weeks.
Read our research in full here: anthropic.com/research/forma…
Pascal Andy retweeted
New on our Frontier Red Team blog: using Claude to formally verify software at large scale.
Formal verification can prove that code is free of certain bugs, but historically it has only been feasible for small codebases (or required a lot of expert time).
With AI models, however, this is beginning to change. In collaboration with @theoremlabs, we used Claude to prove security properties in over one hundred real-world open-source libraries over the span of a few weeks.
Read our research in full here: anthropic.com/research/forma…
No mcp needed. Use t3-cli !
I made a Grok @bot that can drive @t3codes for me via the MCP server. It starts threads, monitors status and can ping you for answers to questions.
Bot template: x.ai/bot/_5NG_WsrOdQs2jtUhZe…
Great way to get access to more models using your other subscriptions. But help centralize work in T3Code which is great if you already use it.
Built on T3 Code, my SDLC gets stronger every day. Each pass adds more depth.
Now I want to push it further: a dashboard that shows what’s happening across all my brigades, live and on replay. Let me know if you see a way to improve the prompt below 👇🏻
Orchestrator V2 is now out on the stable channel in @t3dotcodes! We also thought it was time we drop the Alpha branding, so we call this release 0.1.0 Beta.
Mobile store builds are also updated. Versions 2.0.0 are available to everyone on both App Store and the Play Store!
So meanwhile, as I'm writing this, the EPIC 1 was shipped. And here's a good example of parallel work in the SDLC.
The Head Chef shows that one epic out of six is shipped.
The Chef caring for EPIC 2 is saying that zero out of two PRs are shipped.
At the bottom, you can see that there are two Cooks working in parallel.
and yeah .. can't get enough of this sunny side up eggs emoji 🤗
Now, I understand that it might look like overkill for many of you, but it's really when I want to run big chunks of work.
in this example, I'm running six epics with about 56 issues across them. Eventually, I will be able to run milestones. In this case, my six epics could be in one milestone, and I could ask the brigade to run many milestones.
github.com/pascalandy/nicely…
One could ask, "Hey, why don't you use beads and still bother with all that agile methodology?"
The reason is I still want to make sure that the way I decontruct the work after what I have in mind. I still need to see it. Using Beads at this point would be too soon for me. So once my $grooming-mode skill is mature, I think I will upgrade to a system such as Beads.
Then the head chef creates one or many chefs. Usually, when we have many chefs, it's because we can do some work in parallel.
The Brigade mode is where all the magic and the control plane is taking over. We have three roles in the brigade. The first role is the Head Chef
Pascal Andy retweeted
Orchestrator V2 is now out on the stable channel in @t3dotcodes! We also thought it was time we drop the Alpha branding, so we call this release 0.1.0 Beta.
Mobile store builds are also updated. Versions 2.0.0 are available to everyone on both App Store and the Play Store!
clearly something is wrong here with @grok usage ..
For the context, I sign up using a 2 month promo in X.
I haven't been a believer in AI assistant bots. Hermes, OpenClaw… they never clicked for me
But with the latest upgrade from @bot bringing SOTA models, maybe we have something here
So I cloned the Engineering Lead from @poteto. Its first assignment: observe how I work with my tech lead and my reviewer(s), and learn what's working and what isn't
Nothing flashy on the surface. The interesting part is underneath: my tech lead can actually drive @t3dotcodes
It sees every conversation, every archive, every glitch I run into, and I'm pushing to see how far it can go. The inception is real
Let's see where this goes
Refining your skills is half the job. The other half is making sure the clankers actually follow them.
Update the skill. Check the clankers. Update the skill. Check the clankers. Update the skill…
Who doesn't like a good vanity metric? 😅 Mine: avg agents launched per day. Take your 5 best days out of 7 (2 slow days a week is fair), average them.
That's your weekly number.
Pretty good idea of how much AI you're actually leveraging. What are you guys tracking? 👇