Pinned Tweet
BTC cycle bottom: $40,600 to $50,700, 26 Sep to 31 Oct 2026.
Low so far $58,525 and plenty reckon that was it. In which case I've missed the lot.
My quant tool says lower. Quant's generous.
Posting before the fact so it can be marked after. Egg on my face or a grin ;p
Really like this. Especially the last paragraph.
If an agent writes the tests too, I look for the junk first. Zero assertions. Tests that can't fail. Coverage that looks fine and proves nothing.
Power tools are fine. I'm still the one responsible if it goes wrong ;p
Disengaging from syntax is not disengaging from engineering.
I’m working on a significant project at the moment. It’s in a pre-production state so far. More than a prototype, less than ready for real users.
I’ve been making multiple passes over the structure for each of the following concerns: Architecture, coverage, race & concurrency, security, API, etc. Each of these requires significant interrogation and review of verification procedures; but not review of the code.
At base are the /ub-mode (GitHub.com/unclebob) rules that keep the agents disciplined as far as coding structure and tests are concerned. But above that it’s me making sure that all the bases are covered well enough to keep my client safe from stupid sloppy mistakes.
I am the engineer. I am responsible. The agents are just my tools. They are power tools, to be sure. But power tools in the hands of novices result in missing fingers. So I am very careful.
1
1
18
Bitcoin's Stablecoin Supply Ratio fell eightfold between its two measured bear bottoms: 20.14 to 2.52.
Two readings cannot set a reliable threshold. The dashboard tracks it outside the score.
1
12
Moinerus | Dev, AI, Crypto retweeted
“Your job has fundamentally changed. It is no longer to write code, and LLMs are not here to help you write better code faster.
Your job is to help the LLM write better code faster.”
Bingo.
Replying to @josephfcox
Here is Proton CTO's AI manifesto, that I obtained multiple copies of 404media.co/manifestos-fear-…
Scored here: nitter.gq/moinerus/status/211687…
Scoring my call, as promised.
The lowest daily close in the window was $77,110. Not even close. June's $58,525 still holds.
Quant was generous. My rules weren't ;p
2
Scoring my call, as promised.
The lowest daily close in the window was $77,110. Not even close. June's $58,525 still holds.
Quant was generous. My rules weren't ;p
Bitcoin may already have bottomed, and I may have got the call wrong.
Basically, I don't know. Quant is apparently still hard, even with all these AI advances ;p
So I've built a cycle comparison tool to look at a few ways this could play out.
2
1
30
I can line up the same paths from June's low or October 2025's high, keeping percentage moves and calendar days.
Some cases overlap. Old cycle gains can produce daft prices when rescaled. I haven't shown that any of this can predict the next bottom.
1
10
The comparisons are now live: peakandtrough.com/#compariso…
My original call is here: nitter.gq/moinerus/status/209872…
What would make you rule out June as the bottom?
BTC cycle bottom: $40,600 to $50,700, 26 Sep to 31 Oct 2026.
Low so far $58,525 and plenty reckon that was it. In which case I've missed the lot.
My quant tool says lower. Quant's generous.
Posting before the fact so it can be marked after. Egg on my face or a grin ;p
6
I had a hunch a pipeline was flaky for years. It wasn't taken seriously until I pulled the run data and plotted it.
The DevOps Handbook's Second Way is about feedback. Get the data, test the theory, then decide what to fix.
1
17
Exchange net flow was negative at all three measured Bitcoin bear bottoms: -0.232%, -0.056% and -0.262%.
The 2018 low sat close to zero. A much deeper cut-off would have missed it.
Done isn't merged. Before users arrive, we need dashboards, alerts, a runbook and a rollback path.
Developers should follow the change past the deploy and into the on-call rota.
1
18
Hash Ribbons was negative at all four measured Bitcoin bear bottoms.
The readings ranged from -14.70 to -0.86. A zero flash line catches each of them.
1
2
34
An andon cord nobody answers is decoration.
When main goes red, we need to stop merging and fix the build. The DevOps Handbook's Second Way is about fast feedback. Who calls the stop, and who picks up the fault?
1
23
Moinerus | Dev, AI, Crypto retweeted
If you're at a company with a large legacy codebase and finding it very difficult to move at the velocity of AI while be confident you're not breaking stuff, read this.
Your problem is confidence and the solution is verification. And I'm going to tell you how to get there.
Your agents need a way to verify their work just like humans. When you were building all this stuff by hand you never relied on just looking at the code and test output. You would manually test it. Run it with mock data. Open it up and use the UI. Seed the database with a variety of data. Sometimes you'd even push it to a production-like deployed environment to verify things work outside of local.
Many agents are capable of doing these things too, but it's harder than it should be and some of them aren't so proactive. If you give a paved path for this though, they will.
An idea I got from @poteto which has been very helpful for my agents in this is the control-kody CLI which is a script my agents maintain to make it easy to do all the verification I mentioned above (github.com/kentcdodds/kody/b…). This makes it so much easier for agents to create good PRs in the first place which is good even if you're not ready to just let the agent start merging stuff without you.
So if I worked on a legacy codebase and was watching tech twitter posting about the dozens of PRs each of them is merging every day and wished I could ship improvements to my codebase at the velocity of dozens of improvements each day, the first thing I would do is point my agent at this post and ask them to help me get started.
I strongly believe that any codebase can be turned on autopilot eventually and for a codebase that's earning money and needs to be maintained, the cost is more than worth the improved velocity. Get yourself up on a higher level of abstraction to leverage your agents better.


















