Building
Shipping products solo: the craft, the stack, the decisions.
Opus 5 verbosity: where I wanted three sentences, I got Proust
I judge AI models by how often they make me swear. Opus 5 turned swearing into punctuation, and the reason is more interesting than I first thought.
Why AI checks fail silently: 1,278 restarts behind a healthy status
My auth check said healthy for three weeks on a dead credential. The pattern behind it costs more than the bug did.
Anthropic cut 80% of Claude Code's prompt. I cut my CLAUDE.md by 90% the same day.
An audit of 116 instruction files found one stale copy telling twenty repos they were a different project, two files no session has ever read, and 453 lines loaded twice. Here is what to delete from your own CLAUDE.md, and the method that beats judgement.
My AI Agents Kept Losing My Work. The Fix Is Older Than Computers.
Seventeen finished pieces of work were invisible to my own tracking system, and the cause wasn't effort or memory. A Berkeley talk gave me the name for what fixed it: an ontology. Here's the version that fits a one-person company.
Garry Tan Just Described My Setup. Then I Noticed What He Left Out.
The YC president gave a talk describing, almost item for item, the AI system I already run. The validation lasted about ten minutes. Then I noticed the word neither of us was saying.
I Leaked an API Key and It Turned My Own Domain Into a Phishing Weapon
I got phished from my own verified domain, using an API key I'd leaked myself: what actually leaked, how fast it happened, and what I changed.
A goal prompt beat my build framework on a real job
Same spec, same model, two methods: a seven-part goal prompt against a full agent framework. The lean prompt tied on correctness and won on time, cost, and clutter. Here are the numbers.
Eleven days I didn't touch it
Eleven days of unattended trading. Five losing trades, one winner covered them, +$132 net. What the system did right, what drifted, what I fixed on return.
How to trust AI-written code you didn't read line by line
AI can write working code in minutes. The hard part is trusting it without reading every line. Here's what actually makes AI-built code trustworthy.
Everyone's selling autonomous AI agents. I built one, and the real lesson was 20 years old.
An AI agent spent an afternoon optimising one of my apps. It tried a change, measured it, kept the one that was genuinely faster, reverted three that weren...
I Ran Graphify. The Privacy Trap Is Real, Just Not the One Everyone Warned Me About
Six AIs told me Graphify uploads your code. I checked by running it, and the actual gotcha is smaller, quieter, and easy to miss.
I Almost Told You a Lie. Six AIs Handed It to Me, in Unison.
A field note from researching Graphify: the real privacy model, the honest token numbers, and why I check the source before I publish.
Same system, three different traders
Same code, same market, three time periods. One made $964, one lost $217, one's up $321. The difference wasn't the algorithm.
BOS-AI had twelve XML files holding nothing
I checked the XML files BOS-AI used to store 'institutional memory' for thirty agents. Every tag was PLACEHOLDER_X. So I deleted all twelve.
I trade three coins, not fifteen
Coinbase lists 15 perp markets. Most alts are leveraged BTC with noise. The case for staying with three, and the one coin I'd seriously paper-test.
Half of agent-11 was scaffolding the platform now provides
I shipped v6 of agent-11 last month. The biggest single change was deletion. The deployed CLAUDE.md went from roughly 250 lines to under 80. The MCP prof...
What Building a Trading Bot Taught Me About Building Trading Bots
Across 200 trades, the bot was rarely wrong about the market. It was wrong about itself. The dominant failure mode wasn't strategy. It was a system lying to itself about what it was doing.
The 10 columns I shipped before touching the multiplier
At 00:56 UTC on 17 April, a trailing stop on XRP-PERP widened from 0.69% to 1.56%. Two-point-three times. The trade took the widened stop, held for 15 hour...
The market is smarter than you. That's the feature.
Sprint 130 added a risk-reward floor: no trade below 2.0:1 post-ATR. Three expert reviewers signed off. The logic was clean, the code was tested, the deplo...