AnthonyIsaacson

SDK Test Tweet - 2026-01-21T23:43:37.388Z

hello world

open twitter
50 straight posts about Claude code
close twitter

- Covid
- Epstein files
- Cluely
- AI agents
- (new) Claude code
Enough.

99% of "narrative making" in crypto is just ignorance
"we're the first xyz"
almost everything was tried in 2021, you just never bothered to google it bro

"you're absolutely right" is a crutch llms use to get around partial understanding of problems
jumping in to 1M+ line codebases w/ 4k lines of context on a whim is impossible without assuming the user's prompt is correct
agent mode doesn't solve this. The LLM's prior is "the user is right", which is hard to refute
the next "leap" will be when ai can challenge our assumptions, but I don't think we're getting there with the current patterns.

"there's zero innovation in the sf tech scene, just performative losers building b2b saas"
- performative loser in sf, building b2b saas

x402 scan is the answer to the question
> who is willing to lie to you for money?

rare claude holy shit (?)

the right thing to build is painfully obvious right now

Best part about @mirra?
Our "Get Started" button goes to a product, not a github repo.

a lot of what I’m seeing get built these days is probably making Kim Jong Un quite excited.

unsolved problems w/ x402 as written -
> finality is "great", but llms fail a lot. How do you handle errors?
> most llm interactions are multi-round. Pay-per-message vs pay-per-result.
> if one agent can perform many tasks worth $0.2, $0.5, $1.00, how do you price the endpoint?