defaulting codex to gpt-6.2 astra? check openai's frontiercode chart. sol at max effort about matches astra's best, for a fraction of the cost
how i'd route it:
- sol for the daily loop
- astra only when sol stalls
- ultrafast per call, not global. 3x price now, 6x from march
everyone clipped the 100 years line. the real tell came first. sam altman on ai in science (via @kimmonismus):
"they can discover things that they just weren't going to be able to figure out otherwise"
judging agents on hours saved? also track what shipped that never would have
vps hosts still hand you a server that's open to the whole internet, then people park an agent with api keys on it. lock it first:
- firewall: deny all inbound
- ssh only over tailscale
- web through a tunnel, not an open port
the scanners find your box before your users do
you don't need to give an agent your whole mac to stop the permission spam. macos 27.3 lists what full control means: messages, mail, photos, every keystroke
give it its own box:
- separate user or vm
- scoped tokens, not your logins
- auto-allow reads, ask on deletes and sends
if you're trying iphone duo in claude code, grep for hardcoded screen sizes first. the demo app drew on a fixed phone-size canvas and sat in a tiny box mid-screen
then make claude test closed, half open and open, and fold mid-task
the fold is the new "works on my machine"
still trusting one good agent run? that proves it can, not that it will
vercel's evals guide:
- pass@k: any of k runs passes. fine when retries are cheap (coding + tests)
- pass^k: all k pass. use when a miss costs the user (orders, payments)
5/5 just means no failures seen yet
codex capacity errors fell from 6.82% of daily users to 0.085%, and every fix marked on this chart is retry timing
if your agent loop hits overloads:
- read retry-after
- back off with jitter
- cap retries, then stop
otherwise your while loop is the capacity error
GitHub says some teams are having trouble merging code this morning
if your merge times out, refresh to see whether it went through before retrying. monday has enough duplicate work
parallel agents don't stall on code. they stall waiting on you
claude code's new projects demo has a "waiting on you" lane. the rate limiter thread stops to ask redis or in-memory. make the boring calls first:
- where state lives
- which deps are ok
- what done means