In Countersign, a cloud-cost agent, the model has no write path: it drafts an evidence-backed change request, people countersign, code it cannot reach executes only what they signed, and the evaluation tests its graders before it trusts a score.
Squall Line times each truck route with the hours-of-service rules, reads each Open-Meteo field the way it is defined, at every checkpoint's arrival, ranks the routes with a one-line reason and paints a 48-hour corridor heatmap that blends before it classifies.
A Claude Code plugin can be one text file or a set of seven hooks that sends your diff to Opus, and neither the install count nor the word "official" tells you which.
New York to Los Angeles: why the classic greedy for the gas station problem breaks on real roads, the dynamic program over fuel levels that replaces it, and how to place 6,600 truck stops that have no coordinates.
Every interactive Claude Code session on my Mac now starts on Remote Control, carries its iTerm2 tab's name to my phone, and is archived there when it ends on the Mac. What stock Claude Code already does, the three gaps I closed, and how I verified it.
Anthropic renamed Max 20x to Ultra and is lifting its five-hour limit account by account, after an automatic check. What changed, how to check your account, why the five-hour bar may still be there, what a week-only budget means for agents, and the one rule my status line now follows.
This publication passed 100 members. Most of October's 78 new members likely came through a channel that analytics barely sees. What brought them, what Ghost's growth guide gets right and misses, a checklist any Ghost publisher can run this week, and a measured plan for the next 900.
Claude for Legal now offers ready-made tools, not just a blank chat box: review tables, cite checks, matter boards and intake desks, with Harvey, Goodwin, Foley & Lardner and PwC building inside it. What shipped, what the docs add, and who should worry.
Claude Code, extended instead of wrapped in an app: one command for every project and account, work that moves to another account at a limit, subagents on a benchmarked model table that can retune itself, and a harness that reads its tools' news and repairs itself.
The local agent passed a controlled test and failed real work: tasks Opus 5.5 finishes in 10 to 30 minutes took it hours. It's gone, along with every agent that isn't Claude and about 452 GiB of disk.
And Fuck Sam Altman Too
My harness stopped Claude Code and Codex at 95% of their plan windows, even when usage credits could have paid for more. Now it checks for usable credits first, keeps working until they run low, and says so in the status bar.