30 stories
Six extension mechanisms, and only three of them are different kinds of thing. What each one is, where it lives, when the model reaches for it, and the honest edge cases where two genuinely overlap — plus the two changes that broke everyone's mental model: custom commands are now skills, and MCP…
Everyone writes "put a SKILL.md in ~/.claude/skills". The real picture is two ecosystems with two installers that cannot see each other: marketplace plugins via claude plugin install, plain-git skills via npx skills.
Claude Code as a library — and the field almost everyone misreads. We ran the same agent twice: allowedTools is an auto-approve list, not a sandbox, and the first run called Bash despite Bash not being on it.
Precedence is not the same direction for every mechanism: settings.json lets the repo override your home directory, skills let your home directory override the repo, and CLAUDE.md files never override each other at all.
A complete stdio server in TypeScript, small enough to paste and run, then the parts that decide whether the model ever calls it: the tool description is prompt engineering rather than documentation, and an error should tell the model how to recover instead of just failing.
Three agents that run on a machine that is not yours and hand back a pull request. How a task gets in, what each sandbox can reach, where secrets end up, how the result comes back — and the under-documented part: Jules ships a real REST API, Codex Cloud has only a CLI, and Claude Code on the web…
A skill is not data — it is instructions that steer your agent inside your repository, with your credentials. One realistic attack chain traced end to end, why the permission system rather than the skill file is the trust boundary, why commit-pinning fails outright for most repos through the Skills…
Install commands for macOS, Windows and npm, the OAuth vs API-key sign-in, the real Node floor, the telemetry-on-by-default caveat, and what a K3 session costs on a membership vs the API.
OpenAI's GPT-5.6 family split three ways. What Sol, Terra and Luna actually cost once you count the 272K cliff and the new 1.25x cache write — and which one belongs in your Codex sessions.
Moonshot released Kimi K3's weights on July 27, 2026. What that actually gets you at 2.8 trillion parameters — the real download size, the hardware floor, the bespoke license, and how to choose between self-hosting, the official API, and an aggregator.
A model: key in a subagent's YAML frontmatter picks the model for that leg of the work. Haiku workers plus one Opus judge take an eight-way fan-out from $3.00 to $0.80 — and unlike /model mid-session, it leaves the parent's prompt cache intact.
Claude Opus 5 landed July 24 at $5/$25 per million tokens — the same as Opus 4.8 and exactly half of Fable 5. We break down the pricing table, the five-level effort ladder the press got wrong, Fast mode's 2.5x-for-2x tradeoff, and the ARC-verified benchmark jump.
Showing 30 of 30 stories
You're all caught up.
Get the top AI news and tools straight to your inbox every week.
Weekly · AI news
Join 15,000+ AI enthusiasts
