Topic
162 articles · Page 1 of 5
A phone video was too big to publish. Rather than uploading it somewhere, Willison asked Claude Fable 5.1 for a browser tool that runs FFmpeg via WebAssembly — five presets, and the command line for each one on screen.
A principal product manager at Spotify built a Claude Code plugin that delegates large file reads and boilerplate code to a cheap model. The idea: let the frontier model think, not read. The post has been sitting near the top of Hacker News since Friday night.
On Fable 5.1's launch day, Claude Code follows right along: 2.1.257 makes the new model the default, finally brings configurable time formats and a new auto-mode rule. 2.1.258 clears up a nasty startup failure on macOS Monterey.
Anthropic is permanently raising Claude Code's weekly limits by 25%. Sounds great — but compared to right now, it's a step back. The first post got deleted, then came the honest version.
What you tell Claude in chat, Cowork now knows too - and the other way around. Anthropic merged the two memories, made every entry editable, and lets you choose whether sensitive topics get saved at all.
Anthropic has finished the rollout: Cowork now runs on iPhone, iPad, and the web for every paid plan, not just the priciest tier. The agent that started on the Mac is finally everywhere.
Claude in Slack used to see only the one message right in front of it. Now it reads the whole channel, uses its memory, and jumps in when it has something to add – or stays quiet. At no extra cost.
Starting today, new sessions on Pro, Max, and Team run in auto mode. Instead of asking at every step, a classifier checks each action – and only blocks what turns dangerous.
An independent lab hit Claude Code in auto mode with 720 prompt-injection attacks. Not one got through. For Codex, the scoreboard looked different.
The community keeps reporting that Opus 5 over-engineers tasks — planning and building more than you asked for. Oddly enough, a lower reasoning effort seems to help.
Anthropic is making Auto Mode the default for Pro, Max, and Team on August 14. A classifier checks every tool call - and according to Anthropic's own study, it catches far more dangerous commands than we humans do.
Anthropic rolled out security scanning for skills and plugins. On upload or edit, Claude automatically checks third-party extensions — with three possible results: pass, warn, or fail. Enterprise only, for now.
The stateless rework of the Model Context Protocol has reignited Simon Willison's interest in MCP. He built three tools in a week — and along the way makes the best case for why MCP is often the safer choice over an agent with shell access.
The Model Context Protocol just got its biggest update yet. The new spec release makes MCP stateless, folds Apps and Tasks into an extensions framework, and brings auth up to enterprise grade — and it's rolling out across Claude's products now.
Simon Willison published a compact guide on connecting your own MCP server to both Claude and ChatGPT. The short version: on Claude it's easy, on ChatGPT you have to take a few detours.
Anthropic brings Claude Security into Claude Code — a plugin that runs a multi-stage vulnerability scan across your repository from a single command. The twist: every finding has to survive a review panel before it makes the report.
Simon Willison sat down with Cat Wu and Thariq from the Claude Code team — the video is now online. The highlights: '@Claude' writes 65 percent of the team's PRs, features go to Anthropic staff first, and on prompting, more context beats a pile of prohibitions.
New feature in Claude desktop: record your screen, talk through what you're doing — and Claude turns it into a reusable skill. 'Record a skill' is rolling out to Pro, Max, and Team.
Next week the Model Context Protocol gets an update almost nobody will notice — and it still changes a lot. MCP is dropping its clunky session handling, which finally makes it fit for large-scale deployments.
Simon Willison proved it on his own install: since version 2.1.181, Claude Code uses the Rust port of Bun. Startup got 10% faster — and almost nobody noticed. That's exactly the point.