"Anthropic"
8 results found
How Anthropic reviews code that Claude writes 90% of itself
Anthropic says Claude now writes 90% of Claude Code, with a 7-layer AI review process that cut reverts by 40% while doubling merged PRs.
Anthropic gives Max and Team subscribers monthly API credits
Anthropic now gives Max and Team subscribers $20 to $200 in monthly API credits, though unused credits expire and only one org can be linked.
Claude Opus 5.5 animates entire scenes from JS or Python code
Anthropic released Claude Opus 5.5, which writes full animations frame by frame in JavaScript or Python, no premade assets.
Claude Opus 5 built its own Starfield in 24 hours — and criticized itself along the way
A developer ran Claude Opus 5 through a 24-hour marathon session to build a playable space exploration game from scratch, with a separate adversarial agent grading it against AAA titles like Starfield.
Anthropic publishes an official prompting guide for Claude Opus 5
Anthropic's new prompting guide for Claude Opus 5 confirms the top tip — give it the full task upfront — but also flags less obvious habits from older models that now waste tokens.
Claude Cowork has finally launched on mobile and web
Claude Cowork is no longer tied to the desktop app — background tasks now keep running in the cloud even when your laptop is closed, with approvals routed straight to your phone.
Anthropic researchers found a "hidden thought" space inside Claude — J-space, a workspace the model wasn't designed to have, but developed on its own
Anthropic discovered J-space, an unengineered internal workspace inside Claude that functions as short-term memory for reasoning — and one researchers can read and modify directly.
Anthropic publishes a 52-prompt library directly inside the Claude Code documentation
Anthropic has added a copy-paste prompt library to the Claude Code docs, covering the full development cycle with 52 prompts and "Why this works" explanations for each.