Skills and plugins, not just models, are reshaping AI coding tools
An open-source economy of reusable skills and plugins is growing around Claude Code, Codex, Cursor and Copilot, turning general models into specialized workers.
An open-source economy of reusable skills and plugins is growing around Claude Code, Codex, Cursor and Copilot, turning general models into specialized workers.
Anysphere's Cursor 3 adds an Agents Window for running several AI coding agents at once, a point-and-click Design Mode for front-end work, and handoff between cloud and local environments.
Google has expanded AI Studio from a prototyping tool into a full-stack builder that generates backends, databases and sign-in flows from a prompt, as Firebase Studio winds down.
Anthropic's new Code Review feature runs several AI agents in parallel over a pull request, scores findings by confidence and surfaces only serious issues; it is in preview for Team and Enterprise.
OpenAI is developing its own code repository platform that would compete with GitHub, The Information reported, as tensions with Microsoft persist and Codex passes a million developers.
The February 20 update lets developers run and preview apps, have diffs reviewed automatically and monitor pull requests to merge, as Claude Code passes a $2.5 billion revenue run rate.
Anthropic revised its terms to make clear that Pro and Max subscription tokens may be used only in Claude Code and Claude.ai, shutting out independent agent harnesses.
OpenAI's new Codex app gives developers a single place to run and supervise several AI coding agents in parallel, with worktrees, scheduled automations and built-in code review.