DigestAI news desk

Cut through the AI noise.

Agents & Tools8 min read

OpenAI Codex and Anthropic Claude Code enable computer-use AI; MCP is a standard

The article explains “computer-use,” a feature that lets an AI click, type and read a screen, and compares how OpenAI’s Codex, Anthropic’s Claude Code, and the MCP standard implement it. Official materials up to September 22 2026 confirm that Codex supports computer-use on macOS and Windows desktop apps, useful for GUI‑only bugs or visual checks.

1 source

Key points

  • OpenAI's Codex offers computer-use for macOS and Windows desktop apps, allowing AI to click and type while observing the screen.
  • Anthropic's Claude Code includes computer-use as a research preview for macOS Pro/Max, accessible via CLI /mcp and desktop integration.
  • MCP is a standard for connecting AI tools via APIs, not a screen‑operating feature; Claude Code's computer-use runs as a built‑in MCP server.

Claude Code also offers computer-use, but only as a research preview for macOS Pro/Max. It can be activated from the CLI via the /mcp flag or through a desktop integration, and it includes a separate Chrome‑integration mode for web‑based screens. MCP itself is not a screen‑operating tool; it is a common API‑based way to connect AI apps to external services, and Claude Code’s computer-use runs on a built‑in MCP server.

The piece advises splitting work between dedicated integrations (via MCP or APIs) for data retrieval and computer-use for visual verification, recommending small pilot tasks and clear completion conditions to evaluate reliability and cost.

The story so far

3 episodes →
  1. OpenAI Codex and Anthropic Claude Code enable computer-use AI; MCP is a standardthis story
Full story from note.com · by Yusuke · via Search: ClaudeOpen source ↗

What is "Computer Use" where AI operates a PC? Comparing Codex, Claude Code, and MCP

note.com · 21 September 2026

Loading the full article…

This text was published by note.com and written by Yusuke. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Topics · follow one to build your own front page

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Agents & Tools

All →

Related stories