# AGI Labo (AGIラボ)

> AGI Labo is a Japanese AI media, community, and product studio. It publishes in-depth articles on AI and technology, runs a creator/developer community, and builds AI-powered applications including AGI Cockpit, plus agent-run experiments like AGI Atlas.

## Projects

- [AGI Cockpit](https://agi-labo.com/tools/cockpit): Work OS for AI agents. Run Claude Code, Codex, Antigravity, Cursor, and Grok Build in parallel on one screen, with OpenRouter and LM Studio model support. Master agent splits work into subtasks, Autorun schedules recurring jobs, and a mobile PWA lets you steer tasks from your phone. Free to use on Windows, macOS, and Linux.
- [AGI Cockpit (English)](https://agi-labo.com/en/tools/cockpit): English landing page.
- [AGI Cockpit official documentation](https://agi-labo.com/tools/cockpit/docs): Japanese documentation.
- [AGI Cockpit official documentation (English)](https://agi-labo.com/en/tools/cockpit/docs): English documentation.
- [AGI Cockpit documentation index for AI](https://agi-labo.com/tools/cockpit/docs/llms.txt): Japanese documentation index with Markdown endpoints.
- [AGI Cockpit documentation index for AI (English)](https://agi-labo.com/en/tools/cockpit/docs/llms.txt): English documentation index with Markdown endpoints.
- [AGI Cockpit release notes](https://agi-labo.com/tools/cockpit/docs/releases): Version history and changelog.
- [AGI Atlas](https://atlas.agi-labo.com): An unmanned library of world classics, retranslated into modern Japanese by a team of AI agents. Every edition publishes its source text, rights, and full provenance. Vertical-writing EPUB downloads for members. Free to read (Japanese).
- [AGI Atlas — how the agent-run library works](https://atlas.agi-labo.com/about): Pipeline, provenance system, and credits (Japanese).
- [Projects index](https://agi-labo.com/tools): All AGI Labo projects and applications.
- [Bench](https://agi-labo.com/bench): Open benchmarks for LLM agents, measured in fully blind matches and scored mechanically (Japanese; English at /en/bench).
- [KY-Bench](https://agi-labo.com/bench/ky-bench): Can LLMs read the room? Rankings from Same Scale, a cooperative game measuring how legibly agents signal and how accurately they read each other. Records at https://github.com/tempi-tech/ky-bench.
- [Reversi Bench](https://agi-labo.com/bench/reversi): Does more thinking make a stronger player? Blind Reversi matches between the same model at different reasoning efforts, with thinking cost published alongside results. Records at https://github.com/tempi-tech/reversi-bench.

## Articles & News

- [Practical Guides](https://agi-labo.com/guides): Fully open, hands-on guides for using AI tools and AI agents in real workflows (Japanese).
- [Practical Guides (English)](https://agi-labo.com/en/guides): Fully open, hands-on guides for using AI tools and AI agents in real workflows.
- [Articles](https://agi-labo.com/articles): In-depth articles on AI agents, Claude Code, and practical AI workflows (Japanese).
- [Daily AI News](https://agi-labo.com/daily): Daily AI newspaper (Japanese).

## Community & Courses

- [Courses](https://agi-labo.com/courses): AI courses and hands-on training (Japanese).
- [Community](https://agi-labo.com/community): Creator and developer community.

## Optional

- [Sitemap](https://agi-labo.com/sitemap.xml)
- [Help](https://agi-labo.com/help)