When building a command line tool for AI agents, the way you present information matters a lot. The Agent eXperience Interface (AXI) provides clear rules for creating efficient and agent-friendly CLI tools. These rules help save tokens and avoid confusion.
AXI focuses on token-efficient output using a format called TOON. It also sets limits on how many fields to show by default. This way agents get what they need without wasting resources.
The rules include structured errors, definitive empty states, and pre-computed summaries. All of these make it easier for agents to understand and act on the output. Developers who build CLI tools for agents will find AXI a practical guide.
Add Axi skill to your workflow
Global
mkdir -p ~/.claude/skills/axiProject
mkdir -p .claude/skills/axiSource Repository
More CLI & Terminal Claude Code Skills
Lark Sharedlarksuite/cli
Manage Feishu resources from terminal with user and bot authentication and permission handling
Runcomfy Cliagentspace-so/runcomfy-agent-skills
One binary one login runs hundreds of AI models from your command line
Tzstxixu-me/skills
Create extract and manage tzst archives with safe defaults and scripting support
Xdropxixu-me/skills
Upload and download files securely from your terminal with Xdrop CLI
Runcomfy Clidoany-ai/skills
One binary, one auth, hundreds of AI models from your terminal
Runcomfy Cliruncomfy-com/skills
Control hundreds of AI models from your command line with one simple tool
Hyperframes Cliheygen-com/hyperframes
Scaffold, lint, inspect, preview, and render video compositions from your terminal
Sentry Clisentry/dev
Manage your Sentry projects and issues directly from the terminal