CLI Tools Directory
See how AI agents use these tools, organized by software platform
Frequently asked questions
Can I just pick tools by GitHub stars?
We've all done it. High stars usually mean active maintenance, good docs, and a community that'll answer your Stack Overflow questions at 2am. But stars don't tell you if a tool's output is structured enough for AI agents, or if it'll burn 50K tokens on a help menu. That's why we add editorial "when to use" and "when to skip" sections.
What CLI tools work with Claude Code and Cursor?
Every tool in this directory works with Claude Code, Cursor, Gemini CLI, and Codex CLI out of the box. If your agent can run a shell command, it can use these tools — no plugins, no configuration, no MCP server setup. Popular starting points: GitHub CLI (gh) for PRs and issues, Vercel CLI for deployments, Supabase CLI for database management.
Should I use CLI tools or MCP servers?
CLI tools. They use 9–32× fewer tokens and have near-perfect reliability. MCP servers inject thousands of tokens of schema into every conversation — even when your agent only needs one command. Use MCP when you need discovery or streaming. Use CLI for everything else. We break down the full data in our CLI vs MCP comparison.
How do I add a CLI tool to my AI coding agent?
Install it, authenticate it, done. Run brew install gh, then gh auth login, and Claude Code can immediately run gh pr create or gh issue list. No SDK, no API keys in your prompt, no MCP configuration file. Every tool page in this directory shows the exact install command for macOS, Linux, and npm.
What's the difference between a CLI tool and an MCP server?
A CLI tool is a standalone program you run in your terminal — docker ps, gh pr list, stripe listen. An MCP server is a wrapper that exposes tools through the Model Context Protocol, letting AI agents discover and call them via structured schemas. CLI is simpler and cheaper. MCP is richer and more discoverable. Most platforms now ship both.