AI Coding Tool Cheat Sheets

Start here for fast, source-backed guides to the AI coding tools developers evaluate most often. The five flagship pages now open with a 60-second cheat sheet before the deeper setup, permissions, security, and team-adoption guidance.

Flagship Cheat Sheets

Tool Best starting point when... Cheat sheet
Claude Code you want terminal-first deep repo work and granular permissions Claude Code
OpenAI Codex you want terminal work, local review, scripting, and AGENTS.md OpenAI Codex
Cursor you want an AI-native editor plus terminal agent Cursor
GitHub Copilot your team is GitHub-centric across IDE, CLI, review, and agents GitHub Copilot
Gemini CLI you want a Google-backed terminal agent with sandbox and context controls Gemini CLI

Not sure which one fits? Start with the AI Coding Tools Comparison.

What Every Flagship Page Gives You

  • a 60-second command and configuration reference
  • five copy-ready prompts for common developer work
  • current installation and session commands
  • repository instruction-file guidance
  • permissions and safer defaults
  • MCP / external-tool guidance
  • a team adoption checklist
  • verification status and official sources

All Tool Guides

Terminal and CLI First

Editor and IDE First

Model / Integration Coverage

How To Use a Tool Page

  1. Read the 60-second cheat sheet if the page has one.
  2. Check Status, Last verified, and Verification scope before copying commands.
  3. Compare the repository instruction and permission model with your current engineering standards.
  4. Start with a read-only explanation or review task.
  5. Test interactive editing before headless or broad automation.
  6. Add MCP, plugins, skills, or external systems only after the base workflow is trusted.

Verification Labels

  • Documentation verified means the page was reviewed against official documentation, but the workflow was not fully exercised locally in this repository.
  • Partially verified means some parts were tested or validated more directly than others.
  • Needs verification means the page is directionally useful but should not be treated as approval-ready without re-checking current official docs.
  • Experimental or Platform-specific means rollout assumptions should stay narrow.

A Fair Tool Evaluation

Use the same pilot for each candidate:

  1. explain an unfamiliar repository without edits
  2. review a real diff
  3. plan a small bug fix
  4. implement the fix interactively
  5. run project validation
  6. only then evaluate MCP or non-interactive automation

Record the result with the Tool Evaluation Template.

Verification note: the flagship Claude Code, Codex, Cursor, GitHub Copilot, and Gemini CLI guides were re-checked against official sources on 2026-08-09. Other tool pages retain their own verification dates and should be checked individually.