Choosing the Right Tool — A Framework

~14 min · decision-framework, cross-tool, tool-selection

Level 0🌱 Novice

0 XP0/70 lessons0/11 achievements

0/120 XP to next level120 XP to go0% complete

By job, not by team

The right CLI is the one that fits the job — not the one your friend uses or the one with the loudest blog post. A simple framework: (1) what's the workload (code editing? research synthesis? remote tasks? web automation?), (2) what's the trust profile (your repo or someone else's? sensitive code or experimental?), (3) what's the cost shape (one-off, recurring, parallel fan-out?).

Practical defaults: code editing in your terminal → Claude Code (deepest IDE-like flow), Codex (open-source, multi-provider auth), or Gemini (1M context, free tier). Knowledge work, file-native research → Cowork. Mobile-to-Mac async → Dispatch. Long sessions across machines → Claude Remote. Web tasks, booking/forms → Operator. Quick async questions → Claude.ai or ChatGPT Desktop.

Code

The decision tree·text

1. What kind of work?
   • Code editing                        → Step 2
   • Knowledge work / research / docs    → Cowork
   • Web automation (booking, etc.)      → Operator
   • Quick async questions               → claude.ai / ChatGPT Desktop

2. Code editing — which CLI?
   • Need 1M context / free tier         → Gemini
   • Open source, multi-provider auth    → Codex
   • Deepest skill / hook ecosystem      → Claude Code
   • Already using one — stay there      → Whichever
     (the differences are smaller than the picking-time)

3. Need across-machine?
   • Move sessions cleanly between Macs/cloud → Claude Remote
   • Send tasks from phone to home Mac        → Dispatch
   • Run on a remote with full context        → SSH + your CLI on remote

Common combinations that work·text

Productive setups in 2026:

A) Solo dev, free-tier-first
   Gemini CLI (free tier) for daily code
   Cowork for research / docs
   ChatGPT for quick questions

B) Pro subscriber, Anthropic-heavy
   Claude Code for daily code
   Cowork for research / docs
   Dispatch for mobile-trigger tasks
   Claude Remote for cross-machine

C) Multi-tool team
   AGENTS.md as canonical project memory
   MCP servers for org capabilities
   Skills repo with cross-CLI Skills
   Each dev picks their preferred CLI

External links

Exercise

List your last 10 AI-CLI-shaped tasks. For each, walk the decision tree. Note where you used the wrong tool by reflex and what would have been better. Adjust your defaults accordingly.

Progress

Progress is local-only — sign in to sync across devices.

← PreviousSkills & Custom Commands as Cross-CLI Capability Next →Quiz · 3 questions

Spotted a bug or have feedback on this page?Report an Issue

Comments 0

🔔 Reply notifications (sign in)

No comments yet — be the first.