Builds/cli-devops

Write CLI and scripts with this set

The model comes from Terminal-Bench 4.0. The rules go in AGENTS.md: a repeatable command first, then the script.

Edited 2026-10-10 · Model basis Terminal-Bench 4.0 2026-10-06

RolePrimaryAlternative · when to switch

ModelClaude Opus 5.5Terminal-Bench #1 · 64.8Value：GPT-6.1 Sol（#3 · 58.2，about an eighth of the run cost）RulesAGENTS.mdPrompt · MITTurn it into a skill when the same terminal practice should travel to other repos

Take this set

AGENTS.md

# CLI
- Write a repeatable command or a failing test before changing the script
- Stop and confirm before a destructive command
- After the change, run the check command the repo already documents

Install and use

# Repo rules
Put this page's AGENTS.md at the repo root

Not for

The interface is the product　Page quality is LMArena WebDev, not a terminal pass rate.

Android　Kotlin / Compose uses Android Bench, not this terminal board.

Only the top pass rate　Batch scripts should read cost. The value tier is the alternative on the model row.

Evidence

Scenario board · Write CLI, scripts, and DevOps Prompt · AGENTS.md