mirror of
https://github.com/magnus919/agent-skills.git
synced 2026-09-15 13:36:35 +03:00
Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
CLI Builder — Design Agent-Friendly CLI Tools
A comprehensive design guide and scaffold for building CLI tools that AI agents can actually use. 10 universal design patterns grounded in real failures from building 15+ agent-facing CLIs.
Why Install This Skill
When your agent loads this skill, it can design, build, and refactor CLI tools that agents can discover and use without human help. That means:
- Every
--helpoutput becomes a contract the agent parses to understand your tool - Design
--jsonas a documented contract for machine-readable output agents can consume - Establish or review idempotency so safe reruns are explicit rather than assumed
- Authentication is lazy — help and dry-run work without credentials
- Errors are structured — different exit codes for different failure modes
What You Get
| Directory | Purpose |
|---|---|
SKILL.md |
Concise workflow for discovery, predictable contracts, verification, and maintenance |
templates/ |
Bash CLI scaffold for local wrappers |
references/ |
Python API clients, advanced patterns, readiness checklist, wrapper example, MCP decisions, and improvement cycle |
Triggers
Load this when building a new CLI tool, refactoring an existing tool that causes agent friction, or debugging why your agent keeps failing to use a CLI properly.
Requirements
Bash, Python 3.8+, jq, and a standard Unix CLI environment.
Quick Start
Start with the setup and first workflow in SKILL.md, then use the linked resources for the specific task you need to complete.