What does this skill do, and when should you use it?
This skill guides Codex in building a durable CLI that can run from any repository rather than a one-off script. It covers runtime selection, command design, authentication, stable JSON output, read-first raw requests, installation, and smoke testing. It fits API and internal-tool workflows that need repeatable discovery, lookup, read, write, or upload operations. The repository README says the repository is deprecated and directs users to the OpenAI Plugins repository for current examples.
Reads API documentation, OpenAPI specifications, SDK documentation, curl examples, web applications, admin tools, or local scripts to inventory resources, authentication, pagination, IDs, upload flows, and write actions; designs discovery, resolve, read, write, doctor, and raw-request commands; selects Rust, TypeScript/Node.js, or Python; defines README guidance, configuration paths, and stable JSON success and error shapes; installs the CLI on PATH and checks it from another directory with help, doctor, build, and test commands.
- A developer needs to turn API documentation into a CLI usable across multiple repositories.
- An operator needs repeatable list, search, download, retry, or upload commands for service workflows.
- A team wants to split existing curl commands, SDK calls, or internal scripts into discovery, read, draft, and write phases.
- A Codex user needs environment-variable or local-config authentication, machine-readable JSON, and a diagnostic command.
How do you install this skill?
- This is a static source review only; no commands, installation flow, or live API calls were executed.
- When using the skill, separately confirm permissions, rollback, network reachability, and credential scope before live writes, PATH installation, config storage, or production-data handling.
- License, version/change history, maintenance ownership, Chinese and mainland-China environment guidance, and key-path test evidence should be added.
- Shell / CLI
- Network access
- Local filesystem
Rust or TypeScript/Node.js or Python
The skill is located at skills/.curated/cli-creator/SKILL.md in the repository. The README says curated skills can be installed in Codex with $skill-installer cli-creator; restart Codex after installation. The supplied material does not document more specific installation options.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/openai/skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/.curated/cli-creator" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Create a CLI named `service-cli` from this OpenAPI specification. Start with project discovery, project-ID resolution, task reads, and `--json doctor`.
Use a request such as: “Create a CLI named service-cli from this OpenAPI specification. Start with project discovery, project-ID resolution, task reads, and --json doctor.” Provide the target name, source, and first real jobs so the skill can design the command surface, select a runtime, and implement safe read and write paths.
What are this skill's strengths and limitations?
- Covers the workflow from source inspection through PATH installation and cross-directory smoke testing.
- Emphasizes stable JSON, explicit discovery and ID-resolution commands, and narrowly scoped writes.
- Includes concrete safeguards for authentication precedence, token exposure, raw requests, and live writes.
- Supports Rust, TypeScript/Node.js, and Python implementation paths.
- Requires source material such as API documentation, scripts, or curl examples; it does not implement a specific service by itself.
- Provides no CLI template, test results, or verified platform matrix.
- The README does not declare a repository license; an individual skill license must be checked in its LICENSE.txt file.
- The repository is deprecated, so current skill and plugin examples should be taken from the OpenAI Plugins repository.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Codex CLI Builder this page ✓ OpenAI · Official | 55 · Use with care | ★ 28k | 4mo ago | — |
| MCP Server Builder Guide | 50 · Use with care | ★ 3.1k | 8mo ago | MIT |
| MCP Server Development Guide ✓ Microsoft · Official | 49 · Use with care | ★ 2.7k | 3mo ago | MIT |
| NeMo Relay Installation Guide ✓ NVIDIA · Official | 62 · Recommended | ★ 3.6k | 11d ago | Apache-2.0 |
| FreeHire Tech Job Search Skill | 63 · Recommended | ★ 47k | 17d ago | MIT |
How did FollowSkills review this skill?
The document explicitly prioritizes environment variables, forbids printing or committing credentials, requires confirmation for live writes, and prefers read-only raw requests, making major security and data-flow risks visible. Six points are deducted because PATH installation, local config writes, and external CLI execution still create side effects; permission isolation, config rollback, data retention, dependency supply-chain checks, and license evidence are incomplete.
The workflow and command contract are internally consistent and specify doctor checks, machine-readable errors, pagination, exit codes, fixtures, and cross-directory smoke tests. Two points are deducted because the skill contains no scripts, test suite, or executed implementation evidence; runtime behavior depends on toolchains, networks, and source systems, so abnormal-path handling is prescribed rather than demonstrated. Static calibration caps this at 10.
The audience, scenarios, source inputs, CLI shape, and boundary against one-off scripts are clearly described, and the trigger matches the main use case of creating composable CLIs. Four points are deducted because exclusions for non-API or high-risk services, environment constraints, Chinese-language interaction, and mainland-China reachability are not specified; core use may depend on overseas APIs without addressing that fit issue.
The SKILL.md is readable and uses a referenced file for progressive disclosure; naming, installation, parameters, examples, limitations, and companion-skill guidance are substantial. Seven points are deducted because license, versioning, changelog, maintenance ownership, and update path are unclear; repository README governance cannot fully substitute for skill-local maintenance documentation.
The skill covers source inventory, command design, implementation, installation, and testing, so it plausibly guides construction of a durable CLI. One point is deducted because no representative outputs or execution evidence are included, leaving correctness, completeness, and marginal benefit over manual alternatives unverified; static calibration caps this at 7.
The supplied files are auditable and the reference contains concrete command shapes, JSON policies, and failure-handling rules. Two points are deducted because there is no CI, committed test suite, test result, or third-party execution evidence for independent reproduction; static calibration caps this at 5.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →