OpenAI Skills
OpenAI Skills after deprecation: what the repository still teaches
Understand the deprecated openai/skills catalog, its reusable workflow structure, per-skill licenses and the boundary between studying examples and adopting a current distribution path.
What you will learn
- Start with the deprecation notice
- Read one complete workflow, not only its title
- Choose an adoption boundary you can maintain
Before you start
- Basic Git, Python and command-line concepts
- An explicit boundary for permitted repository inspection and changes
Explain the chapter’s source behavior and apply its acceptance checklist without confusing a catalog with its host.
Key takeaways
- Treat the catalog as deprecated source evidence, not a current product guarantee.
- Separate workflow text, presentation metadata, helpers and host capabilities.
- Check the selected skill’s license and revision before adoption.
Start with the deprecation notice
The pinned openai/skills README begins with a deprecation notice and directs readers to the OpenAI Plugins repository for current examples. This changes the adoption decision: the repository remains useful implementation evidence, but its Trending position does not make its installation examples the preferred starting point for a new rollout. This series studies commit 49f948f, not an imagined future release.
A skill here is a directory combining instructions with optional scripts and resources. It is not another model, a hosted application or a universal permission grant. The catalog supplies workflow materials; the host must discover them, choose when to load them and provide tools. An installed directory alone cannot prove that a particular host has activated the workflow.
Read one complete workflow, not only its title
The gh-fix-ci example gives the abstraction a concrete purpose. Its instructions describe checking GitHub authentication, resolving a pull request, reading failing GitHub Actions checks, explaining the evidence and requesting approval before implementation. The adjacent Python helper retrieves and summarizes diagnostics; it does not implement an arbitrary repair simply because a check is red.
There are already three distinct responsibilities in that small folder: SKILL.md defines the workflow, agents/openai.yaml supplies presentation metadata, and scripts/inspect_pr_checks.py performs deterministic operations. A connector or command-line tool supplies access to the remote service. Mixing these responsibilities makes it difficult to explain whether a failure comes from instructions, credentials, tool behavior or the underlying project.
Choose an adoption boundary you can maintain
The README explicitly places licensing at the individual skill directory. The inspected installer has an Apache-2.0 license file; that observation is not a blanket license conclusion for every skill, external dependency or generated artifact. Record the relevant license alongside the exact revision and review what the selected folder actually includes before redistributing it.
For a new shared workflow, evaluate the current distribution guidance linked by the repository. For an existing installation, first inventory the old folder, its revision, its dependencies and the tasks that depend on it. This review executed isolated source probes, not a live Codex workflow or migration. Compatibility, success rate and production suitability remain questions to test in the intended host.
Implementation steps
- 1
Read the pinned README and identify the selected skill directory.
- 2
Trace its instructions, metadata and helper entry points.
- 3
Record the host, dependencies, revision and allowed actions.
- 4
Choose a current rollout path and a small acceptance task.
Copy-ready example
{"repository":"openai/skills","revision":"49f948faa9258a0c61caceaf225e179651397431","repositoryDeprecated":true,"selectedExample":"gh-fix-ci","liveHostRun":false,"licenseScope":"per skill"}Frequently asked questions
Does deprecation make every example unusable?
No. It changes how the examples should be adopted and maintained. This series analyzes a fixed snapshot and does not promise ongoing compatibility.
Is this repository a model or a service I can deploy?
No. It is a catalog of workflow directories and supporting code. A host and the required tools still execute the work.
Sources
- README.mdSource checked 2026-09-08
- skills/.system/skill-installer/SKILL.mdSource checked 2026-09-08
- skills/.system/skill-installer/LICENSE.txtSource checked 2026-09-08
- skills/.system/skill-installer/scripts/install-skill-from-github.pySource checked 2026-09-08
- skills/.system/skill-installer/scripts/github_utils.pySource checked 2026-09-08
- skills/.system/skill-installer/scripts/list-skills.pySource checked 2026-09-08
- skills/.system/skill-creator/scripts/quick_validate.pySource checked 2026-09-08
- skills/.system/skill-creator/references/openai_yaml.mdSource checked 2026-09-08
- skills/.curated/gh-fix-ci/SKILL.mdSource checked 2026-09-08
- skills/.curated/gh-fix-ci/scripts/inspect_pr_checks.pySource checked 2026-09-08
- skills/.curated/gh-fix-ci/agents/openai.yamlSource checked 2026-09-08
- skills/.system/plugin-creator/SKILL.mdSource checked 2026-09-08
- Current official skills guideSource checked 2026-09-08