What does this skill do, and when should you use it?
This Agent Skill from the openai/codex repository handles questions about OpenAI products and APIs. It prioritizes the OpenAI Developer Docs MCP and uses a Codex manual helper first for broad Codex questions. It also owns model selection, API model migration, and prompt-upgrade guidance, with fallback web searches restricted to official OpenAI domains. The repository is licensed under Apache-2.0, but the source does not document a standalone installation process for this Skill.
Classifies requests and selects the Codex manual helper, OpenAI Developer Docs MCP, or official-domain web search; searches and fetches relevant official documentation; uses an OpenAPI specification tool for API reference, schema, parameter, or required-field questions when available; runs a model resolver to obtain current model, migration-guide, and prompting-guide URLs; and may read or write temporary cache files before producing cited guidance.
- A developer needs to verify how to call an OpenAI API or configure API parameters.
- A Codex user needs guidance on Codex configuration, customization, skills, plugins, MCP, or other product behavior.
- A team is selecting a current model or planning an API model migration.
- An engineer needs prompting guidance for a named GPT-5-family model.
- A reader wants an OpenAI documentation answer supported by official citations.
How do you install this skill?
- Core retrieval depends on the OpenAI Docs MCP and developers.openai.com; restricted mainland-China connectivity or an uninstalled MCP server may make key functionality unavailable.
- The bundled model and prompting references explicitly may drift; current documentation must be checked before presenting current recommendations.
- The script, MCP calls, and migration workflow were not executed in this static review; documented procedures should not be treated as tested behavior.
- The helper writes a temporary cache and accesses remote services; deployment-specific review is still needed for data flow, proxy configuration, and cached content.
- Shell / CLI
- Network access
- Local filesystem
- MCP Server
Node.js 18+
The source only documents installation for the overall Codex CLI collection: on macOS or Linux run curl -fsSL https://chatgpt.com/codex/install.sh | sh; on Windows run powershell -ExecutionPolicy ByPass -c "irm https://chatgpt.com/codex/install.ps1 | iex"; or use npm install -g @openai/codex or brew install --cask codex. It does not document how to install this Skill independently into another client.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/openai/codex.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/codex-rs/skills/src/assets/samples/openai-docs" ~/.claude/skills/
rm -rf "$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:
- Using current official OpenAI documentation, explain how the Responses API supports a multimodal tool-using workflow and provide citations.
In an installed Codex environment, submit a focused request such as: “Using current official OpenAI documentation, explain how the Responses API supports a multimodal tool-using workflow and provide citations.” For Codex questions, ask directly about Codex configuration, skills, or MCP. The source does not specify an independent-client enablement command or complete trigger syntax.
What are this skill's strengths and limitations?
- Explicitly prioritizes official OpenAI documentation and the Codex manual.
- Covers API documentation, Codex self-knowledge, model selection, model migration, and prompt upgrades.
- Can verify API references and OpenAPI specifications when the relevant tool is available.
- Restricts web fallback to official OpenAI domains.
- Depends on the OpenAI Developer Docs MCP, the Codex manual helper, or official network access.
- The model-resolution workflow requires Node.js 18 or newer.
- Temporary caching requires filesystem access.
- No standalone installation procedure, test suite, or cross-client validation evidence is provided.
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 |
|---|---|---|---|---|
| OpenAI Official Docs Assistant this page ✓ OpenAI · Official | 52 · Use with care | ★ 129k | 11d ago | Apache-2.0 |
| OpenAI Developer Docs Assistant ✓ OpenAI · Official | 55 · Use with care | ★ 28k | 4mo ago | — |
| Azure OpenAI Deployment Assistant ✓ Microsoft · Official | 52 · Use with care | ★ 2.7k | 3mo ago | MIT |
| Engram Persistent Memory Protocol | 68 · Recommended | ★ 7.2k | 11d ago | MIT |
| Xberg API Server & MCP Protocol Integration | 52 · Use with care | ★ 9.4k | 11d ago | MIT |
How did FollowSkills review this skill?
The skill restricts retrieval to the official OpenAI Docs MCP and official-domain fallback, and describes an API-key gate plus read-only retrieval boundaries. Deductions apply because MCP data flows, sensitive-credential handling, user confirmation, rollback, and network-dependency risks are not fully specified.
Routing rules, fallbacks, cache validation, timeouts, and error handling are detailed, and the script implements hash verification and atomic writes. Static review cannot execute key paths; MCP availability, Windows/proxy behavior, and remote-response compatibility are not demonstrated by tests, so the score remains below the static ceiling.
The audience and core scenarios are reasonably clear: OpenAI documentation, Codex self-knowledge, model selection, and migration guidance, with some non-fit boundaries. Deductions apply because triggering is complex, Chinese-language output is not specified, and core functionality depends on the developers.openai.com MCP and official network access, which may be difficult from mainland China.
The skill has clear structure, dependency declaration, references, fallback rules, and Apache-2.0 licensing, with substantial progressive disclosure. Deductions apply for the absence of an independent version, changelog, named maintenance owner, and explicit update process; the bundled references also acknowledge drift risk.
For current OpenAI documentation, Codex self-knowledge, and model-migration tasks, the routing and guidance are directly useful and include compatibility and validation checklists. No representative outputs or execution evidence are included, and remote-tool failure may require manual handling, so this is limited to the static-review effectiveness ceiling.
The helper provides HTTP hash checks, cache status, and auditable local manual paths, while tool names and source URLs are specific. Deductions apply because there is no committed key-path test suite, CI execution evidence, or independent corroboration; much of the behavior depends on runtime MCP and network conditions.
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 →