openai-for-ai

A CLI for compiling the OpenAI OpenAPI specification into deterministic, LLM-friendly HTML blocks and discovery manifests.

Usage

pdm install
pdm run openai-for-ai build --spec-url https://app.stainless.com/api/spec/documented/openai/openapi.documented.yml --out site/

This fetches the latest spec (with ETag-based caching), renders operation and component pages beneath site/, and generates helper indexes such as llms.txt, manifest.json, and blocks/index.json.

Output paths are derived from tag and schema names but are sanitized into safe path segments.

Use pdm run openai-for-ai build --help to see all available options.

CLI Examples

pdm run openai-for-ai build --spec-url https://app.stainless.com/api/spec/documented/openai/openapi.documented.yml --out site/

• --spec-url fetches the published OpenAI OpenAPI document and respects the built-in ETag cache to avoid redundant downloads.
• --out site/ writes the rendered HTML fragments and indexes into the site/ directory that ships with the repo.

pdm run openai-for-ai build

• With no flags, the builder uses the default remote spec, writes into site/, and includes the default language examples (curl, python).

pdm run openai-for-ai build --spec-url https://app.stainless.com/api/spec/documented/openai/openapi.documented.yml --lang javascript --out site/

• --lang javascript renders blocks optimized for JavaScript code samples.
• Combining --lang with an explicit --out lets you produce language-specific previews without overwriting other assets.

Tests

pdm run pytest

The integration tests exercise the full build pipeline against the real OpenAI OpenAPI specification bundled under tests/data/openai.documented.yml. Ensure optional dependencies (e.g., Jinja2) are installed so these tests execute instead of being skipped.

License

This project is licensed under the MIT License.