Docs

[burkh4rt]

This pull request introduces significant improvements to documentation and developer experience, focusing on adding a robust documentation build system, enhancing the README.md for clarity and accuracy, and providing detailed API documentation for key pipeline components. The changes also include custom styling for documentation and improved VSCode settings for CSS formatting.

Documentation infrastructure and content:

• Added a .readthedocs.yaml configuration to enable automated documentation builds with Read the Docs, specifying Python 3.12 and integration with mkdocs.
• Created new API documentation pages for the CLI (docs/api/cli.md), Collator (docs/api/collator.md), Tokenizer (docs/api/tokenizer.md), and Winnower (docs/api/winnower.md), providing detailed explanations of each pipeline stage and their outputs. [1] [2] [3] [4]
• Added recipe documentation pages and included the README.md content directly in the documentation index for a unified docs experience. [1] [2] [3] [4] [5]

README and usage improvements:

• Updated the README.md to clarify installation instructions, improve command-line option descriptions, fix and clarify links to configuration files and recipes, and update file exclusion patterns for data syncing. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]

Developer experience and styling:

• Added a custom extra.css stylesheet for documentation, including self-hosted Gotham fonts and a custom color scheme for Material for MkDocs.
• Updated VSCode settings to use Prettier as the default CSS formatter for a consistent code style.

[Copilot]

Pull request overview

This PR adds a MkDocs-based documentation site (built on Read the Docs) and updates existing user-facing docs to improve clarity around installation, configuration, and pipeline usage.

Changes:

• Introduces MkDocs + Material configuration, RTD build config, and custom docs styling/assets.
• Adds API reference pages for key pipeline components and includes recipe/README content into the docs site.
• Updates CLI/README/recipes to align option help text and usage examples with current behavior.

Reviewed changes

Copilot reviewed 18 out of 24 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
src/cocoa/cli.py	Updates package version lookup and clarifies CLI help text for tokenizer reuse and verbosity.
recipes/date-based-generative-inference.md	Refreshes the recipe instructions and command example for running the pipeline with config overrides.
recipes/configure-new-token.md	Clarifies how to apply collation config changes and how to pass the config via CLI.
README.md	Improves installation instructions and updates documentation links and option descriptions.
pyproject.toml	Adds documentation-related optional dependencies and expands project URLs.
mkdocs.yml	Adds MkDocs site configuration, navigation, and plugin setup for API/recipe content.
docs/svg/emoji_u2615.svg	Adds a documentation logo/favicon asset for the MkDocs theme.
docs/stylesheets/extra.css	Adds custom styling and self-hosted font configuration for the docs theme.
docs/recipes/tokenizer-transfer.md	Adds a docs wrapper page that includes the existing recipe content.
docs/recipes/index.md	Adds a docs landing page that includes the recipes README content.
docs/recipes/date-based-generative-inference.md	Adds a docs wrapper page that includes the existing recipe content.
docs/recipes/configure-new-token.md	Adds a docs wrapper page that includes the existing recipe content.
docs/index.md	Includes the repository README directly into the docs home page.
docs/api/winnower.md	Adds API documentation for the Winnower stage and its outputs/behavior.
docs/api/tokenizer.md	Adds API documentation for the Tokenizer stage and tokenizer reuse semantics.
docs/api/collator.md	Adds API documentation for the Collator stage and config/security notes.
docs/api/cli.md	Adds CLI documentation describing commands and typical usage patterns.
.vscode/settings.json	Sets Prettier as the default formatter for CSS files in VSCode.
.readthedocs.yaml	Configures Read the Docs to build the MkDocs site using Python 3.12.
.gitignore	Updates ignored local/build artifacts (docs build output, dist/build, etc.).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.