AI resources
Use Liquidify's Markdown, JSON catalog, and read-only MCP documentation interfaces.
Liquidify documentation is made for agents, too. The same generated component contract is available in human-readable and machine-readable forms. Prefer these routes over scraping rendered HTML.
Text and Markdown
| Route | Contents |
|---|---|
/llms.txt | Short index of the current documentation and important routes |
/llms-full.txt | Consolidated current-version documentation corpus |
/docs/<path>.md | Processed Markdown for an individual documentation page |
Use the per-page Markdown route for focused context and llms-full.txt only when the client can accommodate the complete corpus. Code examples in these outputs follow the site's pnpm-only installation policy.
Structured catalog
| Route | Contents |
|---|---|
/api/catalog.json | Packages, components, maturity, categories, and canonical links |
/api/components/<slug>.json | Generated API, SwiftUI mapping, tokens, and metadata for one component |
The catalog is versioned with the documentation and is generated from public package exports, TypeScript declarations, component specifications, and consumer-safe Storybook metadata.
MCP
The read-only endpoint at /api/mcp exposes bounded documentation operations:
search_docsget_componentget_component_apilist_componentsget_installationmap_swiftui_api
MCP resources use liquidify:// identifiers for documentation pages, components, packages, and versions. The server does not execute consumer code, mutate the repository, publish packages, or expose internal test and fidelity artifacts.
Prompting guidance
Include the target React and Liquidify versions, framework, component name, and whether state should be controlled. Ask the model to cite a catalog or Markdown route when exact props matter. Treat every current component as experimental and pin its package version in generated installation guidance.