import { Card, Cards } from 'fumadocs-ui/components/card';

The Parse SDK is the low-level layer of datamark. It turns Markdown strings into fully typed syntax trees and gives you the tools to query, transform, and extract data from them.

Use it directly when you need fine-grained control over parsing, or as the foundation for higher-level formats built with the [Format SDK](/docs/template).

<Cards>
  <Card title="parse()" description="Turn Markdown into a typed Document with a section tree" href="/docs/parse/parse" />

  <Card title="Frontmatter" description="Extract and split YAML frontmatter from Markdown strings" href="/docs/parse/frontmatter" />

  <Card title="YAML" description="Parse and stringify YAML with thin built-in helpers" href="/docs/parse/yaml" />

  <Card title="Tree Utilities" description="Find, filter, split, and extract data from the AST" href="/docs/parse/tree-utils" />
</Cards>

What the AST SDK provides [#what-the-ast-sdk-provides]

| Capability      | Functions                                                                                                          |
| --------------- | ------------------------------------------------------------------------------------------------------------------ |
| **Parsing**     | `parse()`, `parseBlocks()`                                                                                         |
| **Frontmatter** | `extractFrontmatter()`, `splitFrontmatter()`                                                                       |
| **YAML**        | `parseYaml()`, `stringifyYaml()`                                                                                   |
| **Querying**    | `find()`, `findAll()`, `filter()`, `sectionsAtDepth()`, `sectionsByHeading()`                                      |
| **Splitting**   | `splitBy()`, `between()`, `after()`, `before()`                                                                    |
| **Extraction**  | `codeBlocks()`, `inlineText()`, `textContent()`, `extractTodoItems()`                                              |
| **Type guards** | `isHeading()`, `isCodeBlock()`, `isTodoItem()`, `isSection()`, `isBlockNode()`, `isInlineNode()`, `isParentNode()` |

What the Format SDK adds on top [#what-the-format-sdk-adds-on-top]

The [Format SDK](/docs/template) builds on the Parse SDK and adds:

* **`datamark()`** — a factory that wraps `parse`, `stringify`, `schema`, and `examples` into a single typed format
* **Typed frontmatter** — `frontmatterSchema` validates and types `doc.frontmatter` before your `parse` function runs
* **Builder primitives** — `heading()`, `list()`, `codeBlock()`, `frontmatter()` from `datamark/stringify` replace fragile string concatenation
* **Self-testing** — `examples` are validated by `.test()` automatically

When to use the Parse SDK directly [#when-to-use-the-parse-sdk-directly]

* You need to parse Markdown without defining a full format
* You want to inspect or transform an AST interactively
* You're building custom tooling on top of datamark's parser
* You need to extract data from an AST without serialization

For most use cases — especially when you want typed data, validation, and round-trip serialization — the [Format SDK](/docs/template) is the better starting point.