diff --git a/packages/markdown-parser/src/core/parser.ts b/packages/markdown-parser/src/core/parser.ts
index d244548..c105f8a 100644
--- a/packages/markdown-parser/src/core/parser.ts
+++ b/packages/markdown-parser/src/core/parser.ts
@@ -3,6 +3,7 @@ import { BlockState, InlineState } from './state';
 import { linkify } from '@fuyeor/linkify';
 import type {
   ASTNode,
+  ASTTransform,
   BlockRule,
   InlineRule,
   Linkifier,
@@ -23,6 +24,7 @@ function linkifyWithCandidateCheck(text: string): ReturnType<typeof linkify> {
 export class MarkdownParser {
   #blockRuleMap = new Map<string, BlockRule[]>();
   #inlineRuleMap = new Map<string, InlineRule[]>();
+  #astTransforms: ASTTransform[] = [];
   #idSequence = 0;
   #isPreflight = false;
   readonly #maxNestingDepth: number;
@@ -82,11 +84,21 @@ export class MarkdownParser {
     return this;
   }
 
+  // Register a post-parse AST transform for extensions such as Twemoji.
+  addAstTransform(transform: ASTTransform): this {
+    this.#astTransforms.push(transform);
+    return this;
+  }
+
   build(): (content: string) => ASTNode[] {
     return (content: string) => {
       this.#idSequence = 0;
       const state = new BlockState(content);
-      return this.#parseBlocks(state);
+      const nodes = this.#parseBlocks(state);
+      return this.#astTransforms.reduce(
+        (currentNodes, transform) => transform(currentNodes),
+        nodes,
+      );
     };
   }
 
@@ -165,7 +177,9 @@ export class MarkdownParser {
             // ` for code block and inline code
             firstChar === 96 ||
             // ~ equals `
-            firstChar === 126;
+            firstChar === 126 ||
+            // $ for block math
+            firstChar === 36;
 
           if (mayInterrupt) {
             const rules = this.#blockRuleMap.get(
@@ -182,6 +196,7 @@ export class MarkdownParser {
                     'blockquote',
                     'list',
                     'code_block',
+                    'math_block',
                     'ffm_blocks',
                   ].includes(rule.name)
                 ) {
diff --git a/packages/markdown-parser/src/core/render.ts b/packages/markdown-parser/src/core/render.ts
index dc00400..5ce79d7 100644
--- a/packages/markdown-parser/src/core/render.ts
+++ b/packages/markdown-parser/src/core/render.ts
@@ -31,6 +31,29 @@ const getTableAlignment = (align: unknown) =>
     ? ` align="${align}"`
     : '';
 
+const displayLanguageNames: Record<string, string> = {
+  javascript: 'JavaScript',
+  typescript: 'TypeScript',
+  cpp: 'C++',
+  csharp: 'C#',
+  xml: 'XML',
+  json: 'JSON',
+  yaml: 'YAML',
+  ffm: 'Fuyeor Flavored Markdown',
+  fon: 'FON',
+  fer: 'Fer',
+};
+
+// Render a stable fenced-code contract for asynchronous highlight.js processing.
+const getCodeLanguage = (lang: unknown): string => {
+  if (typeof lang !== 'string') return '';
+  return lang.trim().split(/\s+/u)[0] ?? '';
+};
+
+const getDisplayLanguageName = (language: string): string =>
+  displayLanguageNames[language] ??
+  (language ? language.charAt(0).toUpperCase() + language.slice(1) : 'Text');
+
 export function render(nodes?: ASTNode[]): string {
   let html = '';
 
@@ -52,6 +75,19 @@ export function render(nodes?: ASTNode[]): string {
       case 'text':
         html += node.content ? escapeHtml(node.content) : '';
         break;
+      case 'emoji': {
+        const emoji = String(node.content ?? '');
+        const alt = String(node.alt ?? emoji);
+        const src = String(node.src ?? '');
+        html += `<img class="emoji" draggable="false" alt="${escapeHtml(alt)}" src="${escapeHtml(src)}"/>`;
+        break;
+      }
+      case 'math_inline':
+        html += `<span class="math-inline">${escapeHtml(String(node.content ?? ''))}</span>`;
+        break;
+      case 'math_block':
+        html += `<div class="math-block">${escapeHtml(String(node.content ?? ''))}</div>\n`;
+        break;
       case 'bold':
         html += `<strong>${render(node.children)}</strong>`;
         break;
@@ -83,13 +119,19 @@ export function render(nodes?: ASTNode[]): string {
           : render(node.children);
         break;
       }
-      case 'code_block':
+      case 'code_block': {
+        const language = getCodeLanguage(node.lang);
+        const displayLanguage = getDisplayLanguageName(language);
+        const preAttributes = language
+          ? `class="hljs language-${escapeHtml(language)}" data-language="${escapeHtml(language)}"`
+          : 'class="hljs language-plaintext"';
         html += `<div class="code-block-wrapper">`;
-        if (node.lang) {
-          html += `<div class="code-lang">${escapeHtml(node.lang)}</div>`;
+        if (language) {
+          html += `<div class="code-lang">${escapeHtml(displayLanguage)}</div>`;
         }
-        html += `<pre><code${node.lang ? ` class="language-${escapeHtml(node.lang)}"` : ''}>${node.content ? escapeHtml(node.content) : ''}\n</code></pre></div>\n`;
+        html += `<pre ${preAttributes}><code>${escapeHtml(String(node.content ?? ''))}\n</code></pre></div>\n`;
         break;
+      }
       case 'list': {
         const tag = node.ordered ? 'ol' : 'ul';
         const start =
@@ -102,6 +144,26 @@ export function render(nodes?: ASTNode[]): string {
       case 'list_item':
         html += `<li>${render(node.children)}</li>\n`;
         break;
+      case 'mermaid':
+        html += `<div class="language-mermaid">${escapeHtml(String(node.content ?? ''))}</div>\n`;
+        break;
+      case 'abc':
+        html += `<div class="language-abc">${escapeHtml(String(node.content ?? ''))}</div>\n`;
+        break;
+      case 'smiles_inline':
+        html += `<span class="language-smiles smiles-inline" data-smiles="${escapeHtml(String(node.content ?? ''))}"></span>`;
+        break;
+      case 'smiles_block': {
+        const blocks = String(node.content ?? '')
+          .split('\n')
+          .filter((line) => line.trim().length > 0)
+          .map(
+            (line) =>
+              `<div class="language-smiles smiles-block" data-smiles="${escapeHtml(line.trim())}"></div>`,
+          );
+        html += blocks.length > 0 ? `${blocks.join('\n')}\n` : '';
+        break;
+      }
       case 'table': {
         html += '<table>\n';
         if (node.headers) {
diff --git a/packages/markdown-parser/src/default.ts b/packages/markdown-parser/src/default.ts
index 8c5ac97..e44aa12 100644
--- a/packages/markdown-parser/src/default.ts
+++ b/packages/markdown-parser/src/default.ts
@@ -19,6 +19,8 @@ import {
   strikeRule,
 } from './rules/inlines';
 import { ffmBlockRule } from './rules/ffm';
+import { latexPlugin } from './plugins/latex';
+import { extensionsPlugin } from './plugins/extensions';
 
 export function createMarkdownParser(options: MarkdownParserOptions = {}) {
   return (
@@ -57,6 +59,8 @@ export function createFuyeorMarkdownParser(
       .addBlockRule(tableRule)
       .addBlockRule(hrRule)
       .addBlockRule(blockquoteRule)
+      .use(latexPlugin)
+      .use(extensionsPlugin)
 
       // inline order: Inline code -> Links -> Bold/Underline/Strikethrough
       .addInlineRule(hardBreakRule)
diff --git a/packages/markdown-parser/src/index.spec.ts b/packages/markdown-parser/src/index.spec.ts
deleted file mode 100644
index 7cf9cec..0000000
--- a/packages/markdown-parser/src/index.spec.ts
+++ /dev/null
@@ -1,179 +0,0 @@
-// @fuyeor/markdown-parser/src/index.spec.ts
-import { describe, it, expect } from 'vitest';
-import { MarkdownParser } from './core/parser';
-import { createFuyeorMarkdownParser } from './default';
-import { render } from './core/render';
-import { headingRule, codeBlockRule, tableRule } from './rules/blocks';
-import { boldRule, linkRule } from './rules/inlines';
-
-// build parser
-const parse = new MarkdownParser()
-  .addBlockRule(tableRule)
-  .addBlockRule(codeBlockRule)
-  .addBlockRule(headingRule)
-  .addInlineRule(boldRule)
-  .addInlineRule(linkRule)
-  .build();
-
-describe('test @fuyeor/markdown-parser', () => {
-  it('parse and filter malicious code', () => {
-    const ast = parse('text <script>alert(1)</script>');
-
-    expect(ast[0].type).toBe('paragraph');
-    expect(ast[0].children![0].content).toBe('text <script>alert(1)</script>');
-  });
-
-  it('parse ATX headings and internal bolding', () => {
-    const ast = parse('## Hello **World**');
-
-    expect(ast[0].type).toBe('heading');
-    expect(ast[0].level).toBe(2);
-    expect(ast[0].children![0].type).toBe('text');
-    expect(ast[0].children![0].content).toBe('Hello ');
-    expect(ast[0].children![1].type).toBe('bold');
-  });
-
-  it('parse fenced code block', () => {
-    const content = '```ts\nconst a = 1;\n```';
-    const ast = parse(content);
-
-    expect(ast[0].type).toBe('code_block');
-    expect(ast[0].lang).toBe('ts');
-    expect(ast[0].content).toBe('const a = 1;');
-  });
-
-  it('parse standard links', () => {
-    const ast = parse(
-      'visit www.fuyeor.com or [click here](https://fuyeor.com)',
-    );
-    const children = ast[0].children!;
-
-    // automatic completion protocol
-    expect(children[1].type).toBe('link');
-    expect(children[1].url).toBe('https://www.fuyeor.com');
-
-    // generate standard link
-    expect(children[3].type).toBe('link');
-    expect(children[3].url).toBe('https://fuyeor.com');
-  });
-
-  it('uses linkify for internationalized fuzzy domains', () => {
-    const ast = createFuyeorMarkdownParser()('Visit fuyeor.xn--p1ai');
-    const link = ast[0].children?.find((node) => node.type === 'link');
-
-    expect(link).toMatchObject({
-      url: 'https://fuyeor.рф',
-      children: [{ type: 'text', content: 'fuyeor.рф' }],
-    });
-  });
-
-  it('parse table', () => {
-    const content = `
-| title 1 | title 2 |
-|---|---|
-| content 1 | content 2 |
-`;
-    const ast = parse(content);
-    const tableNode = ast.find((n) => n.type === 'table')!;
-
-    expect(tableNode).toBeDefined();
-    expect(tableNode.headers).toHaveLength(2);
-    expect(tableNode.children![0].type).toBe('table_row');
-  });
-
-  it('renders aligned headed tables', () => {
-    const ast = createFuyeorMarkdownParser()(
-      '| 属性 | 类型 | 说明 |\n| :--- | :---: | ---: |\n| name | string | 用户名 |',
-    );
-
-    expect(render(ast)).toBe(`<table>
-<thead>
-<tr>
-<th align="left">属性</th>
-<th align="center">类型</th>
-<th align="right">说明</th>
-</tr>
-</thead>
-<tbody>
-<tr>
-<td align="left">name</td>
-<td align="center">string</td>
-<td align="right">用户名</td>
-</tr>
-</tbody>
-</table>
-`);
-  });
-
-  it('renders separator-first tables without a header', () => {
-    const ast = createFuyeorMarkdownParser()(
-      '| :--- | ---: |\n| Ray ID | a2c5ac427b7aa727 |\n| IP 地址 | 172.214.47.18 |',
-    );
-    const tableNode = ast.find((node) => node.type === 'table');
-
-    expect(tableNode?.headers).toBeUndefined();
-    expect(tableNode?.children).toHaveLength(2);
-    expect(render(ast)).toBe(`<table>
-<tbody>
-<tr>
-<td align="left">Ray ID</td>
-<td align="right">a2c5ac427b7aa727</td>
-</tr>
-<tr>
-<td align="left">IP 地址</td>
-<td align="right">172.214.47.18</td>
-</tr>
-</tbody>
-</table>
-`);
-  });
-
-  it('rejects unsafe link schemes', () => {
-    const ast = parse('[click](javascript:alert(1))');
-
-    expect(ast[0].children?.some((node) => node.type === 'link')).toBe(false);
-  });
-
-  it('bounds recursive block parsing', () => {
-    const boundedParse = createFuyeorMarkdownParser({ maxNestingDepth: 8 });
-
-    expect(() => boundedParse(`${'>'.repeat(100)} value`)).not.toThrow();
-  });
-
-  it('fails fast for an invalid nesting depth', () => {
-    expect(() => new MarkdownParser({ maxNestingDepth: 0 })).toThrow(
-      RangeError,
-    );
-  });
-
-  it.each([
-    ['modern two-space indentation', '  - example'],
-    ['legacy three-space marker alignment', '   - example'],
-  ])('recognizes %s as a nested list', (_description, childLine) => {
-    const ast = createFuyeorMarkdownParser()(`1. example\n${childLine}`);
-    const rootList = ast[0];
-
-    expect(rootList.type).toBe('list');
-    expect(rootList.children).toHaveLength(1);
-    expect(
-      rootList.children?.[0].children?.some((node) => node.type === 'list'),
-    ).toBe(true);
-  });
-
-  it('uses two-space steps for deeper list nesting', () => {
-    const ast = createFuyeorMarkdownParser()(
-      '1. root\n   - level 1\n    - level 2',
-    );
-    const rootItem = ast[0].children?.[0];
-    const levelOneList = rootItem?.children?.find(
-      (node) => node.type === 'list',
-    );
-    const levelOneItem = levelOneList?.children?.[0];
-    const levelTwoList = levelOneItem?.children?.find(
-      (node) => node.type === 'list',
-    );
-
-    expect(levelOneList).toBeDefined();
-    expect(levelTwoList).toBeDefined();
-  });
-});
diff --git a/packages/markdown-parser/src/index.ts b/packages/markdown-parser/src/index.ts
index aaa247b..1414985 100644
--- a/packages/markdown-parser/src/index.ts
+++ b/packages/markdown-parser/src/index.ts
@@ -19,11 +19,37 @@ export {
   underlineRule,
   strikeRule,
 } from './rules/inlines';
+export { latexPlugin, latexInlineRule, latexBlockRule } from './plugins/latex';
+export {
+  twemojiPlugin,
+  nativeEmojiPlugin,
+  twemojiTransform,
+  transformTwemojiNode,
+  getTwemojiUrl,
+} from './plugins/twemoji';
+export { highlightPlugin, highlightTransform } from './plugins/highlight';
+export {
+  abcPlugin,
+  abcTransform,
+  mermaidPlugin,
+  mermaidTransform,
+  smilesBlockPlugin,
+  smilesBlockTransform,
+  smilesInlinePlugin,
+  smilesPlugin,
+  smilesInlineRule,
+  SPECIAL_BLOCK_LANGUAGES,
+  specialBlockPlugin,
+  specialBlockTransform,
+  transformSpecialBlockNode,
+} from './plugins/async-blocks';
+export { extensionsPlugin, extensionsTransform } from './plugins/extensions';
 export { createMarkdownParser, createFuyeorMarkdownParser } from './default';
 
 export type {
   NodeType,
   ASTNode,
+  ASTTransform,
   ParserContext,
   BlockRule,
   InlineRule,
diff --git a/packages/markdown-parser/src/plugins/ast.ts b/packages/markdown-parser/src/plugins/ast.ts
new file mode 100644
index 0000000..db9fd36
--- /dev/null
+++ b/packages/markdown-parser/src/plugins/ast.ts
@@ -0,0 +1,37 @@
+// @fuyeor/markdown-parser/src/plugins/ast.ts
+import type { ASTNode } from '#/types';
+
+type ASTNodeReplacement = ASTNode | readonly ASTNode[];
+
+// Recursively transform only changed branches and preserve unchanged node arrays.
+export const mapAstNodes = (
+  nodes: ASTNode[],
+  transformNode: (node: ASTNode) => ASTNodeReplacement,
+): ASTNode[] => {
+  let transformedNodes: ASTNode[] | undefined;
+
+  for (let index = 0; index < nodes.length; index++) {
+    const node = nodes[index];
+    const transformedChildren = node.children
+      ? mapAstNodes(node.children, transformNode)
+      : node.children;
+    const transformedNode =
+      transformedChildren && transformedChildren !== node.children
+        ? { ...node, children: transformedChildren }
+        : node;
+    const replacement = transformNode(transformedNode);
+    const replacementNodes = Array.isArray(replacement)
+      ? replacement
+      : [replacement];
+
+    if (
+      !transformedNodes &&
+      (replacementNodes.length !== 1 || replacementNodes[0] !== node)
+    ) {
+      transformedNodes = nodes.slice(0, index);
+    }
+    transformedNodes?.push(...replacementNodes);
+  }
+
+  return transformedNodes ?? nodes;
+};
diff --git a/packages/markdown-parser/src/plugins/async-blocks.ts b/packages/markdown-parser/src/plugins/async-blocks.ts
new file mode 100644
index 0000000..1af6b5a
--- /dev/null
+++ b/packages/markdown-parser/src/plugins/async-blocks.ts
@@ -0,0 +1,94 @@
+// @fuyeor/markdown-parser/src/plugins/async-blocks.ts
+import { InlineState } from '#/core/state';
+import type {
+  ASTNode,
+  ASTTransform,
+  InlineRule,
+  MarkdownPlugin,
+} from '#/types';
+import { mapAstNodes } from './ast';
+import { normalizeHighlightLanguage } from './highlight';
+
+const SMILES_INLINE_PATTERN = /^#\[smiles\s*=\s*`([^`]+)`\]/u;
+export const SPECIAL_BLOCK_LANGUAGES = new Set(['mermaid', 'abc', 'smiles']);
+
+// Parse the FFM #[smiles = `...`] inline placeholder syntax.
+export const smilesInlineRule: InlineRule = {
+  name: 'smiles_inline',
+  markers: ['#'],
+  parse(state: InlineState) {
+    if (state.currentChar !== '#') return null;
+
+    const match = state.content.slice(state.pos).match(SMILES_INLINE_PATTERN);
+    if (!match) return null;
+
+    return {
+      node: { type: 'smiles_inline', content: match[1].trim() },
+      consumedChars: match[0].length,
+    };
+  },
+};
+
+// Convert one fenced code block into its asynchronous renderer placeholder.
+export const transformSpecialBlockNode = (
+  node: ASTNode,
+  languages: ReadonlySet<string>,
+): ASTNode => {
+  if (node.type !== 'code_block') return node;
+
+  const language = normalizeHighlightLanguage(String(node.lang ?? ''));
+  if (!languages.has(language)) return node;
+
+  return {
+    type: language === 'smiles' ? 'smiles_block' : language,
+    content: String(node.content ?? ''),
+  } satisfies ASTNode;
+};
+
+// Build a lazy transform for a selected set of special fenced languages.
+const createSpecialBlockTransform =
+  (languages: ReadonlySet<string>): ASTTransform =>
+  (nodes) =>
+    mapAstNodes(nodes, (node) => transformSpecialBlockNode(node, languages));
+
+export const mermaidTransform = createSpecialBlockTransform(
+  new Set(['mermaid']),
+);
+export const abcTransform = createSpecialBlockTransform(new Set(['abc']));
+export const smilesBlockTransform = createSpecialBlockTransform(
+  new Set(['smiles']),
+);
+export const specialBlockTransform = createSpecialBlockTransform(
+  SPECIAL_BLOCK_LANGUAGES,
+);
+
+// Register Mermaid, ABC, and fenced Smiles block placeholders together.
+export const specialBlockPlugin: MarkdownPlugin = (parser) => {
+  parser.addAstTransform(specialBlockTransform);
+};
+
+// Register only Mermaid fenced blocks for consumers that opt in selectively.
+export const mermaidPlugin: MarkdownPlugin = (parser) => {
+  parser.addAstTransform(mermaidTransform);
+};
+
+// Register only ABC fenced blocks for consumers that opt in selectively.
+export const abcPlugin: MarkdownPlugin = (parser) => {
+  parser.addAstTransform(abcTransform);
+};
+
+// Register only fenced Smiles blocks for consumers that opt in selectively.
+export const smilesBlockPlugin: MarkdownPlugin = (parser) => {
+  parser.addAstTransform(smilesBlockTransform);
+};
+
+// Register the FFM inline Smiles syntax.
+export const smilesInlinePlugin: MarkdownPlugin = (parser) => {
+  parser.addInlineRule(smilesInlineRule);
+};
+
+// Register both inline and fenced Smiles syntax.
+export const smilesPlugin: MarkdownPlugin = (parser) => {
+  smilesInlinePlugin(parser);
+  smilesBlockPlugin(parser);
+};
diff --git a/packages/markdown-parser/src/plugins/extensions.ts b/packages/markdown-parser/src/plugins/extensions.ts
new file mode 100644
index 0000000..0086a1e
--- /dev/null
+++ b/packages/markdown-parser/src/plugins/extensions.ts
@@ -0,0 +1,32 @@
+// @fuyeor/markdown-parser/src/plugins/extensions.ts
+import type { ASTNode, ASTTransform, MarkdownPlugin } from '#/types';
+import { mapAstNodes } from './ast';
+import { transformTwemojiNode } from './twemoji';
+import {
+  SPECIAL_BLOCK_LANGUAGES,
+  transformSpecialBlockNode,
+  smilesInlinePlugin,
+} from './async-blocks';
+import { normalizeHighlightLanguage } from './highlight';
+
+// Normalize code languages and special blocks in the same pass as Twemoji.
+const transformExtensionNode = (
+  node: ASTNode,
+): ASTNode | readonly ASTNode[] => {
+  if (node.type === 'text') return transformTwemojiNode(node);
+  if (node.type !== 'code_block') return node;
+
+  const language = normalizeHighlightLanguage(String(node.lang ?? ''));
+  const normalizedNode =
+    language && language !== node.lang ? { ...node, lang: language } : node;
+  return transformSpecialBlockNode(normalizedNode, SPECIAL_BLOCK_LANGUAGES);
+};
+
+export const extensionsTransform: ASTTransform = (nodes) =>
+  mapAstNodes(nodes, transformExtensionNode);
+
+// Register all five asynchronous/frontend-compatible extensions together.
+export const extensionsPlugin: MarkdownPlugin = (parser) => {
+  parser.addAstTransform(extensionsTransform);
+  smilesInlinePlugin(parser);
+};
diff --git a/packages/markdown-parser/src/plugins/highlight.ts b/packages/markdown-parser/src/plugins/highlight.ts
new file mode 100644
index 0000000..442c76d
--- /dev/null
+++ b/packages/markdown-parser/src/plugins/highlight.ts
@@ -0,0 +1,41 @@
+// @fuyeor/markdown-parser/src/plugins/highlight.ts
+import type { ASTNode, ASTTransform, MarkdownPlugin } from '#/types';
+import { mapAstNodes } from './ast';
+
+const languageAliases: Record<string, string> = {
+  'c++': 'cpp',
+  'c#': 'csharp',
+  cs: 'csharp',
+  js: 'javascript',
+  ts: 'typescript',
+  sh: 'bash',
+  yml: 'yaml',
+  html: 'xml',
+  md: 'markdown',
+  rs: 'rust',
+  rb: 'ruby',
+  pl: 'perl',
+  ps1: 'powershell',
+  py: 'python',
+};
+
+// Normalize fenced-code aliases to the language names used by highlight.js.
+export const normalizeHighlightLanguage = (rawLanguage: string): string => {
+  const language = rawLanguage.trim().split(/\s+/u)[0]?.toLowerCase() ?? '';
+  return languageAliases[language] ?? language;
+};
+
+export const highlightTransform: ASTTransform = (nodes) =>
+  mapAstNodes(nodes, (node) => {
+    if (node.type !== 'code_block') return node;
+
+    const language = normalizeHighlightLanguage(String(node.lang ?? ''));
+    return language && language !== node.lang
+      ? { ...node, lang: language }
+      : node;
+  });
+
+// Register language normalization for asynchronous highlight.js processing.
+export const highlightPlugin: MarkdownPlugin = (parser) => {
+  parser.addAstTransform(highlightTransform);
+};
diff --git a/packages/markdown-parser/src/plugins/latex.ts b/packages/markdown-parser/src/plugins/latex.ts
new file mode 100644
index 0000000..8e561c7
--- /dev/null
+++ b/packages/markdown-parser/src/plugins/latex.ts
@@ -0,0 +1,99 @@
+// @fuyeor/markdown-parser/src/plugins/latex.ts
+import { BlockState, InlineState } from '#/core/state';
+import type { BlockRule, InlineRule, MarkdownPlugin } from '#/types';
+
+const MATH_INLINE_MARKER = '$';
+const MATH_BLOCK_MARKER = '$$';
+
+// Find an unescaped dollar delimiter without interpreting LaTeX escapes.
+const findMathDelimiter = (
+  content: string,
+  from: number,
+  delimiter: string,
+): number => {
+  let position = content.indexOf(delimiter, from);
+  while (position !== -1) {
+    let backslashCount = 0;
+    for (
+      let index = position - 1;
+      index >= 0 && content[index] === '\\';
+      index--
+    )
+      backslashCount++;
+    if (backslashCount % 2 === 0) return position;
+    position = content.indexOf(delimiter, position + delimiter.length);
+  }
+  return -1;
+};
+
+// Parse inline $...$ formulas while leaving unmatched dollars as plain text.
+export const latexInlineRule: InlineRule = {
+  name: 'math_inline',
+  markers: [MATH_INLINE_MARKER],
+  parse(state: InlineState) {
+    if (state.currentChar !== MATH_INLINE_MARKER) return null;
+    if (state.content.startsWith(MATH_BLOCK_MARKER, state.pos)) return null;
+
+    const endIndex = findMathDelimiter(
+      state.content,
+      state.pos + MATH_INLINE_MARKER.length,
+      MATH_INLINE_MARKER,
+    );
+    if (endIndex === -1) return null;
+
+    return {
+      node: {
+        type: 'math_inline',
+        content: state.content
+          .slice(state.pos + MATH_INLINE_MARKER.length, endIndex)
+          .trim(),
+      },
+      consumedChars: endIndex - state.pos + MATH_INLINE_MARKER.length,
+    };
+  },
+};
+
+// Parse a $$...$$ formula as one block, including formulas spanning lines.
+export const latexBlockRule: BlockRule = {
+  name: 'math_block',
+  markers: [MATH_INLINE_MARKER],
+  parse(state: BlockState) {
+    const firstLine = state.currentLine;
+    if (!firstLine) return null;
+
+    const trimmedLine = firstLine.trimStart();
+    if (!trimmedLine.startsWith(MATH_BLOCK_MARKER)) return null;
+
+    const contentLines: string[] = [];
+    let consumedLines = 1;
+    let currentContent = trimmedLine.slice(MATH_BLOCK_MARKER.length);
+    let endIndex = findMathDelimiter(currentContent, 0, MATH_BLOCK_MARKER);
+
+    while (endIndex === -1) {
+      contentLines.push(currentContent);
+      if (state.lineIndex + consumedLines >= state.lineCount) return null;
+      currentContent = state.lines[state.lineIndex + consumedLines];
+      consumedLines++;
+      endIndex = findMathDelimiter(currentContent, 0, MATH_BLOCK_MARKER);
+    }
+
+    contentLines.push(currentContent.slice(0, endIndex));
+    const trailingContent = currentContent.slice(
+      endIndex + MATH_BLOCK_MARKER.length,
+    );
+    if (trailingContent.trim() !== '') return null;
+
+    return {
+      node: {
+        type: 'math_block',
+        content: contentLines.join('\n').trim(),
+      },
+      consumedLines,
+    };
+  },
+};
+
+// Register LaTeX block and inline rules on a markdown-parser instance.
+export const latexPlugin: MarkdownPlugin = (parser) => {
+  parser.addBlockRule(latexBlockRule).addInlineRule(latexInlineRule);
+};
diff --git a/packages/markdown-parser/src/plugins/twemoji.ts b/packages/markdown-parser/src/plugins/twemoji.ts
new file mode 100644
index 0000000..cd311db
--- /dev/null
+++ b/packages/markdown-parser/src/plugins/twemoji.ts
@@ -0,0 +1,67 @@
+// @fuyeor/markdown-parser/src/plugins/twemoji.ts
+import type { ASTNode, ASTTransform, MarkdownPlugin } from '#/types';
+import { mapAstNodes } from './ast';
+
+const r = String.raw;
+const baseEmojiPattern = r`\p{Emoji}(?:\p{EMod}|[\u{E0020}-\u{E007E}]+\u{E007F}|\uFE0F?\u20E3?)`;
+const emojiPattern = r`\p{RI}{2}|(?![#*\d](?!\uFE0F?\u20E3))${baseEmojiPattern}(?:\u200D${baseEmojiPattern})*`;
+const emojiRegex = new RegExp(emojiPattern, 'gu');
+
+// Convert a raw Emoji sequence to the Twemoji CDN codepoint path.
+export const getTwemojiUrl = (emoji: string): string => {
+  const codepoints = Array.from(emoji)
+    .map((character) => character.codePointAt(0)!.toString(16))
+    .filter((codepoint) => codepoint !== 'fe0f')
+    .join('-');
+  return `https://deliver.fuyeor.net/@libs/twemoji-new/svg/${codepoints}.svg`;
+};
+
+// Split text nodes into plain text and emoji image nodes without touching code.
+export const transformTwemojiNode = (node: ASTNode): ASTNode[] => {
+  const content = node.content;
+  if (typeof content !== 'string' || content.length === 0) return [node];
+
+  emojiRegex.lastIndex = 0;
+  if (!emojiRegex.test(content)) return [node];
+
+  const nodes: ASTNode[] = [];
+  let lastIndex = 0;
+  emojiRegex.lastIndex = 0;
+  let match: RegExpExecArray | null;
+
+  while ((match = emojiRegex.exec(content)) !== null) {
+    const emoji = match[0];
+    const matchIndex = match.index;
+    if (matchIndex > lastIndex) {
+      nodes.push({
+        type: 'text',
+        content: content.slice(lastIndex, matchIndex),
+      });
+    }
+    nodes.push({
+      type: 'emoji',
+      content: emoji,
+      alt: emoji,
+      src: getTwemojiUrl(emoji),
+    });
+    lastIndex = matchIndex + emoji.length;
+  }
+
+  if (lastIndex < content.length) {
+    nodes.push({ type: 'text', content: content.slice(lastIndex) });
+  }
+  return nodes;
+};
+
+export const twemojiTransform: ASTTransform = (nodes) =>
+  mapAstNodes(nodes, (node) =>
+    node.type === 'text' ? transformTwemojiNode(node) : node,
+  );
+
+// Register Twemoji conversion as a post-parse AST transform.
+export const twemojiPlugin: MarkdownPlugin = (parser) => {
+  parser.addAstTransform(twemojiTransform);
+};
+
+// Keep the name used by the existing Interactify markdown-it integration.
+export const nativeEmojiPlugin = twemojiPlugin;
diff --git a/packages/markdown-parser/src/types.ts b/packages/markdown-parser/src/types.ts
index 460d45d..bec0595 100644
--- a/packages/markdown-parser/src/types.ts
+++ b/packages/markdown-parser/src/types.ts
@@ -17,6 +17,13 @@ export type NodeType =
   | 'table'
   | 'table_row'
   | 'table_cell'
+  | 'math_inline'
+  | 'math_block'
+  | 'emoji'
+  | 'mermaid'
+  | 'abc'
+  | 'smiles_inline'
+  | 'smiles_block'
   | 'bold'
   | 'italic'
   | 'link'
@@ -81,4 +88,5 @@ export interface MarkdownParserOptions {
   linkifier?: Linkifier;
 }
 
+export type ASTTransform = (nodes: ASTNode[]) => ASTNode[];
 export type MarkdownPlugin = (parser: MarkdownParser) => void;