首次导出可能自动下载 Chrome Stable,需要网络;v2.1.0 起默认按 GFM 规则过滤 script、iframe、style 等原始 HTML;v2.0.0 起标题锚点改为 GitHub 兼容规则,旧文档内链可能失效
在 VS Code 扩展市场安装 Markdown PDF 后,打开 Markdown 文件按 F1 输入 export,选择 markdown-pdf: Export (pdf)README 未说明硬件要求;导出依赖本机安装的 Chrome 或 Edge 浏览器,也可自动下载 Chrome Stable
VS Code 扩展,把 Markdown 一键导出为 PDF、HTML、PNG 或 JPEG
VS Code 扩展,把 Markdown 文档一键导出成 PDF,支持 Mermaid、PlantUML 图表和语法高亮,适合写文档后直接出成品。
核心能力
- 支持导出 PDF、HTML、PNG、JPEG 四种格式
- 用 highlight.js 做代码块语法高亮
- 支持 Mermaid 与 PlantUML 图表渲染
- 用 KaTeX 渲染 LaTeX 数学公式
- 支持 :[label](path.md) 内嵌其他 Markdown 片段
容易踩的坑
- 2.2.0最新正式版2026-07-29 · 2026-07-29 15:52 UTC
2.2.0 (2026/07/28)
Breaking Changes
- Sanitization behavior change: with
markdown-pdf.sanitizeset to"gfm"(default) or"gfm-allow-style", block-level<style>,<script>, and<iframe>elements in the Markdown body are now removed entirely — tag and content — instead of having the opening<escaped and the content left as visible text in the output. This prevents CSS / JavaScript source code from appearing as literal text in exported files. Inline occurrences are still escaped as before. When something is removed during an export, a notification summarizes what was removed: a toast with a "Show Details" button on manual export, or an entry in the "Markdown PDF" output channel on automatic convert-on-save #437
Changes
- Add
Markdown PDF: Output Diagnosticscommand (Command Palette) that writes environment and configuration diagnostics — extension version, VS Code / OS info, Chromium path and its resolution source, and relevant settings — to the "Markdown PDF" output channel. Home directory paths are masked. Attach its output when filing an issue - Add a "Markdown PDF" output channel (log level aware) that collects all extension logs, and log an environment snapshot and conversion context at the start of each export to make failure reports diagnosable #437
- Improve error notifications: toasts now describe what failed in plain language instead of showing internal function names, add a one-line hint for common causes (Chromium launch failure, locked output file / permission denied, disk full, invalid output directory — with an "Open Settings" shortcut where applicable), and provide a "Show Details" button that opens the log with the full context, error message, and stack trace #437
- Chromium resolution failures now report the underlying reason — network / proxy download failure,
markdown-pdf.chromium.autoDownloaddisabled, or no usable browser found — instead of a generic "Chromium does not exist" message #436 - Error logs now follow
Error.causechains and expandAggregateError, and missing shared system libraries (a common Chromium launch failure on minimal Linux environments) are detected and called out explicitly
Fixes
- Fix: Documents containing an unmatched backtick before a fenced code block no longer have a chunk of content duplicated as raw Markdown in the export. The
markdown-it-includecode-region scanner could pair an opening backtick with a closing backtick on the far side of a fenced block, producing overlapping protected regions that were emitted twice. Affects 2.0.0 through 2.1.0; the include scan runs on every export becausemarkdown-pdf.markdown-it-include.enabledefaults totrue#443 #444 - Fix: Export no longer hangs indefinitely when
markdown-pdf.sanitizeis"none"and the document contains scripts that open blocking dialogs (alert(),confirm(),prompt(),beforeunload). Such dialogs are now auto-dismissed during rendering and recorded in the output channel - Fix: When HTML generation fails, the export is now skipped instead of writing a corrupted file. Previously the literal string
undefinedor template-wrapped garbage was written to the output path, silently overwriting a previous good export - Fix: The
Exporting (pdf) ...progress notification no longer stays open indefinitely when post-export cleanup stalls. Browser shutdown is now bounded by a timeout, and the temporary HTML file is reliably removed #374
- Sanitization behavior change: with
- 2.1.02026-05-24 · 2026-05-24 13:32 UTC
查看更新说明
2.1.0 (2026/05/24)
Breaking Changes
- Security hardening: Raw HTML in Markdown is now sanitized by default according to the GFM Disallowed Raw HTML extension. The following are removed from Markdown body content:
- Tags:
<script>,<iframe>,<style>,<textarea>,<title>,<xmp>,<noembed>,<noframes>,<plaintext>(opening<is escaped to<, content is preserved as visible text) on*event handler attributes (onclick,onload, etc.)href/srcattributes whose value begins withjavascript:
- Tags:
- The behavior is controlled by the new
markdown-pdf.sanitizesetting ("gfm"/"gfm-allow-style"/"none", default"gfm"). See README for details and migration notes. - To preserve pre-change behavior, set
markdown-pdf.sanitizeto"none". To keep inline<style>only, use"gfm-allow-style". Existing layout CSS can also be migrated to external files viamarkdown-pdf.styles.
Changes
- Add
markdown-pdf.sanitizesetting for raw HTML sanitization (see Breaking Changes above) - Add support for
```plantumlfenced code blocks as a PlantUML syntax in addition to the existing@startuml/@endumlblock markers. Both are supported on equal footing (the fence form is the same one used by VS Code preview, GitHub, and GitLab) #92 #162 #389 - Automatically download the latest Chrome Stable build when Chromium is needed, instead of using only the build id pinned by
puppeteer-core. If the Chrome for Testing API is unreachable, the extension falls back to the latest cached build, and finally to thepuppeteer-corebundled build id. Add the newmarkdown-pdf.chromium.autoDownloadsetting (defaulttrue); set it tofalseto opt out of the automatic download and rely on an installed system browser (Chrome / Chromium / Edge) ormarkdown-pdf.executablePath. - Add math rendering support via KaTeX for
$...$,$$...$$,\(...\),\[...\], and```mathfenced code blocks (matches VS Code's built-in Markdown preview). Controlled by the newmarkdown-pdf.math.enabled(defaulttrue) andmarkdown-pdf.math.katex.macrossettings.
- Security hardening: Raw HTML in Markdown is now sanitized by default according to the GFM Disallowed Raw HTML extension. The following are removed from Markdown body content:
- 2.0.12026-04-13 · 2026-04-13 16:18 UTC
查看更新说明
2.0.1 (2026/04/14)
Fixes
- Fix: Self-closing
<div class="page" />now correctly triggers a page break #428
- Fix: Self-closing
- 2.0.02026-04-12 · 2026-04-12 20:06 UTC
查看更新说明
2.0.0 (2026/04/13)
Breaking Changes
- Heading IDs now follow GitHub-compatible VS Code slug generation. Existing internal anchors in your documents may change.
- Highlight.js upgraded from v9 to v11. Some highlight style names have been renamed or removed. Legacy names are mapped to current names when possible, and
tomorrow.cssis used as a fallback when no mapping is available. - Front matter parsing is now handled by a custom implementation. YAML sequences and non-plain objects in front matter are now rejected.
- Chromium download and cache management moved to a built-in
chromium-resolver. The previous temporary directory fallback has been removed; the managed Chromium is stored under the VS Code global storage directory.
Changes
- Migrate source code to TypeScript and bundle the extension with esbuild
- Bundle
puppeteer-coreand manage Chromium via the built-inchromium-resolver, preferring an installed Chrome/Edge and falling back to an automatic download - Replace
markdown-it-include,markdown-it-named-headers, andmarkdown-it-checkboxwith in-repo custom implementations - Remove
cheerio,mustache, andgray-matterdependencies in favor of internal implementations - Add unit and integration test suites using
vscode-test-cli, including sample generation and HTML snapshot comparison - Simplify VS Code variant detection and remove the obsolete temp-cache fallback
Fixes
- Fix global storage path resolution for the Chromium cache
- Gracefully handle include errors in the custom
markdown-it-includeplugin - Image
srctransformation now handles quoted attributes, flexible spacing, and raw-text contexts correctly - Front matter parsing now handles BOM-prefixed files and rejects invalid structures
- 1.5.02023-09-08 · 2023-09-08 14:50 UTC
查看更新说明
1.5.0 (2023/09/08)
- Improve: The default date format for headers and footers has been changed to the ISO-based format (YYYY-MM-DD).
- Support different date formats in templates #197
- Improve: Avoid TimeoutError: Navigation timeout of 30000 ms exceeded and TimeoutError: waiting for Page.printToPDF failed: timeout 30000ms exceeded #266
- Fix: Fix description of outputDirectoryRelativePathFile #238
- README
- Add: Specification Changes
- Fix: Broken link
- Improve: The default date format for headers and footers has been changed to the ISO-based format (YYYY-MM-DD).
用户点评
登录用户的真实使用记录,可按版本和硬件筛选还没有人点评。第一个说出真实体验的人最有价值。
Markdown PDF
This VS Code extension converts Markdown files to pdf, html, png or jpeg files.
Table of Contents
- What's New
- Breaking Changes
- Features
- Chromium
- Usage
- Extension Settings
- Options
- FAQ
- Known Issues
- Change Log
- License
- Sponsor
- Special thanks
What's New
Highlights of new features and improvements since v2. See Breaking Changes for changes that may affect existing behavior.
2.2.0
- Improve error notifications with actionable hints and a Show Details button (details)
- Add detailed logging to the new Markdown PDF output channel (details)
- Add
Markdown PDF: Output Diagnosticscommand for bug reports (details)
2.1.0
- Add PlantUML fenced code block support (details)
- Add math rendering via KaTeX (details)
- Auto-download latest Chrome Stable (details)
Breaking Changes
Changes since v2 that may affect existing behavior. See the FAQ section for details.
2.2.0
- With
markdown-pdf.sanitize: "gfm"(default) or"gfm-allow-style", block-level<style>,<script>, and<iframe>elements are now removed together with their content, instead of being escaped and left as visible text in the output. A notification reports what was removed during an export.
2.1.0
- Security hardening: To mitigate XSS-like risk (#411), raw HTML in Markdown is now sanitized by default following the GFM Disallowed Raw HTML extension. Tags such as
<script>,<iframe>,<style>, andon*/javascript:attributes are stripped from Markdown body content. The behavior is controlled by the new markdown-pdf.sanitize setting.
2.0.0
- Heading IDs now follow GitHub-compatible VS Code slug generation. Existing internal anchors in your documents may change.
- Details: Why did my heading anchors change?
- Highlight.js upgraded from v9 to v11. Some highlight style names have been renamed or removed.
- Front matter parsing is now stricter. Some previously accepted formats may be rejected.
- Chromium is resolved from an installed Chrome/Edge browser first, or auto-downloaded on first use.
Features
Markdown PDF adds the following authoring features on top of the default Markdown renderer when converting to PDF, HTML, PNG, or JPEG.
List
| Category | Feature | Description | Example |
|---|---|---|---|
| Basic syntax extensions | Syntax highlighting | Code block highlighting via highlight.js | ```js |
| Emoji | Emoji shortcodes | :smile: | |
| Checkbox | GitHub-style task lists | - [ ] / - [x] | |
| Heading IDs | GitHub-compatible heading anchors | # Heading → #heading | |
| Content composition | Container | Admonition-like blocks | ::: warning |
| Include | Embed Markdown fragments | :[label](https://github.com/yzane/vscode-markdown-pdf/blob/HEAD/path.md) | |
| Diagrams & math | PlantUML | UML diagrams from code blocks | @startuml … @enduml |
| Mermaid | Diagrams from fenced code blocks | ```mermaid | |
| Math | LaTeX math via KaTeX | $E = mc^2$ |
Basic syntax extensions
Checkbox
Render - [ ] / - [x] task-list items as disabled checkboxes, mirroring GitHub's task list rendering. Useful for status reports and checklists that should stay visible in the exported output.
Markdown
- [ ] Task A
- [x] Task B
Preview

Heading IDs
Headings receive GitHub-compatible anchor IDs automatically, so internal links such as [Section](#section) resolve the same way they do on GitHub. ASCII headings are lowercased with spaces replaced by hyphens; non-ASCII headings keep their original characters.
| Heading | Generated ID |
|---|---|
# My Heading | #my-heading |
# API Reference | #api-reference |
# 日本語見出し | #日本語見出し |
See also: Why did my heading anchors change? in the FAQ.
Content composition
Container
Admonition-like blocks via markdown-it-container. The identifier after ::: becomes the block's CSS class, so you can style warnings, tips, and notes by pairing it with markdown-pdf.styles.
Markdown
::: warning
**Warning:** here be dragons
:::
Stylesheet (for example markdown-pdf.css)
.warning {
border-left: 4px solid #f0ad4e;
background: #fff8e1;
padding: 12px 16px;
margin: 8px 0;
}
Settings
"markdown-pdf.styles": ["markdown-pdf.css"]
Preview

See also: markdown-pdf.styles.
Include
Embed the content of another Markdown file inline using :[alternate-text](https://github.com/yzane/vscode-markdown-pdf/blob/HEAD/relative-path-to-file.md). If a referenced fragment cannot be read (missing file, permission error, etc.), the extension reports the error at the include site and continues exporting the rest of the document.
Given the following directory layout (where README.md is the document being exported):
├── [plugins]
│ └── README.md
├── CHANGELOG.md
└── README.md
Markdown
README Content
:[Plugins](https://github.com/yzane/vscode-markdown-pdf/blob/HEAD/plugins/README.md)
:[Changelog](https://github.com/yzane/vscode-markdown-pdf/blob/HEAD/CHANGELOG.md)
Preview
Content of README.md
Content of plugins/README.md
Content of CHANGELOG.md
See also: markdown-pdf.markdown-it-include.enable.
Diagrams & math
PlantUML
Render UML diagrams via PlantUML using markdown-it-plantuml. Two equivalent syntaxes are supported; both produce the same <img> tag and share the markdown-pdf.plantumlServer setting.
Fenced code block
A ```plantuml fenced code block. This is the common fence convention used across the PlantUML ecosystem (for example, GitLab renders this form natively when the PlantUML integration is enabled).
Markdown
```plantuml
Bob -[#red]> Alice : hello
Alice -[#0000FF]->Bob : ok
```
Block markers
@startuml / @enduml block markers. The markers can be customized via markdown-pdf.plantumlOpenMarker and markdown-pdf.plantumlCloseMarker.
Markdown
@startuml
Bob -[#red]> Alice : hello
Alice -[#0000FF]->Bob : ok
@enduml
Preview (either form produces the same image)

See also: markdown-pdf.plantumlServer.
Mermaid
Render diagrams from fenced code blocks via Mermaid. The Mermaid library is loaded from the URL configured in markdown-pdf.mermaidServer (defaults to a CDN).
Markdown
Preview

Math
Render LaTeX math via KaTeX. Uses @vscode/markdown-it-katex (the same plugin VS Code's built-in Markdown preview ships) for $…$, $$…$$, and \begin{env}…\end{env}, plus a small in-house plugin for \(…\) and \[…\] bracket delimiters. Rendering runs in Node, so no network access is required.
Supported notations:
-
Inline:
$E = mc^2$,\(E = mc^2\) -
Display:
$$\int_0^\infty f(x)\,dx$$,\[\alpha\] -
LaTeX environments:
\begin{aligned}a &= b\\c &= d\end{aligned} -
Fenced code block:
```math \sum_{i=1}^{n} i = \frac{n(n+1)}{2} ```
Markdown
Preview

See also:
- markdown-pdf.math.enabled — disable math rendering
- markdown-pdf.math.katex.macros — custom KaTeX macros
Sample files
This README converted to each output format:
Chromium
Markdown PDF uses a Chromium-based browser for PDF/PNG/JPEG export. It tries the following sources in order:
- The path specified in markdown-pdf.executablePath
- An installed Google Chrome, Microsoft Edge, or Chromium on your system
- A managed Chromium automatically downloaded on first use and refreshed to track the latest Chrome Stable on subsequent VS Code launches (can be disabled with markdown-pdf.chromium.autoDownload)
See How is the Chromium browser selected? and Where is Chromium downloaded? in the FAQ for details.
If you are behind a proxy, set the http.proxy option in settings.json and restart Visual Studio Code.
Usage
Command Palette
- Open the Markdown file
- Press
F1orCtrl+Shift+P - Type
exportand select belowmarkdown-pdf: Export (settings.json)markdown-pdf: Export (pdf)markdown-pdf: Export (html)markdown-pdf: Export (png)markdown-pdf: Export (jpeg)markdown-pdf: Export (all: pdf, html, png, jpeg)
To collect environment information for a bug report, run Markdown PDF: Output Diagnostics. See How do I troubleshoot a failed export? in the FAQ.

Menu
- Open the Markdown file
- Right click and select below
markdown-pdf: Export (settings.json)markdown-pdf: Export (pdf)markdown-pdf: Export (html)markdown-pdf: Export (png)markdown-pdf: Export (jpeg)markdown-pdf: Export (all: pdf, html, png, jpeg)

Auto convert
- Add
"markdown-pdf.convertOnSave": trueoption to settings.json - Restart Visual Studio Code
- Open the Markdown file
- Auto convert on save
Extension Settings
Visual Studio Code User and Workspace Settings
- Select File > Preferences > UserSettings or Workspace Settings
- Find markdown-pdf settings in the Default Settings
- Copy
markdown-pdf.*settings - Paste to the settings.json, and change the value

Options
List
Save options
markdown-pdf.type
- Output format: pdf, html, png, jpeg
- Multiple output formats support
- Default: pdf
"markdown-pdf.type": [
"pdf",
"html",
"png",
"jpeg"
],
markdown-pdf.convertOnSave
- Enable Auto convert on save
- boolean. Default: false
- To apply the settings, you need to restart Visual Studio Code
markdown-pdf.convertOnSaveExclude
- Excluded file name of convertOnSave option
"markdown-pdf.convertOnSaveExclude": [
"^work",
"work.md$",
"work|test",
"[0-9][0-9][0-9][0-9]-work",
"work\\test" // All '\' need to be written as '\\' (Windows)
],
markdown-pdf.outputDirectory
- Output Directory
- All
\need to be written as\\(Windows)
"markdown-pdf.outputDirectory": "C:\\work\\output",
- Relative path
- If you open the
Markdown file, it will be interpreted as a relative path from the file - If you open a
folder, it will be interpreted as a relative path from the root folder - If you open the
workspace, it will be interpreted as a relative path from each root folder
- If you open the
"markdown-pdf.outputDirectory": "output",
- Relative path (home directory)
- If path starts with
~, it will be interpreted as a relative path from the home directory
- If path starts with
"markdown-pdf.outputDirectory": "~/output",
- If you set a directory with a
relative path, it will be created if the directory does not exist - If you set a directory with an
absolute path, an error occurs if the directory does not exist
markdown-pdf.outputDirectoryRelativePathFile
- If
markdown-pdf.outputDirectoryRelativePathFileoption is set totrue, the relative path set with markdown-pdf.outputDirectory is interpreted as relative from the file - It can be used to avoid relative paths from folders and workspaces
- boolean. Default: false
Styles options
markdown-pdf.styles
- A list of local paths to the stylesheets to use from the markdown-pdf
- If the file does not exist, it will be skipped
- All
\need to be written as\\(Windows)
"markdown-pdf.styles": [
"C:\\Users\\<USERNAME>\\Documents\\markdown-pdf.css",
"/home/<USERNAME>/settings/markdown-pdf.css",
],
- Relative path
- If you open the
Markdown file, it will be interpreted as a relative path from the file - If you open a
folder, it will be interpreted as a relative path from the root folder - If you open the
workspace, it will be interpreted as a relative path from each root folder
- If you open the
"markdown-pdf.styles": [
"markdown-pdf.css",
],
- Relative path (home directory)
- If path starts with
~, it will be interpreted as a relative path from the home directory
- If path starts with
"markdown-pdf.styles": [
"~/.config/Code/User/markdown-pdf.css"
],
- Online CSS (https://xxx/xxx.css) is applied correctly for JPG and PNG, but problems occur with PDF #67
"markdown-pdf.styles": [
"https://xxx/markdown-pdf.css"
],
markdown-pdf.stylesRelativePathFile
- If
markdown-pdf.stylesRelativePathFileoption is set totrue, the relative path set with markdown-pdf.styles is interpreted as relative from the file - It can be used to avoid relative paths from folders and workspaces
- boolean. Default: false
markdown-pdf.includeDefaultStyles
- Enable the inclusion of default Markdown styles (VSCode, markdown-pdf)
- boolean. Default: true
Syntax highlight options
markdown-pdf.highlight
- Enable Syntax highlighting
- boolean. Default: true
markdown-pdf.highlightStyle
- Set the current
highlight.jsstyle file name. Examples:github.css,monokai.css,base16/solarized-dark.css - file name list
- demo site : https://highlightjs.org/demo
"markdown-pdf.highlightStyle": "github.css",
Markdown options
markdown-pdf.breaks
- Enable line breaks
- boolean. Default: false
Emoji options
markdown-pdf.emoji
- Enable emoji. EMOJI CHEAT SHEET
- boolean. Default: true
Configuration options
markdown-pdf.executablePath
- Path to a Google Chrome, Microsoft Edge, or Chromium executable to run instead of the bundled Chromium
- See How is the Chromium browser selected? in the FAQ for how this setting interacts with installed browser detection and the managed Chromium download
- All
\need to be written as\\(Windows) - To apply the settings, you need to restart Visual Studio Code
"markdown-pdf.executablePath": "C:\\Program Files (x86)\\Google\\Chrome\\Application\\chrome.exe"
markdown-pdf.chromium.autoDownload
- Automatically download a managed Chromium when no installed browser is found
- boolean. Default: true
- When
false, Markdown PDF does not download Chromium and relies only on markdown-pdf.executablePath or an installed Google Chrome / Microsoft Edge / Chromium. Export fails if none is available. - See How is the Chromium browser selected? in the FAQ for the full resolution order
"markdown-pdf.chromium.autoDownload": true
Common Options
markdown-pdf.scale
- Scale of the page rendering
- number. Default: 1
"markdown-pdf.scale": 1
PDF options
- pdf only. puppeteer page.pdf options
markdown-pdf.displayHeaderFooter
- Enables header and footer display
- boolean. Default: true
- Activating this option will display both the header and footer
- If you wish to display only one of them, remove the value for the other
- To hide the header
"markdown-pdf.headerTemplate": "", - To hide the footer
"markdown-pdf.footerTemplate": "",
markdown-pdf.headerTemplate
- Specifies the HTML template for outputting the header
- To use this option, you must set
markdown-pdf.displayHeaderFootertotrue <span class='date'></span>: formatted print date. The format depends on the environment<span class='title'></span>: markdown file name<span class='url'></span>: markdown full path name<span class='pageNumber'></span>: current page number<span class='totalPages'></span>: total pages in the document%%ISO-DATETIME%%: current date and time in ISO-based format (YYYY-MM-DD hh:mm:ss)%%ISO-DATE%%: current date in ISO-based format (YYYY-MM-DD)%%ISO-TIME%%: current time in ISO-based format (hh:mm:ss)- Default (version 1.5.0 and later): Displays the Markdown file name and the date using
%%ISO-DATE%%"markdown-pdf.headerTemplate": "<div style=\"font-size: 9px; margin-left: 1cm;\"> <span class='title'></span></div> <div style=\"font-size: 9px; margin-left: auto; margin-right: 1cm; \">%%ISO-DATE%%</div>", - Default (version 1.4.4 and earlier): Displays the Markdown file name and the date using
<span class='date'></span>"markdown-pdf.headerTemplate": "<div style=\"font-size: 9px; margin-left: 1cm;\"> <span class='title'></span></div> <div style=\"font-size: 9px; margin-left: auto; margin-right: 1cm; \"> <span class='date'></span></div>",
markdown-pdf.footerTemplate
- Specifies the HTML template for outputting the footer
- For more details, refer to markdown-pdf.headerTemplate
- Default: Displays the {current page number} / {total pages in the document}
"markdown-pdf.footerTemplate": "<div style=\"font-size: 9px; margin: 0 auto;\"> <span class='pageNumber'></span> / <span class='totalPages'></span></div>",
markdown-pdf.printBackground
- Print background graphics
- boolean. Default: true
markdown-pdf.orientation
- Paper orientation
- portrait or landscape
- Default: portrait
markdown-pdf.pageRanges
- Paper ranges to print, e.g., '1-5, 8, 11-13'
- Default: all pages
"markdown-pdf.pageRanges": "1,4-",
markdown-pdf.format
- Paper format
- Letter, Legal, Tabloid, Ledger, A0, A1, A2, A3, A4, A5, A6
- Default: A4
"markdown-pdf.format": "A4",
markdown-pdf.width
markdown-pdf.height
- Paper width / height, accepts values labeled with units(mm, cm, in, px)
- If it is set, it overrides the markdown-pdf.format option
"markdown-pdf.width": "10cm",
"markdown-pdf.height": "20cm",
markdown-pdf.margin.top
markdown-pdf.margin.bottom
markdown-pdf.margin.right
markdown-pdf.margin.left
- Paper margins.units(mm, cm, in, px)
"markdown-pdf.margin.top": "1.5cm",
"markdown-pdf.margin.bottom": "1cm",
"markdown-pdf.margin.right": "1cm",
"markdown-pdf.margin.left": "1cm",
PNG, JPEG options
- png and jpeg only. puppeteer page.screenshot options
markdown-pdf.quality
- jpeg only. The quality of the image, between 0-100. Not applicable to png images
"markdown-pdf.quality": 100,
markdown-pdf.clip.x
markdown-pdf.clip.y
markdown-pdf.clip.width
markdown-pdf.clip.height
- An object which specifies clipping region of the page
- number
// x-coordinate of top-left corner of clip area
"markdown-pdf.clip.x": 0,
// y-coordinate of top-left corner of clip area
"markdown-pdf.clip.y": 0,
// width of clipping area
"markdown-pdf.clip.width": 1000,
// height of clipping area
"markdown-pdf.clip.height": 1000,
markdown-pdf.omitBackground
- Hides default white background and allows capturing screenshots with transparency
- boolean. Default: false
PlantUML options
markdown-pdf.plantumlOpenMarker
- Opening delimiter for the
@startuml/@endumlblock marker syntax. Change this if you want to use a different start marker. - Default: @startuml
markdown-pdf.plantumlCloseMarker
- Closing delimiter for the
@startuml/@endumlblock marker syntax. Change this if you want to use a different end marker. - Default: @enduml
markdown-pdf.plantumlServer
- Plantuml server. e.g. http://localhost:8080
- Default: http://www.plantuml.com/plantuml
- For example, to run Plantuml Server locally #139 :
plantuml/plantuml-server - Docker Hubdocker run -d -p 8080:8080 plantuml/plantuml-server:jetty
markdown-it-include options
markdown-pdf.markdown-it-include.enable
- Enable markdown-it-include.
- boolean. Default: true
mermaid options
markdown-pdf.mermaidServer
- mermaid server
- Default: https://unpkg.com/mermaid/dist/mermaid.min.js
math options
markdown-pdf.math.enabled
-
Enable math rendering via KaTeX for
$…$,$$…$$,\(…\),\[…\], and```mathfenced code blocks. -
Matches the behavior of VS Code's built-in Markdown preview.
-
Set to
falseto keep the raw$,\(,\[, and```mathtext (use this if your document contains$X$-style placeholders that should not be parsed as math). -
To disable math in a single document only, escape the
$as\$at the call site, or override this setting via YAML front matter:--- math: enabled: false --- -
boolean. Default: true
markdown-pdf.math.katex.macros
-
User-defined KaTeX macros passed to the KaTeX renderer.
-
Example:
{ "\\RR": "\\mathbb{R}" } -
Per-document macros can be supplied via YAML front matter, which takes precedence over this setting:
--- math: katex: macros: "\\RR": "\\mathbb{R}" --- -
Default: {}
Sanitize options
markdown-pdf.sanitize
- Sanitization mode for raw HTML in Markdown
"gfm": Strip GFM's disallowed tags and dangerous attributes (default)"gfm-allow-style": Same as"gfm"but keeps<style>elements"none": Disable sanitization (legacy behavior, not recommended)- Default:
"gfm"
"markdown-pdf.sanitize": "gfm",
FAQ
How can I change emoji size ?
- Add the following to your stylesheet which was specified in the markdown-pdf.styles
.emoji {
height: 2em;
}
Auto guess encoding of files
Using files.autoGuessEncoding option of the Visual Studio Code is useful because it automatically guesses the character code. See files.autoGuessEncoding
"files.autoGuessEncoding": true,
Output directory
If you always want to output to the relative path directory from the Markdown file.
For example, to output to the "output" directory in the same directory as the Markdown file, set it as follows.
"markdown-pdf.outputDirectory" : "output",
"markdown-pdf.outputDirectoryRelativePathFile": true,
Page Break
Please use either of the following to insert a page break.
<div class="page"/>
<div class="page"></div>
Why did my heading anchors change?
Starting with 2.0.0, Markdown PDF generates heading IDs using a custom markdown-it-named-headers implementation that follows GitHub-compatible VS Code slug generation. Compared to the previous implementation, the new slug generator preserves CJK characters and underscores while removing unsupported punctuation, which can cause existing internal anchors (e.g. #some-heading) to resolve differently.
If your Markdown relies on specific anchor strings (for example, a table of contents or cross-document links), re-check the generated anchors after exporting and update the links as needed.
Why did my syntax highlight style stop working?
Starting with 2.0.0, Markdown PDF uses highlight.js v11 (previously v9). Some style names from v9 have been renamed or removed. Markdown PDF maps legacy style names to current names where possible and shows a warning message when a configured style cannot be found. If no mapping is available, the extension falls back to tomorrow.css.
Please check the available styles and update your markdown-pdf.highlightStyle setting to a current style name.
Why is my front matter no longer parsed?
Starting with 2.0.0, Markdown PDF parses YAML front matter with a custom implementation instead of gray-matter. The new parser is stricter and rejects the following cases that the old parser may have accepted:
- Top-level YAML sequences (arrays) as front matter
- Front matter that does not parse into a plain object
- Malformed YAML structures
A valid front matter must be a YAML mapping (object) at the top level, for example:
---
title: My Document
"markdown-pdf":
displayHeaderFooter: true
---
BOM-prefixed files are still supported.
Why is my raw HTML being escaped or removed?
Earlier versions of this extension passed all raw HTML in Markdown through to the renderer without validation. Tags such as <script> and <iframe> could therefore execute during preview or PDF rendering, creating XSS-like risk when opening untrusted Markdown files (#411).
Starting with 2.1.0, raw HTML inside the Markdown body is sanitized by default per the GFM Disallowed Raw HTML extension. Behavior is controlled by markdown-pdf.sanitize:
| Mode | Behavior |
|---|---|
"gfm" (default) | Strip GFM's disallowed tags and dangerous attributes. Recommended when opening Markdown files authored by others. |
"gfm-allow-style" | Same as "gfm" but keeps <style> so you can embed CSS directly in a Markdown file to produce a self-contained PDF. Use only with content you trust — even without <script>, CSS can issue requests to attacker-controlled URLs via url(...) / @import / @font-face and leak information (known as CSS exfiltration). |
"none" | Disable sanitization. Legacy behavior. Not recommended. |
What "gfm" removes
Disallowed tags:
<title>, <textarea>, <style>, <xmp>, <iframe>, <noembed>, <noframes>, <script>, <plaintext>
Starting with 2.2.0, block-level <style> / <script> / <iframe> elements are removed together with their content, so CSS or JavaScript source no longer appears as literal text in the output. All other cases — the remaining disallowed tags, and inline occurrences of any disallowed tag — have their opening < escaped to < (content preserved as visible text). In 2.1.0, everything was escaped.
Attributes:
on*event handlers (onclick,onload, …)href/srcwhose value starts withjavascript:
Sanitize notification
Starting with 2.2.0, when elements are removed or attributes are stripped during an export, a notification summarizes what was neutralized: a toast with a "Show Details" button on manual export, or an entry in the "Markdown PDF" output channel on Auto convert.
Migrating from inline <style>
If you used to customize PDF layout by writing <style> directly inside a Markdown file, move that CSS into a .css file and reference it via markdown-pdf.styles. External stylesheets are loaded from your VS Code settings, not from the Markdown body, so they are not affected by sanitization.
Caveats for external CSS:
- CSS can still make outbound network requests through
@import url(...),background: url(...), attribute selectors withurl(...), etc. Only reference stylesheet files you trust. - When
markdown-pdf.stylesRelativePathFileistrue, the stylesheet path is resolved relative to the opened Markdown file. Be cautious about opening Markdown from untrusted locations that may ship a malicious sibling.css.
Sanitization scope
Sanitized:
- Raw HTML written inside the Markdown body (rendered via markdown-it's
html_block/html_inline) - Content pulled in by the Include feature (
:[label](https://github.com/yzane/vscode-markdown-pdf/blob/HEAD/path.md)) — it goes through the same renderer
Not sanitized:
- External CSS loaded via
markdown-pdf.styles(by design — user-configured trust boundary) - The extension's built-in stylesheets and HTML template
- HTML emitted by the extension itself (mermaid, highlight.js, emoji, PlantUML)
How do I troubleshoot a failed export?
Starting with 2.2.0, the extension writes detailed logs to the Markdown PDF output channel. Open it via View > Output and select Markdown PDF in the dropdown.
- When an export fails, the error toast describes what failed and, for common causes (Chromium launch failure, locked output file, permission denied, disk full, invalid output directory), adds a one-line hint. The Show Details button opens the log with the full context, error message, and stack trace.
- Every export logs an environment snapshot and conversion context block at the start, so the log alone is usually enough to see what happened.
- Run
Markdown PDF: Output Diagnosticsfrom the Command Palette to print environment and configuration diagnostics: extension version, VS Code / OS information, the Chromium path and how it was selected, and relevant settings. Home directory paths are masked.
When filing an issue, please attach the output of Markdown PDF: Output Diagnostics and the log shown by Show Details.
How is the Chromium browser selected?
Markdown PDF resolves a Chromium-based browser in the following order:
- The path specified in markdown-pdf.executablePath, if the file exists.
- An installed browser on your system. Google Chrome (stable) is detected via @puppeteer/browsers at its standard OS install location; Microsoft Edge and Chromium are probed at the fixed paths listed below.
- A managed Chromium that Markdown PDF automatically downloads on first use.
The first match wins. The per-OS scan order for installed Edge and Chromium is:
Windows
- Google Chrome (stable install, detected via
@puppeteer/browsers) %LOCALAPPDATA%\Microsoft\Edge\Application\msedge.exe%LOCALAPPDATA%\Chromium\Application\chrome.exe%PROGRAMFILES%\Microsoft\Edge\Application\msedge.exe%PROGRAMFILES%\Chromium\Application\chrome.exe%PROGRAMFILES(X86)%\Microsoft\Edge\Application\msedge.exe%PROGRAMFILES(X86)%\Chromium\Application\chrome.exe
macOS
- Google Chrome (stable install, detected via
@puppeteer/browsers) /Applications/Chromium.app/Contents/MacOS/Chromium/Applications/Microsoft Edge.app/Contents/MacOS/Microsoft Edge
Linux
- Google Chrome (stable install, detected via
@puppeteer/browsers) /usr/bin/chromium-browser/usr/bin/chromium/usr/bin/microsoft-edge/usr/bin/microsoft-edge-stable
Where is Chromium downloaded?
If no installed browser is found, Markdown PDF downloads a managed Chromium on first use. The download is stored under the extension's VS Code global storage directory:
| OS | Download path |
|---|---|
| Windows | %APPDATA%\Code\User\globalStorage\yzane.markdown-pdf\ |
| macOS | ~/Library/Application Support/Code/User/globalStorage/yzane.markdown-pdf/ |
| Linux | ~/.config/Code/User/globalStorage/yzane.markdown-pdf/ |
If you use VS Code Insiders or VSCodium, the base path changes accordingly (for example Code - Insiders or VSCodium instead of Code).
During the download, Installing Chromium is shown in the status bar.
Which Chromium build is downloaded?
Markdown PDF tries to fetch the latest Chrome Stable build id from the Chrome for Testing API. The latest build id is checked once per VS Code session: when a newer Chrome Stable has been released, Markdown PDF downloads the new build at the next export and removes the previously cached build. Within the same VS Code session the build id is memoized — restart VS Code to pick up a freshly released build.
If the API is unreachable, it falls back in this order:
- The most recently cached build under the global storage directory shown above
- The build id pinned by the bundled
puppeteer-core(last-resort fallback)
Disabling the auto-download
Set markdown-pdf.chromium.autoDownload to false to skip the download entirely. Markdown PDF will then rely only on markdown-pdf.executablePath or an installed Google Chrome / Microsoft Edge / Chromium, and export will fail with an error if none is available.
Known Issues
markdown-pdf.styles option
- Online CSS (https://xxx/xxx.css) is applied correctly for JPG and PNG, but problems occur with PDF. #67
Change Log
See CHANGELOG.md for the full change history.
License
MIT
Sponsor
If you find Markdown PDF useful, you can support continued development via GitHub Sponsors.