Web2Markdown Settings

Markdown Output

ATX writes headings with # symbols and supports every heading level. Setext underlines only the first two heading levels. ATX is the most widely compatible choice.
Chooses the character placed before unordered list items. Dash, asterisk, and plus render the same in Markdown viewers; this changes source appearance only.
Fenced blocks wrap code in triple backticks and are easier to copy and edit. Indented blocks use four leading spaces. Fenced is usually clearer and more reliable.
Download to assets saves copies beside the Markdown so the capture can work offline. Keep remote URLs makes a smaller capture that still depends on the original website.
Adds a structured block containing the page title, source URL, capture date, author when available, capture ID, and quality counts. Markdown tools and agents can read these details without changing the article text.
Creates a linked list from the headings found in the captured article. Use it for long documents; leave it off for short pages or when another tool generates its own contents list.
Removes marketing analytics tags (such as utm_source, utm_medium, fbclid, and gclid) from captured Markdown links while preserving essential query parameters, anchor hashes, and target destinations.
Replaces runs of 3 or more consecutive newlines with 2 newlines to preserve clean paragraph spacing without large empty gaps left behind by ad containers or spacers.

Site Tree Capture

Site Tree starts with the current page and follows links on the same website. These defaults control how much of the site it visits.

A link level is one click farther from the page where you start. Level 1 captures the start page and pages linked directly from it. Higher levels find more pages, take longer, and can create many more files.
Level 2 is a practical default for a documentation section or small website.
Optional regular expression matched against the path after the domain. For example, ^/docs(?:/|$) keeps /docs and pages below it while ignoring /blog and /account. Leave blank to allow every same-site path.
Leave blank unless you need to keep a crawl inside one section of a website.
When Site Tree starts, the extension checks Sitemap records in robots.txt. If none are listed, it tries /sitemap.xml. Eligible same-site URLs join the normal crawl, but Max pages, the path filter, robots rules, pacing, and cancellation still apply.
Useful for pages listed for search engines but not linked near the page where you start.
When enabled, the popup starts with Split by Headings turned on. Each captured web page can become several Markdown files at its H2 or H3 headings. This is useful for large guides but creates more files.
When disabled (default), captures behave as a human visitor browsing the site directly. Enable this only when running unattended automated crawls where robots.txt Disallow directives must be honored.
Disabled by default because you are manually capturing the page as an interactive user, not a robot crawler.

Export

Download to folder saves Markdown, local images, and a completion receipt. Copy to clipboard copies a one-file capture as text; multi-file captures still download because they cannot fit in one clipboard item.
This is the main folder that contains each capture's page folder, Markdown files, assets, and receipt. The default is Downloads/Web2Markdown. Browse can select or create a folder on another drive. If Chrome cannot use that folder for an action, the action saves under Downloads/Web2Markdown and reports the fallback.
Choose any available drive, or keep Downloads/Web2Markdown. An unavailable chosen folder falls back to Downloads automatically for that capture.
Captures use a unique folder so Markdown, local assets, and the completion receipt stay together. A separate Save As prompt could break desktop detection. Choose your capture root with Browse instead, or copy preview text for use elsewhere.
Capture bundles save automatically inside the selected root. Choose Browse to change that root.
After Chrome finishes a download, opens Explorer or Finder and highlights the completed file. This applies to Downloads mode. Direct-folder mode already uses a folder you selected through the system picker.
Automatically opens a Chrome download in Explorer/Finder. Direct folders are already chosen through the system picker.

Site Profiles

Apply bounded selectors, folders, filenames, and metadata to matching sites. Profiles never run code.

New site profile An enabled profile automatically supplies capture defaults when its host or URL pattern matches the active page. Disable it to keep the profile saved without applying it. A private label that helps you recognize this rule in Settings and capture metadata, such as Company Knowledge Base or Build Rite Projects. Enter a hostname or full URL with optional * wildcards. *.example.com matches its subdomains. https://example.com/docs/* matches only that site's docs section. The first enabled matching profile is used. Page captures the readable content of the whole active page. Selection captures only text you selected before opening the extension. This chooses the starting mode; you can still change it in the popup. Optional. Restricts extraction to the first page element matching a CSS selector, such as main article, #documentation, or .knowledge-base. Leave blank to let readability detection choose the main content. Removes repeated page elements before conversion, such as .newsletter, nav.sidebar, or aside.related. Enter up to 10 valid CSS selectors, one per line. Organizes matching captures beneath your selected capture root. For example, Research/{domain}/{date} creates nested folders using the page's domain and capture date. Absolute paths and parent-folder traversal are rejected. Builds the filename from safe tokens, such as {title}-{date}. Do not include .md; the extension adds it. If a filename already exists, Web2Markdown creates a numbered copy instead of overwriting it. Adds your own simple fields to the YAML block, such as project=Research or status=inbox. Built-in source and capture identity fields cannot be replaced. Up to 20 fields are allowed.

Available tokens: {title}, {domain}, {date}, and {captureId}. A maximum of 25 profiles, 10 exclude selectors, and 20 metadata fields is enforced.

Agent Access

Adds a movable-looking but fixed page overlay that a browser agent can click to capture the current signed-in page. Turn it off whenever it covers page controls or you are not using agent capture.
Adds a visible button that a browser agent can use to capture the current signed-in page. Each capture writes Markdown and a capture.json completion receipt under Downloads/Web2Markdown/Agent.
Allow local agent tools to request a captureConnects the extension to the Web2Markdown desktop native host on this computer. A trusted local CLI or MCP tool can request capture of only the active HTTP or HTTPS tab. It cannot supply another URL, crawl settings, scripts, or headers.
Requires the Web2Markdown desktop native host. The bridge can capture only the active HTTP or HTTPS tab and accepts no URL parameters.

Workspace and Support

Match system follows your operating system. Light and Dark override it for capture, Settings, and Help. This choice saves immediately.

Connect your capture folder to the editor

In the desktop Capture Inbox, add the same folder selected here. Folder names alone cannot prove they are the same location. Save a test and match its ID in the editor. Ordinary captures do not require the automation bridge.

Use the persistent side panel for longer capture work. Diagnostics contain versions and component health, but never captured page content.

Select Copy diagnostics to refresh this report.