Help Center

Getting started

1. Open the page you want to capture.
2. Select Single Page, Site Tree, Sitemap, or Selection in the extension popup.
3. For Single Page, choose Markdown or HTML snapshot in Save page as. Choose Preview to review the source or Download to save immediately. Selection and Site Tree remain Markdown; Sitemap downloads its evidence bundle.
4. Open saved Markdown or HTML in Web2Markdown Editor. HTML snapshots retain selected computed styles and local images, with scripts, forms, frames, media, and external stylesheets omitted. They are rendered-page snapshots rather than complete offline archives or original server source.

First-run checklist

1. Pin Web2Markdown Capture to the Chrome toolbar.
2. Open Settings and choose the capture folder and local or remote image handling.
3. Capture one ordinary HTTP or HTTPS page and confirm the Markdown and final receipt appear in the selected capture root.
4. Install Web2Markdown Editor and confirm the new capture opens from the watched folder.
5. Open the persistent side panel for review and progress. Chrome controls its width; drag the inside edge or choose Open wide capture workspace for a full tab bound to the source page.
6. Enable the local agent bridge only if this computer has the packaged native host installed.

What's New in 1.7.3

Capture reusable webpage elements for your local Artifact Library. Choose Capture Element, select a button, card, text block, or image, review it, then save the artifact file. In desktop 2.8.5 or later, open Artifact Library to import, name, tag, edit, insert, or export it. Captures preserve selected styles and available images; website scripts and behavior are excluded.

The library keeps your original capture separately from your edits. Change text, colors, fonts, spacing, borders, and rounded corners in the desktop editor. Missing or unsupported resources are reported; a capture is not a complete recreation of a site's CSS or interactive states.

Earlier release: 1.7.2

Version 1.7.2 adds a 680-pixel popup, a wide capture workspace, and Single Page HTML snapshots. HTML source review is inert; open the saved file in the desktop editor for visual preview. HTML receipts require desktop 2.8.4 or later for Capture Inbox detection.

Earlier releases: 1.7.0 and 1.7.1

Fixed duplicate right-click menu errors after an extension reload or update. Menu setup now replaces old entries once and records callback failures in diagnostics. Reload the extension to apply the fix, then clear historical errors before checking for new ones. Captures and settings are unchanged.

• Interactive human-session robots exemption: Captures taken while browsing as an actual human visitor do not block or abort on robots.txt Disallow directives or errors by default. Strict compliance remains available in Settings.
• Multi-root destination registry & quick chips: Save captures across multiple direct folders on any drive and switch between them or Downloads with one click directly in the popup.
• One-click permission renewal: When browser restarts suspend directory permissions, a renewal badge appears in the popup and options for instant re-authorization.
• Incremental multi-highlight clipping: In Selection mode, + Append Highlight lets you accumulate multiple excerpts into your active draft note with a clean separator before saving.
• Dynamic path & frontmatter templating: Single and double curly brace tokens support {{title}}, {{domain}}, {{date}}, {{year}}, {{month}}, {{day}}, and {{captureId}}.
• Markdown Cleaner & tracking parameter removal: Automatically strips marketing tracking query parameters (such as utm_*, fbclid, gclid) from URLs and Markdown links, and collapses excessive blank lines for clean, distraction-free captures.
• Each capture gets its own operation folder so repeated and simultaneous captures stay separate.
• Preview supports split-file selection, editable source, passive rendered output, and bounded local draft recovery with cross-panel conflict protection.
• Site Tree history supports explicit interrupted-job recovery, failed-page retry, and partial saves.
• System, light, and dark appearance now work across the popup, side panel, Settings, and Help.
• Connection test captures help pair the extension with the desktop Capture Inbox without commands.
• Optional page SEO evidence saves observed DOM facts and a readable local summary.
• Capture actions automatically use their normal folder under Downloads/Web2Markdown when Chrome cannot use a chosen custom folder.
• The popup reports the actual saved folder and explains when automatic fallback was used.
• Expected capture failures now stay in the popup and Diagnostics instead of appearing as Chrome extension faults.
• Failure details retain their capture context and actionable message without exposing raw local paths.
• Sitemap mode continues to save raw sitemap evidence without opening or converting listed pages.
• The approved GDK-branded Web2Markdown artwork remains consistent across Chrome and the desktop editor.

Capture modes

Single Page

Extracts the readable content from the active HTTP or HTTPS page into one Markdown document.

Site Tree

Follows same-site links within the configured depth, page cap, path pattern, pacing, and robots rules. It can also use the site's sitemap as an optional bounded discovery source.

Sitemap

Downloads the readable sitemap documents, a deterministic URL inventory, and a completion receipt without opening or converting the pages listed inside them.

Selection

Captures the text currently selected on the page. In Selection mode with an active draft, use + Append Highlight to collect multiple excerpts into the same note before downloading.

Preview and edit

Preview generates Markdown without saving files to disk. Select any split file before copying or editing its source. Rendered view shows sanitized output without fetching remote images or running captured scripts. Save writes every file in this preview, including your edits.

Preview drafts and capture recovery

A preview stays bound to its original source URL, capture settings, and file set even when you browse elsewhere. The displayed destination and matched Site Profile explain the capture. Use the profile selector for an automatic match, no profile, or a one-capture override.

The extension retains one local draft for up to 24 hours, 50 files, and 2 MiB of UTF-8 data. These drafts can contain private page content. Clear draft removes the retained preview, not previously saved files. A storage error leaves the current preview open and explains that it is not protected after closing.

The popup and side panel share a revision-checked draft. If another panel changed it, local edits are not silently overwritten. Resolve the conflict before relying on stored recovery. A failed save leaves the preview available for correction and retry.

Expired drafts and jobs are cleaned when their storage is next accessed, not by a background timer while Chrome is closed. Use Clear draft or Clear inactive history for immediate removal.

Capture history retains up to 30 Site Tree jobs for 7 days within a 6 MiB local budget. Interrupted work never resumes automatically. Choose Resume for remaining queued URLs, Retry failed for failed pages, or Save partial to publish retained successful pages. Clearing inactive history removes recovery content, not files already on disk.

Page SEO evidence

Enable the one-capture page evidence option to save page-evidence.json and seo-summary.md beside your Markdown. Open either from the desktop Capture Inbox. The option observes rendered page title, description, language, canonical and robots metadata, alternate languages, headings, links, images, and a JSON-LD inventory.

Collection is bounded to 20,000 inspected elements, 200 links, 100 images, 100 headings, and 20 JSON-LD blocks per page. Large or malformed evidence produces warnings or truncation rather than unbounded output. Scripts and linked pages are not executed or fetched by the evidence viewer.

These are observed DOM facts, not an SEO score, indexing verdict, HTTP-header audit, link-status check, or performance measurement. Page captures and their evidence stay together in one folder; separate Save As prompts are disabled for evidence bundles. Use Sitemap mode for sitemap-only output without opening its listed pages.

Files and assets

Local image handling downloads supported images into an assets folder and rewrites Markdown image paths. Remote image mode leaves original URLs unchanged.

• YAML frontmatter records title, source and canonical URLs, stable capture ID, capture kind, capture time, evidence counts, and the matched Site Profile when enabled.
• Generated article content is enclosed in explicit boundaries so the desktop editor can refresh that region without replacing user content outside it.
• Split by headings creates multiple Markdown files at H2 or H3 boundaries.
• Browse offers a system folder picker for selecting or creating a capture root on any available drive. Chrome exposes only the chosen folder's name to the extension, not its full path.
• Manage multiple saved capture directories in Settings and quickly switch between them and Downloads using the popup's destination quick chips.
• When browser restarts suspend directory permissions, a one-click renewal badge appears in the popup and Settings for immediate re-authorization.
• The Downloads mode also lists recent folders under Downloads/Web2Markdown and accepts a safe relative subfolder.
• If a chosen folder is unavailable when an action starts, that capture uses its normal folder under Downloads/Web2Markdown. The popup names the destination and explains the fallback. Your chosen-folder setting remains available to renew later.
• Capture bundles keep Markdown, assets, and receipts together automatically; a separate Save As prompt is unavailable. Use Browse to change the capture root. Show file in folder applies only to Downloads mode.
• A final capture.json receipt is written after Markdown and asset downloads reach terminal states.
• Each new transaction uses a unique capture-<run ID> subfolder. The stable source ID still lets the desktop compare recaptures without overwriting earlier bundles.

Settings reference

Every Settings field has a ? control. Hover over it, click it, or focus it with Tab to read an explanation. Press Escape to close a focused tooltip.

Markdown output

Heading appearance

ATX uses # symbols and supports all heading levels. Setext uses underlines for only the first two levels.

Bullet symbol

Changes the source character before unordered list items. Dash, asterisk, and plus render the same.

Code block appearance

Fenced blocks use triple backticks. Indented blocks use four spaces. Fenced blocks are generally easier to edit.

Where captured images live

Local assets support offline captures. Remote URLs save space but continue to depend on the source website.

Clean tracking parameters

Automatically removes marketing telemetry parameters (such as utm_*, fbclid, and gclid) from captured page URLs and links, keeping Markdown links pure and uncluttered.

Collapse consecutive blank lines

Collapses three or more consecutive blank lines down to a clean two-line paragraph separator.

YAML frontmatter

A structured block at the top of the Markdown containing source, capture, author, and quality information.

Site Tree

Respect robots.txt

When disabled (default), interactive captures taken while browsing as an actual human visitor bypass robots.txt disallow directives. When enabled, Site Tree strictly follows robots.txt disallow rules and halts on robots fetch failures.

Link level 1

Captures the starting page and pages linked directly from it.

Link level 2

Also follows links found on those directly linked pages. This is the recommended default.

Link level 3

Follows one more level. It can find substantially more pages and therefore takes longer and creates more files.

Page path filter

An optional advanced pattern that keeps Site Tree inside part of a site. Leave it blank for all same-site paths. ^/docs(?:/|$) keeps only /docs and pages below it.

Find pages from the site's sitemap

Checks absolute Sitemap records in robots.txt. If none are listed, it tries /sitemap.xml. Sitemap pages still have to match the starting site, URL path filter, robots rules, file-type rules, and Max pages limit.

Split by headings

Lets one long web page become several Markdown files at H2 or H3 headings.

Export and Site Profiles

Download mode supports Markdown files, local assets, multiple files, and receipts. Clipboard mode copies a one-file capture as text. Site Profiles are optional rules for sites you capture repeatedly; they can match pages, select or remove page areas with CSS selectors, organize output with safe folder and filename patterns, and add simple metadata.

Site Profiles

Create bounded, declarative profiles in Settings for sites you capture repeatedly. The first enabled host or URL pattern that matches supplies the default page or selection mode.

• An include selector can narrow extraction to one content region.
• Up to 10 exclude selectors can remove recurring page elements.
• Folder and filename templates support tokens such as {title}, {domain}, {date}, {year}, {month}, {day}, and {captureId} (also accepted in double curly braces like {{title}}).
• Additional frontmatter uses simple key-value fields. Provenance fields cannot be overridden.
• Profiles cannot run scripts, traverse parent folders, or use absolute output paths.

Site Tree capture

Site Tree is deliberately bounded. It stays on the starting origin, supports a maximum depth of 3 and maximum page count of 50, waits between pages, and respects crawl safety boundaries. When you capture interactively as a human browsing user, robots.txt disallow rules are not enforced by default. Optional sitemap discovery reads at most 10 sitemap files, two index levels, 1 MB per file, and 2,000 URL records before the normal 50-page capture cap is applied.

• Interactive captures taken as a human visitor do not block or abort on robots.txt disallow directives unless "Respect robots.txt" is enabled in Settings.
• Use a URL path pattern such as ^/docs(?:/|$) to keep a crawl inside one site section.
• Pages that fail independently are reported without discarding successful pages.
• HTTP 429, server errors, unreadable robots responses, and network failures stop the crawl with an actionable error when robots compliance is enabled.
• A missing, malformed, oversized, or unsupported sitemap does not bypass safety rules. Site Tree continues with normal links and records sitemap warnings in the completion receipt.
• Use Cancel to stop new crawl work. Already completed downloads remain on disk.

Sitemap-only capture

Choose Sitemap when you need the site's sitemap evidence without downloading page content. The capture checks absolute Sitemap records in robots.txt and falls back to /sitemap.xml only when none are declared.

• Raw readable sitemap documents are saved as numbered XML or text files.
• sitemap-inventory.json records source URLs, redirects, formats, discovered URLs, counts, and warnings.
• capture.json is written last so tools can detect a complete bundle.
• The same 1 MB per file, 10-file, two-index-level, and 2,000-URL limits apply.
• The listed pages are not opened, rendered, or converted to Markdown.

Desktop workflow

Web2Markdown Editor opens Markdown and a comprehensive catalog of UTF-8 text formats without changing their source format. Markdown documents also support preview, outline, formatting tools, templates, and rich export.

1. Add the same capture root in the desktop Capture Inbox. Chrome exposes a folder name, not its full path; compare full paths in your system file browser.
2. Use the registered Windows Markdown association, File > Open, drag and drop, Quick Open, or the watched Downloads folder.
3. Use the desktop Capture Inbox to inspect source identity, receipt state, duplicates, and asset health across configured folders.
4. Use conflict comparison and File History when a file changes outside the editor.
5. Export Markdown to PDF, HTML, plain text, or DOCX from the editor.

Select Create connection test capture in extension Settings or the popup. Find that exact test ID in the desktop Capture Inbox and confirm it there. Saving a test proves extension output only; the desktop must independently detect it. Completed bundles arrive as one notification with Open primary, Open selected, and View Inbox actions. Automatic opening is limited to a single-file capture when the editor is focused and empty.

Agent access

Agent Mode adds a visible page control for trusted clicks. The optional local bridge lets a trusted local process request capture of the active Chrome page.

1. Install the Windows desktop app. In desktop Settings > Optional Chrome automation, paste the exact extension ID and select Register installed host. This current-user registration needs no command or administrator access; company policy can still prohibit it.
2. Use Capture for agent button in the extension popup to show or hide the floating page control. The choice persists and updates supported pages immediately.
3. Select Enable bridge, grant the optional nativeMessaging permission, and reload the extension if prompted.
4. Keep the target HTTP or HTTPS tab foregrounded before invoking the local CLI or MCP tool.

After installing or reloading the extension itself, reload an older target tab once so its content script is current. Normal popup toggle changes do not require another reload.

Troubleshooting

The page cannot be captured

Confirm the active tab uses HTTP or HTTPS, reload the page, and retry. Chrome internal pages and extension pages cannot be captured.

Selection is empty

Select page text before opening the extension popup. Some embedded viewers do not expose selection to extensions.

Images remain remote

Use Download to folder with local image handling. Clipboard output and one-file Save As cannot guarantee a stable sibling asset folder.

The selected folder no longer works

Chrome can revoke or forget a saved folder grant. The current capture automatically saves under Downloads/Web2Markdown and the popup names the folder. Use Settings > Browse later if you want to renew the custom destination.

The bridge says it is not running

Open Settings, select Enable bridge, confirm it is connected to com.web2markdown.agent, then keep a normal web page foregrounded.

Site Tree stops before downloading

Review the displayed robots, rate-limit, network, or validation error. Single Page and Selection remain available.

Diagnostics

Open Settings and select Copy diagnostics to collect the extension version, capture-folder mode, bridge state, crawl state, and recent sanitized service-worker errors.

The report excludes page content, captured Markdown, credentials, and full local filesystem paths. Chrome may stop an idle extension service worker, so its in-memory recent-error list resets when that worker restarts.

Privacy and permissions

Web2Markdown is local-only. It has no cloud account, analytics service, remote content API, or store submission.

• Page access is used to extract content from pages you choose to capture.
• Downloads permission writes Markdown and local assets.
• Storage keeps local settings, bounded preview drafts, and capture recovery history. Drafts and job checkpoints can contain captured private content; clear them from the popup or side panel when no longer needed.
• A direct-folder selection stores a browser-managed directory handle in extension-local IndexedDB. The extension does not receive or report the folder's full filesystem path.
• Tabs and scripting support active-page extraction and bounded Site Tree work.
• nativeMessaging is optional and used only after you enable the local agent bridge.
• sidePanel keeps the extension interface beside a page when you explicitly open it.

Shortcuts

About Web2Markdown Capture

Web2Markdown Capture is a GDK Digital tool and the browser half of Web2Markdown. The Electron desktop editor opens captured files from disk and does not receive page content directly from the extension.