HTML to Markdown & Reader Mode Extractor
AIDEVELOPER TOOLSNEWS
Independent tool. Trademarks referenced on this page are the property of their respective owners and are used only to identify the public website this tool reads. No affiliation or endorsement.
HTML to Markdown & Reader Mode Extractor
Convert any URL or raw HTML to clean markdown, plaintext, and reader-mode metadata. A FREE, one-call primitive for LLM pipelines, text-processing workflows, and content extraction.
What it does
Pass a URL or raw HTML, get back:
- Markdown — clean, ready for LLM ingestion
- Plaintext — stripped of all markup
- Reader-mode metadata — title, byline, published date, language, word count, reading time
- Images —
[{src, alt}]list - Links —
[{href, text}]list
All in a single run. No crawler config needed.
Extraction modes
| Mode | Best for | How it works |
|---|---|---|
readability (default) |
News articles, blog posts, documentation | Mozilla Readability strips navigation, ads, and boilerplate; then Turndown converts the clean article HTML to markdown |
turndown |
Preserving the full page structure | Direct Turndown conversion of the full HTML without article extraction — no stripping |
trafilatura |
Noisy pages, forums, user-generated content | Cheerio removes script/style/nav/header/footer elements, targets the main content area, then converts to markdown |
Input
| Field | Type | Default | Description |
|---|---|---|---|
inputs |
array | — | Required. Array of {url?: string, html?: string} objects |
mode |
string | readability |
Extraction mode: readability / turndown / trafilatura |
output |
string | markdown |
Output format: markdown / plaintext / both |
includeImages |
boolean | true |
Include images list in output |
preserveLinks |
boolean | true |
Include links list and preserve hyperlinks in markdown |
renderJs |
boolean | false |
Fetch JS-rendered pages via browser (requires more memory) |
maxItems |
integer | 15 |
Max inputs to process |
Example input
{
"inputs": [
{ "url": "https://en.wikipedia.org/wiki/Markdown" },
{ "html": "<html><body><h1>Hello</h1><p>World</p></body></html>" }
],
"mode": "readability",
"output": "markdown"
}
Output
One dataset record per input:
| Field | Type | Description |
|---|---|---|
input |
string (JSON) | Original input {url?, html?} |
title |
string | Extracted page title |
byline |
string | Author / byline |
publishedAt |
string | Published timestamp (ISO 8601, when detectable) |
language |
string | Detected language code (ISO 639-1) |
markdown |
string | Markdown output |
plaintext |
string | Plain-text output |
wordCount |
number | Word count |
readingTimeMin |
number | Estimated reading time (at 240 wpm) |
images |
string (JSON) | [{src, alt}] image references |
links |
string (JSON) | [{href, text}] hyperlinks |
mode |
string | readability / turndown / trafilatura |
finalUrl |
string | URL after redirects (URL inputs only) |
status |
string | success / timeout / error |
errorMsg |
string | Error message on failure |
Example output record
{
"input": "{\"url\":\"https://en.wikipedia.org/wiki/Markdown\"}",
"title": "Markdown",
"byline": "Contributors to Wikimedia projects",
"publishedAt": "2005-08-09T19:56:00Z",
"language": "en",
"markdown": "# Markdown\n\nMarkdown is a lightweight markup language...",
"wordCount": 2459,
"readingTimeMin": 11,
"images": "[{\"src\":\"/static/images/icons/enwiki-25.svg\",\"alt\":\"Wikipedia\"}]",
"links": "[{\"href\":\"/wiki/Main_Page\",\"text\":\"Main page\"}]",
"mode": "readability",
"finalUrl": "https://en.wikipedia.org/wiki/Markdown",
"status": "success",
"errorMsg": null
}
Pricing
FREE. Compute-billed only — you pay for the Apify platform run time, not per-record. Ideal as a preprocessing step before paid downstream actors (e.g., translation, summarization).
Use cases
- LLM context preparation — fetch and clean web pages before passing to GPT/Claude/Gemini
- RAG pipeline ingestion — extract clean text from arbitrary URLs at scale
- Content archiving — convert live pages to portable markdown
- Article extraction — strip ads and navigation from news/blog content
- Text analytics — feed clean plaintext into NLP models
Technical notes
- HTTP fetch via got-scraping (Chrome TLS fingerprint, follows redirects)
- Concurrency: up to 20 inputs processed in parallel per worker
- Memory: 512 MB default; increase to 1024 MB if
renderJs: true - Timeout: 30 seconds per URL fetch
Further reading: Technical Data Utilities: DNS Audits, SSL Checks, Satellite Elements, and Bulk Parsing
