Introducing Agents.·Read
OverviewAgentsWorkflowsExecutionDeveloper ToolsSecurityLibrary
Updated Sep 21, 2026

Utilities

Fetch article URLs and keep the clean body text

Article body text plus title, author, date, and language from a URL list.

Agent
Extract one articleExtract a list of URLsKeep table text

overview

What trafilatura does

You have an article URL or a file of URLs, and the next node needs the body text, not the chrome. Set --url for one page or --urls-file for a list. The wrapper fetches HTML, runs Trafilatura, and writes articles.jsonl plus results.jsonl. Each row keeps title, author, date, language, and the cleaned body.

--min-chars drops extractions shorter than N characters. Default is 200, which can drop example.com. Set --min-chars 1 when you want a short public page. --max-chars truncates a long body. --concurrency and --timeout bound the fetch. --include-tables and --fail-empty are booleans; leave them disconnected unless you mean to send the flag.

trafilatura reads the HTML the worker can download. It does not render JavaScript and it does not bypass a login wall. Use browser-fetch when the page is empty without Chromium. Use rss-read when the source is a feed, not a page.

source github.com/adbar/trafilatura

use cases

Where trafilatura fits

Extract one article

Set --url and --min-chars 1 if the page is short. articles.jsonl holds title and body.

Extract a list of URLs

Pass --urls-file. --max-urls caps the batch.

Keep table text

Connect --include-tables only when you want table cells in the body.

Fail if nothing extracted

Connect --fail-empty when an empty articles.jsonl should fail the node.

reference

trafilatura inputs and flags

12 inputs
NameTypeFlagDescription
urlSTRING--urlSingle article URL.
urls-fileFILE--urls-fileFile of article URLs.
min-charsSTRING--min-charsDrop extractions shorter than N characters. Default 200.
inFOLDER--indirUpstream folder holding URLs.

Showing key inputs. trafilatura exposes 12 inputs in total.

Full flag reference (12 inputs)
NameTypeFlagDescription
inFOLDER--indirUpstream folder holding URLs (JSONL with a url field, JSON array, or one URL per line)
urlSTRING--urlSingle article URL
urls-fileFILE--urls-fileFile of article URLs (JSONL / JSON array / one per line)
urls-textSTRING--urls-textInline URL list. Avoid unquoted '|' in values.
concurrencySTRING--concurrencyParallel fetches (default: 8)
timeoutSTRING--timeoutPer-URL download timeout in seconds (default: 25)
min-charsSTRING--min-charsDrop extractions shorter than N characters (default: 200)
max-charsSTRING--max-charsTruncate body text at N characters (default: 20000)
max-urlsSTRING--max-urlsCap URLs processed this run (0 = no cap)
user-agentSTRING--user-agentOverride HTTP User-Agent
include-tablesBOOLEAN--include-tablesKeep table content in the extracted text
fail-emptyBOOLEAN--fail-emptyExit non-zero when no article was extracted

example

Run trafilatura

trafilatura · command
trafilatura-fetch --url https://example.com --min-chars 1 --outdir /hive/out
sample output
{"url": "https://example.com", "title": "Example Domain", "lang": "en", "text": "Example Domain\nThis domain is for use in documentation examples without needing permission. Avoid use in operations.\nLearn more", "text_chars": 127}

guidance

Choosing trafilatura

Use trafilatura when the page is static HTML and you want the article body. Use browser-fetch when the HTML is empty without Chromium. Use rss-read when the source is a feed.

browser-fetch

Renders the URL in headless Chromium. Use it when trafilatura gets an empty or interstitial page.

rss-read

Parses a public feed. Use it when you have an RSS URL, not an article URL.

wget

Downloads the raw response. It does not extract title and body.

faq

trafilatura questions

Set --url. Add --min-chars 1 if the default 200-character floor would drop the page.

related

More Utilities tools

  1. browser-fetch

    Headless Chromium render: requested URL, final URL, status, and title next to the page.

  2. cewl

    Spider a URL and return a wordlist for password crackers.

  3. csvkit

    SQL on one CSV. The table name is the file stem.

  4. dnsgen

    Wordlist and mined-word subdomain permutation.

  5. duckdb

    In-process SQL on CSV, JSON, and Parquet. No server.

  6. edge-tts

    Microsoft Edge neural speech from text or a script file. No API key.

Run trafilatura yourself

https://example.com feeds trafilatura. articles.jsonl is from a completed run.

Open trafilatura in the libraryBuild a workflow with it

Facts on this page come from the live Trickest tool library.