Zendesk Help Center Scraper

Extract every article from any Zendesk Help Center as clean plain text, with sections, categories, labels and vote counts. Built for RAG pipelines, AI support agents, knowledge base migration and competitive research.

Thousands of companies run their documentation on Zendesk Guide: Discord, Figma, Zendesk itself, and countless others. Every one of them exposes a public API. Paste the help centre URL and go. No API key, no login, no configuration.

What you get

Field	Description
id	Zendesk article ID
title	Article title
url	Public article link
body	Full article as clean plain text
wordCount	Words in the body
section / sectionId	Section name, not just an ID
category / categoryId	Category name, not just an ID
labels	Article labels
locale	Article language
createdAt / updatedAt / editedAt	Timestamps
voteSum / voteCount	Reader feedback
promoted / outdated	Zendesk article flags
helpCenter	Source help centre
scrapedAt	ISO timestamp

Why this one

Locale is detected for you. Zendesk redirects /api/v2/help_center/articles.json to a locale-specific path, and scrapers that ignore this get a redirect instead of data. This Actor resolves the locale on the first call and then goes direct.

Category and section names, not IDs. Zendesk returns section_id: 4402024389911, which is useless on its own. This Actor fetches the full taxonomy first and attaches real names, so your records are readable and filterable without a second pass.

Plain text, not HTML. Article bodies come back as HTML full of markup. Feeding that to a model wastes tokens. This strips it properly, including entity decoding, so records drop straight into a vector store.

Honest rate limiting. Zendesk returns Retry-After on 429 and this Actor respects it rather than hammering and failing.

Use it for

• Building a RAG corpus from your own or a competitor's documentation
• Feeding an AI support agent with accurate product knowledge
• Migrating a knowledge base off Zendesk
• Auditing which articles are outdated or poorly rated

Example input

{
  "helpCenterUrl": "https://support.discord.com",
  "maxArticles": 500,
  "includeBody": true
}

Filters

• Locale — leave blank to auto-detect, or force one such as de, fr, ja
• Category ID or Section ID — restrict to one part of the help centre
• Max articles — caps the run so cost is predictable
• Include full text — turn off for a fast title-and-URL index

Notes

• Uses only the public, unauthenticated Zendesk Help Center API. No login, no agent-side data, no tickets.
• Works on any Zendesk Guide instance, including custom domains.