Replace text in multiple Freshdesk articles

Freshdesk has no find and replace for solution articles, and its API has no bulk update endpoint. To change a name, price or link across many articles, you edit each one by hand, write a script against the API, or use a tool that previews, snapshots and can undo the batch.

This guide compares the three routes, shows a safe script sketch, and ends with a checklist for renames.

What Freshdesk's bulk actions can and can't do

In the knowledge base list view, Freshdesk lets you select several articles and act on them together: move them to another folder, change visibility, change the author, and manage tags. Advanced article bulk actions are a Pro and Enterprise feature.

None of those actions touch article text. The article editor has no replace function either; your browser's Ctrl+F only finds text on the screen.

The API has the same gap. There is an endpoint to update one article (PUT /api/v2/solutions/articles/[id]) and nothing to update many. When someone asked the Freshworks developer community about bulk updates, the answer was to use the UI.

• Knowledge base features by plan
• Freshdesk API v2 reference: Solutions
• Bulk update articles (Freshworks developer community)

Checked 2026-09-19 · Freshdesk docs, API reference and developer community

Option A: edit each article by hand

The default route. Search the knowledge base for the old term, open each result, find every occurrence, edit, and save.

Our estimate of the effort, with the arithmetic: about 2 minutes per article (open, find, edit title and body, check the SEO fields, save). 100 articles is roughly 3.3 hours. Each translation is a separate article, so 3 languages triples that.

What goes wrong:

• Missed places. Search finds the body text, but the old name also hides in titles, meta titles, meta descriptions, image alt text and link text.
• Missed variants. "Workspace", "workspace", "workspaces" and "Work space" all need a decision.
• No undo. If you replace something you shouldn't have, you fix it by hand again. Article versioning (Pro and Enterprise) lets you restore one article at a time.

Fine for 10 articles. Painful for 100. Not realistic for 400 in four languages.

Option B: an API script

If someone on your team writes code, the Freshdesk API can do the job one article at a time. Authentication is HTTP Basic with the API key as the username and X as the password. Lists return at most 100 items per page.

The endpoints you need:

Rate limits are per account, per minute, and shared with every other integration: 100 on Growth, 400 on Pro, 700 on Enterprise. Invalid requests count too.

A sketch in TypeScript (Node 22). It runs as a dry run unless you pass --apply, writes a JSON snapshot of each article before changing it, stays under half the Growth limit, and stops after 3 errors in a row. It only replaces text between HTML tags, so link URLs and attributes are left alone.

// replace.ts — dry run: npx tsx replace.ts · apply: npx tsx replace.ts --apply
// env: FD_DOMAIN=yourcompany FD_KEY=... FIND=Workspace REPLACE=Project
import { mkdir, writeFile } from 'node:fs/promises';

const { FD_DOMAIN, FD_KEY, FIND = '', REPLACE = '' } = process.env;
const APPLY = process.argv.includes('--apply');
const base = `https://${FD_DOMAIN}.freshdesk.com/api/v2`;
const auth = 'Basic ' + Buffer.from(`${FD_KEY}:X`).toString('base64');
const sleep = (ms: number) => new Promise((r) => setTimeout(r, ms));
const find = new RegExp(`\\b${FIND.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')}\\b`, 'g');

async function api(path: string, init: RequestInit = {}): Promise<any> {
  await sleep(1200); // ~50 calls/min: half of Growth's 100/min
  const res = await fetch(base + path, { ...init, headers: { Authorization: auth, 'Content-Type': 'application/json' } });
  if (res.status === 429) {
    await sleep(Number(res.headers.get('retry-after') ?? 60) * 1000);
    return api(path, init);
  }
  if (!res.ok) throw new Error(`${init.method ?? 'GET'} ${path} → ${res.status}`);
  return res.json();
}

async function pages(path: string): Promise<any[]> {
  const out: any[] = [];
  for (let page = 1; ; page++) {
    const batch = await api(`${path}${path.includes('?') ? '&' : '?'}per_page=100&page=${page}`);
    out.push(...batch);
    if (batch.length < 100) return out;
  }
}

// Replace only in text nodes, never inside tags (keeps href, src, alt untouched).
const replaceText = (html: string) => html.split(/(<[^>]+>)/).map((part) => (part.startsWith('<') ? part : part.replace(find, REPLACE))).join('');

await mkdir('snapshots', { recursive: true });
let errors = 0;
for (const category of await api('/solutions/categories')) {
  for (const folder of await api(`/solutions/categories/${category.id}/folders`)) {
    for (const article of await pages(`/solutions/folders/${folder.id}/articles`)) {
      const title = article.title.replace(find, REPLACE);
      const description = replaceText(article.description);
      if (title === article.title && description === article.description) continue;
      console.log(`${APPLY ? 'UPDATE' : 'would update'} ${article.id} ${article.title}`);
      if (!APPLY) continue;
      await writeFile(`snapshots/${article.id}.json`, JSON.stringify(article, null, 2));
      try {
        await api(`/solutions/articles/${article.id}`, { method: 'PUT', body: JSON.stringify({ title, description }) });
        errors = 0;
      } catch (e) {
        console.error(e);
        if (++errors >= 3) throw new Error('3 errors in a row, stopping');
      }
    }
  }
}

Why this is still risky:

• No preview for a person to approve. The dry run prints titles, not the changed sentences. Review the diff before --apply, or you publish every match blind.
• Drafts and status. Freshdesk's reference says that passing only status: 1 unpublishes an article. The sketch doesn't send a status, but test on one article first and check your portal before running on all of them.
• Your snapshots are your only undo. The API has no versions endpoint. If the snapshot folder is lost, so is the way back. A restore script (PUT each snapshot's title and description) is worth writing before you apply.
• Translations are separate. The sketch covers primary articles only. Each language needs its own pass through the translation endpoint, with its own words to find.
• Nested folders and SEO fields. Accounts with subfolders (Enterprise) need to walk them too, and meta titles live in seo_data.
• Rate limits are shared. A long run can starve other integrations that use the same Freshdesk account.

Option C: TidyKB

TidyKB is built for this job: one search across titles, bodies and SEO fields, every match shown as a diff, and a batch you can undo.

The safety model is the same one the script above tries to copy. It is described on the security page.

Checklist before a rename

Go through this list before you replace a product or feature name across your help center.

1. List every variant. Singular, plural, lowercase, hyphenated, abbreviations, and the old name inside URLs.
2. Decide what stays. Release notes and changelog articles should often keep the old name, because they describe history.
3. Check titles and SEO fields. Meta titles and descriptions are easy to forget and show up in search results.
4. Check link text and image alt text. "See Workspace settings" is text; the URL behind it may also change.
5. Plan for translations. Decide the new term per language before anyone translates, and add it to your glossary.
6. Screenshots. Text changes won't fix a screenshot that still shows the old name. List those articles for a second pass.
7. Links. If article URLs change with the title, update links that pointed at them. The guide on how to find broken links in Freshdesk covers this.
8. Test with Freddy. If you use Freddy AI Agent, ask it three questions that mention the product by its new name.

Freshdesk and Freddy are trademarks of Freshworks Inc. TidyKB is an independent product and is not affiliated with, endorsed or sponsored by any company named on this page.