Managing multilingual content at scale is one of the most common challenges for global content teams. Whether you're translating navigation menus, product catalogs, or marketing pages, the process often involves tedious manual work or complex scripting.

With the Hygraph MCP server, you can now handle mass localization through a simple conversation—no code required.

#The challenge

Imagine you have hundreds of navigation items that need Swedish translations. Traditionally, you'd either update each entry manually through the UI (time-consuming) or write custom scripts using the Management API (requires development resources).

There's now a better way.

#The solution: Conversational content updates

By connecting Claude to your Hygraph project via the MCP server, you can update localized content in bulk using natural language. Here's how it works.

Step 1: Prepare your translation file

Structure your CSV with the content IDs and localized values. Column names should follow the pattern fieldName_localeCode:

id,name_sv,url_sv
cmfl1ohv4hla006w84a2m9d,Möbler,mobler
cmfl1ohv4hla206w8ddfyeg,Förvaringsmöbler,forvaringsmobler
cmfl1r7othp0l06w85hgf03,Bord,bord
cmfl3vxtcleqp06w8s882u2,Hyllor,hyllor
cmfl4scnrn0b308w8oi09am,Kontorsmöbler,kontorsmobler

A few things to keep in mind:

• Use complete entry IDs (the full string from Hygraph)
• The locale code (sv for Swedish) must already be configured in your Hygraph project
• Column naming convention: fieldName_localeCode makes it easy for Claude to understand which fields to update

Step 2: Connect Claude to Hygraph

Set up the Hygraph MCP server with your project credentials. This gives Claude secure, direct access to read your schema and update content entries.

Step 3: Have a conversation

Upload your CSV and describe what you need:

"I need to add Swedish translations to my navigation items. The content type is NavigationItem and the localized fields are 'name' and 'url'. Here's my CSV with the translations. Can you update these entries?"

Claude will:

1. Check your schema to confirm the fields exist and are localized
2. Show you what it's about to do
3. Update each entry with the Swedish translations
4. Report back on successes and any failures

Want more control? Ask Claude to generate the mutations first so you can review them before execution.

Step 4: Verify the results

After the updates complete, you can verify the changes:

"Can you show me one of the entries you just updated so I can confirm the Swedish content is there?"

#Tips for success

Start small. Test with 2-3 entries before running hundreds. This lets you catch any issues with field names or data formatting early.

Confirm your locale is configured. The target locale (like sv for Swedish) must already exist in your Hygraph project settings before you can add translations.

Use complete IDs. Hygraph content IDs are long strings—make sure you're using the full ID, not a truncated version.

Know your content type. Tell Claude the exact model name (NavigationItem, MenuItem, Category, etc.) so it can locate the right schema.

#Scaling up for larger projects

For bulk updates with hundreds or thousands of entries, batch processing keeps things manageable:

"Please process these in batches of 20 and let me know after each batch is done."

This approach lets you monitor progress, catch errors early, and avoid overwhelming the API with too many simultaneous requests.

#Why this matters

This workflow eliminates the traditional trade-off between manual effort and technical complexity. Content teams can handle localization tasks directly, while developers can focus on building features instead of writing one-off migration scripts.

The combination of structured data (your CSV) and natural language instructions (your conversation with Claude) creates a flexible system that adapts to your specific content model without custom code.

Ready to streamline your localization workflow? Connect the Hygraph MCP server to Claude and start updating content through conversation.