A standalone editor for SDL Trados SDLXLIFF bilingual files – designed for translators who work with LLM Chatbots… and a few useful tag features!

What is it?

SDLXLIFF Refiner is a Windows desktop application that lets you open, edit, and save SDLXLIFF bilingual files without needing Trados Studio’s editor. It’s built specifically for translators who want to leverage Large Language Models (like ChatGPT, Gemini, Claude, or DeepSeek for example) as part of their translation workflow, but it works just as well as a lightweight standalone editor for quick fixes and bulk operations.

I built this because working with LLMs and Trados Studio bilingual files shouldn’t be painful. Copy your source text out with row numbers and a customisable prompt, paste it into your LLM of choice, then paste the translation straight back – and SDLXLIFF Refiner handles the alignment, the tags, and the formatting for you.

The application also supports working with a CLI (Command Line Interface) and this enables the ability to work with multifariousMCP if you are a user of Claude Desktop.

Key Features

Trados Studio Project Integration

SDLXLIFF Refiner reads your Trados Studio project infrastructure directly. Point it at your projects.xml file and you get a built-in project browser – navigate projects, language pairs, and bilingual files without leaving the application. No Studio API dependency, no plugins to install.

Visual Tag Editing

Inline tags are resolved from the SDLXLIFF header and displayed as colour-coded placeholders with WYSIWYG formatting previews. You can see at a glance what {cf1 bold} actually means, and the tag inventory panel gives you a complete overview of every tag type in the file with styled samples and occurrence counts.

Tracked Change Support

If your source documents contain tracked changes, SDLXLIFF Refiner displays them with familiar styling red strikethrough for deletions, blue underline for insertions. The built-in LLM prompt includes specific guidance for handling tracked changes, so your LLM knows to translate the accepted version while preserving the revision markup. Also solves a long-standing question of how do you translate tracked changes… now you can.

In-Cell Editing

Sometimes you just need to fix a typo or tweak a phrase without going through a full clipboard round-trip. Double-click any target cell to edit it directly – you’re working with the plain text and tag placeholders, so what you see is what you get. This is handy for quick corrections after an LLM pass, but it’s not a substitute for translating in Trados Studio. There’s no fuzzy match leverage, no termbases, no QA checks – it’s a lightweight editor for targeted fixes when needed. This is not the purpose of this tool.

LLM Clipboard Workflow

This is the heart of the application. The clipboard workflow is designed to make LLM-assisted translation as smooth as possible:

• Copy Source (Numbered) – copies source text with row numbers for alignment
• Copy Source (with Prompt) – prepends your customisable LLM prompt with language-specific instructions
• Paste Target – smart paste that matches translations back by row number, converts line-break markers, and skips locked segments
• Automatic chunking – large files are automatically split into LLM-friendly chunks under your configured word limit, with a visual dialog tracking which chunks have been copied and pasted

Per-Project LLM Prompts

Each Trados Studio project can have its own LLM prompt. The default prompt covers row alignment, tag preservation, line-break handling, and tracked change rules – but you can customise it per project for domain-specific instructions. Template variables [SOURCE] and [TARGET] are automatically resolved to the actual language names.

Search & Replace

Three modes to cover different needs, providing good search & replace operations for text and also tags:

• Plain text – with tag protection enabled by default so you can’t accidentally corrupt formatting markup
• Regex – full .NET regex support with capture groups
• Tag pair operations – remove or replace specific formatting tag types across the entire file, with optional text filtering and WYSIWYG previews

Now it’s possible to search and replace tags easily, add tags to translatable content (as long as they are already used in the document format), remove specific tags etc.

Command-Line Interface

For batch processing and scripting, SDLXLIFF Refiner includes a companion CLI tool (sdlxliff-refiner.exe). It exposes the same search and replace engine, tag removal, and tag replacement operations from the command line – so you can standardise terminology across an entire project, strip review highlighting from all German files, or chain operations in a PowerShell script. Version 3.0.0 adds a translate command that sends segments directly to OpenAI, Anthropic, or Google Gemini and writes the results back into the file, using the same numbered-chunk format and project prompts as the GUI clipboard workflow – no manual copy-paste required. Additional commands cover:

• segment inspection (list-segments),
• direct target editing (set-target),
• project browsing (list-projects), and
• prompt management (prompt).

Every command supports --dry-run for safe previewing and --json for structured output, making the CLI suitable for AI agent workflows as well as scripting. The CLI is a self-contained portable executable with no installation required.

LLM Provider Configuration

To use the translate command – whether you’re running it directly from the command line or via multifariousMCP in Claude Desktop – you need to connect SDLXLIFF Refiner to at least one LLM provider. The Settings window has an API Keys tab for this.

You can store keys for OpenAI, Anthropic, and Google Gemini directly in the application. They’re encrypted using Windows DPAPI so they’re only accessible under your own Windows user account. Once saved, the CLI picks them up automatically – no need to pass --api-key on the command line every time, and no risk of keys ending up in your shell history or scripts.

Beyond the three built-in providers, you can define named provider profiles. A profile combines a name, a provider type, a model identifier, and – if you’re using a custom endpoint like Azure OpenAI – an API URL. Each profile can have its own stored key. Instead of spelling out --provider openai --model gpt-4o every time, you save a profile called something like gpt4o and reference it by name.

This is where multifariousMCP fits in. multifariousMCP is a companion MCP server that exposes the SDLXLIFF Refiner CLI as tools that Claude can call directly within a Claude Desktop session. Claude can inspect your files, read segment content, apply terminology fixes, run translations via your configured provider, and verify the results – all within a normal conversation, without you switching between applications or typing commands manually. The API keys and provider profiles you set up here are the same ones multifariousMCP uses when it drives the CLI on your behalf, so there’s nothing extra to configure.

SDL Review Comments

The CLI now supports reading and writing SDL review comments directly in SDLXLIFF files. You can list all comments with their row numbers, add comments to specific segments, delete individual comments by ID, or strip all comments from a file or an entire project in one pass. Useful for bulk comment cleanup before delivery, or for AI agent workflows that need to flag segments for human review.

Locked Segment Handling

Segments locked in Trados Studio are visually marked, cannot be edited, and are automatically skipped during paste operations. You’ll always know which segments are locked and you won’t accidentally overwrite them.

Two-Tier Backup System

Every save creates a .bak rolling backup. On top of that, a permanent “original backup” preserves the file exactly as it was when you first opened it. If someone else edits the file in Trados Studio while you’re working on it, SDLXLIFF Refiner detects the external change and gives you options. You can restore any file to its original state at any time.

Merged Document Support

SDLXLIFF files based on merged documents are notoriously tricky – many external tools struggle with them. SDLXLIFF Refiner handles both within-paragraph merges and cross-paragraph merges correctly, filtering out ghost placeholder segments that would otherwise clutter your view.

Light & Dark Themes

Full light and dark theme support with customisable colours. The dark theme is VS Code-inspired and easy on the eyes for long editing sessions.

What’s in the Download?

The download zip contains three files:

• SdlxliffRefinerSetup-2.6.0.exe – installer for the desktop application (includes Start Menu and optional desktop shortcuts)
• sdlxliff-refiner.exe – the CLI tool as a standalone portable executable (no installation needed – just put it wherever you like)
• CLI-User-Guide.docx – command reference for the CLI tool

The sdlxliff-refiner.exe is used within the multifariousMCP.

Silent Installation

The installer supports Inno Setup’s silent install switches for automated deployment:

SdlxliffRefinerSetup-2.6.0.exe /VERYSILENT /SUPPRESSMSGBOXES
SdlxliffRefinerSetup-2.6.0.exe /VERYSILENT /SUPPRESSMSGBOXES /DesktopIcon=1

/SILENT shows a progress bar but no prompts. /VERYSILENT runs with no UI at all. Add /DesktopIcon=1 to create a desktop shortcut during silent install.

Requirements

• Windows with .NET 8 runtime (desktop application) or any 64-bit Windows (CLI – self-contained, no runtime needed)
• No Trados Studio installation required (but reads Studio project files if available)
• No third-party dependencies

Who is it for?

SDLXLIFF Refiner is for professional translators who use Trados Studio and want to incorporate LLM technology into their workflow without using the APIs and being able to use the full context of the file. It’s also useful if you simply need a quick way to view or edit SDLXLIFF files outside of Studio – for batch text replacements, tag cleanup, or reviewing translations. If you manage multiple projects or language pairs and need to apply consistent changes at scale, the CLI tool lets you script those operations without opening the desktop application.

If you’ve ever tried copying segments out of Trados Studio into an LLM and then struggled to get the translations back in with the right alignment and tags intact, this tool was built for exactly that problem.