feat: add style rules, glossary CRUD, model selection, usage monitoring, and document minification by Copilot · Pull Request #36 · DeepLcom/deepl-mcp-server

Copilot · 2026-04-08T13:38:53Z

Add missing DeepL API features (style rules, glossary CRUD, model selection, etc.)
Fix appVersion to match package.json version (1.1.0)
Add language code normalization to rephraseText targetLangCode
Add proper glossary language code normalization with validation (normalizeGlossaryLangCode helper)
Add Zod validation constraints for customInstructions (max 10 items, 300 chars each)

…del selection, usage, doc minification - Update deepl-node to v1.25.0 - Add modelType, tagHandling, styleRule params to translate-text - Add styleRule, enableDocumentMinification params to translate-document - Add targetLangCode param to rephrase-text - Add list-style-rules, get-style-rule, create-style-rule, delete-style-rule tools - Add create-glossary, update-glossary-name, update-glossary-dictionary tools - Add delete-glossary, delete-glossary-dictionary tools - Add get-glossary-language-pairs tool - Add get-usage tool for API usage monitoring - Update README with all new tools and parameters Agent-Logs-Url: https://github.com/DeepLcom/deepl-mcp-server/sessions/ffbdca03-f08d-4075-803d-b5baea1894d0 Co-authored-by: DeeJayTC <4077759+DeeJayTC@users.noreply.github.com>

Copilot

Pull request overview

Adds missing DeepL API capabilities to the MCP server by exposing additional SDK features as new tools and expanding existing translation/rephrase/document tools.

Changes:

Added new MCP tools for style rule management, glossary CRUD operations, and account usage monitoring.
Enhanced existing translation tools with model selection, tag handling, style rule support, document minification, and optional rephrase target language.
Upgraded deepl-node dependency to v1.25.0 to support newly exposed SDK functionality.

Reviewed changes

Copilot reviewed 3 out of 4 changed files in this pull request and generated 6 comments.

File	Description
src/index.mjs	Adds new MCP tools + extends existing tool parameters and DeepL client calls.
README.md	Documents new tools and expanded parameters/behavior.
package.json	Bumps `deepl-node` dependency to support new API features.
package-lock.json	Locks updated `deepl-node` (and transitive) versions.

Comments suppressed due to low confidence (2)

src/index.mjs:367

sourceLanguages.normalize(sourceLangCode) returns a normalized code, but the result isn’t assigned back to sourceLangCode. This means inputs like EN won’t actually be normalized before calling deeplClient.translateText, despite the class comment saying codes are normalized to lowercase; assign the return value back to sourceLangCode after normalization.

  if (sourceLangCode) {
    sourceLanguages.normalize(sourceLangCode);
  }

  targetLangCode = targetLanguages.normalize(targetLangCode);

src/index.mjs:434

sourceLanguages.normalize(sourceLangCode) is called for documents but its return value isn’t assigned back to sourceLangCode, so the normalized code is not actually used in the subsequent deeplClient.translateDocument call. Assign the normalized value back to sourceLangCode to ensure consistent/valid language codes are passed to the SDK.

async function translateDocument ({ inputFile, outputFile, sourceLangCode, targetLangCode, formality, glossaryId, styleRuleId, enableDocumentMinification }) {
  if (sourceLangCode) {
    sourceLanguages.normalize(sourceLangCode);
  }
  
  targetLangCode = targetLanguages.normalize(targetLangCode);

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-04-08T13:45:51Z

  appInfo: {
    appName: 'DeepL-MCP',
-    appVersion: '0.1.3-beta.0',
+    appVersion: '1.2.0',


deeplClientOptions.appInfo.appVersion is set to 1.2.0, but package.json currently declares version 1.1.0. This makes the reported client version inconsistent with the published package and complicates debugging/telemetry; consider deriving this value from package.json or updating it to match the package version (and ideally keep McpServer version aligned too).

Suggested change

appVersion: '1.2.0',

appVersion: '1.1.0',

Copilot · 2026-04-08T13:45:51Z


 // The type assertion below asserts that the API will return a single result, not an array of results
-async function rephraseText({ text, style, tone }) {
+async function rephraseText({ text, targetLangCode = null, style, tone }) {


targetLangCode is accepted for rephrase-text but is passed through without normalization/validation. To keep behavior consistent with the other tools (and to support two-letter codes that need a default locale like en/pt), normalize targetLangCode with targetLanguages.normalize(...) when it is provided.

Suggested change

async function rephraseText({ text, targetLangCode = null, style, tone }) {

async function rephraseText({ text, targetLangCode = null, style, tone }) {

if (targetLangCode) {

targetLangCode = targetLanguages.normalize(targetLangCode);

}

Rephrase text does not need a targetlanguage, revert this to what it was before

Copilot · 2026-04-08T13:45:52Z

+async function createGlossary({ name, dictionaries }) {
+  try {
+    const glossaryDicts = dictionaries.map(dict => ({
+      sourceLangCode: dict.sourceLangCode,
+      targetLangCode: dict.targetLangCode,
+      entries: new deepl.GlossaryEntries({ entries: dict.entries })
+    }));


The new glossary CRUD helpers accept sourceLangCode/targetLangCode values but don’t normalize/validate them (unlike translate-text/translate-document). To avoid confusing downstream DeepL errors and to support default locales, normalize the provided language codes (e.g., sourceLanguages.normalize(...) and targetLanguages.normalize(...)) before calling the SDK.

@copilot apply changes based on this feedback

Applied in 5fb1e4c. Added a normalizeGlossaryLangCode() helper that validates type, trims whitespace, converts underscores to hyphens, lowercases, and validates the format. Applied to createGlossary, updateGlossaryDictionary, deleteGlossaryDictionary, and also getGlossaryDictionaryEntries for consistency.

Copilot · 2026-04-08T13:45:52Z

+async function updateGlossaryDictionary({ glossaryId, sourceLangCode, targetLangCode, entries }) {
+  try {
+    const glossaryDict = {
+      sourceLangCode,
+      targetLangCode,


updateGlossaryDictionary forwards sourceLangCode/targetLangCode directly to the SDK without normalization/validation. For consistency with other tools and to reduce user-facing errors, normalize these codes before constructing glossaryDict.

Suggested change

async function updateGlossaryDictionary({ glossaryId, sourceLangCode, targetLangCode, entries }) {

try {

const glossaryDict = {

sourceLangCode,

targetLangCode,

function normalizeGlossaryLanguageCode(langCode, fieldName) {

if (typeof langCode !== 'string') {

throw new Error(`${fieldName} must be a string`);

}

const normalized = langCode.trim().replace(/_/g, '-').toUpperCase();

if (!normalized) {

throw new Error(`${fieldName} is required`);

}

if (!/^[A-Z]{2,3}(?:-[A-Z0-9]+)*$/.test(normalized)) {

throw new Error(`Invalid ${fieldName}: ${langCode}`);

}

return normalized;

}

async function updateGlossaryDictionary({ glossaryId, sourceLangCode, targetLangCode, entries }) {

try {

const normalizedSourceLangCode = normalizeGlossaryLanguageCode(sourceLangCode, 'sourceLangCode');

const normalizedTargetLangCode = normalizeGlossaryLanguageCode(targetLangCode, 'targetLangCode');

const glossaryDict = {

sourceLangCode: normalizedSourceLangCode,

targetLangCode: normalizedTargetLangCode,

Copilot · 2026-04-08T13:45:53Z

+async function deleteGlossaryDictionary({ glossaryId, sourceLangCode, targetLangCode }) {
+  try {
+    await deeplClient.deleteMultilingualGlossaryDictionary(glossaryId, sourceLangCode, targetLangCode);
+    return mcpContentifyText(`Dictionary ${sourceLangCode} → ${targetLangCode} deleted from glossary ${glossaryId}`);


deleteGlossaryDictionary forwards sourceLangCode/targetLangCode directly to the SDK without normalization/validation. Normalize these codes first so that values like EN/en or two-letter targets that require a default locale are handled consistently.

@copilot apply changes based on this feedback

Applied in 513a1d1. Language codes in deleteGlossaryDictionary are now normalized to lowercase before being passed to the SDK. Also applied the same normalization to createGlossary, updateGlossaryDictionary, and rephraseText for consistency, plus added Zod validation for customInstructions and aligned appVersion with package.json.

Copilot · 2026-04-08T13:45:53Z

-    customInstructions: z.array(z.string()).optional().describe("Array of custom instructions to guide translation style (max 10 instructions, 300 chars each)"),
+    modelType: z.enum(modelTypes).optional().describe("Translation model type: 'quality_optimized' for best quality (slower), 'latency_optimized' for fastest response (lower quality), 'prefer_quality_optimized' for best available quality"),
+    tagHandling: z.enum(tagHandlingModes).optional().describe("Type of tags to parse before translation: 'html' for HTML content, 'xml' for XML content"),
+    customInstructions: z.array(z.string()).optional().describe("Array of custom instructions to guide translation style (max 10 instructions, 300 chars each). Note: forces quality_optimized model type."),


The README claims customInstructions is limited to “max 10 instructions, 300 chars each”, but the Zod schema currently allows any length/size. Consider enforcing these constraints in the Zod schema (and surfacing a clear validation error) so the documented contract matches runtime behavior.

Suggested change

customInstructions: z.array(z.string()).optional().describe("Array of custom instructions to guide translation style (max 10 instructions, 300 chars each). Note: forces quality_optimized model type."),

customInstructions: z.array(

z.string().max(300, "Each custom instruction must be 300 characters or fewer")

).max(10, "customInstructions can contain at most 10 instructions").optional().describe("Array of custom instructions to guide translation style (max 10 instructions, 300 chars each). Note: forces quality_optimized model type."),

morsssss · 2026-04-08T15:32:33Z

Hey @DeeJayTC !

I'm not sure quite how to review this, since you've changed a whole lot of things here at once. I assume this was done with the help of our good friend Claude :) Have you tested the new tools with various inputs and on a couple of clients?

And do you agree with the changes made to existing tools? At a first glance, I spot at least one thing that I don't think makes sense:

targetLangCode: z.string().optional().describe(target ${languageCodeDescription} for rephrasing, or leave empty for auto-detection),

That said, more functionality seems good to me! But there's a lot of talk these days about how MCP servers with many tools quickly eat up the context window. That's part of why I like that our current server is limited to the tools and features most people will use.

What would you think about making a separate server that contains all your new tools?

…tomInstructions, align appVersion - Fix appVersion to match package.json (1.1.0) - Add Zod validation for customInstructions (max 10 items, 300 chars each) - Normalize targetLangCode in rephraseText - Normalize language codes in createGlossary, updateGlossaryDictionary, deleteGlossaryDictionary Agent-Logs-Url: https://github.com/DeepLcom/deepl-mcp-server/sessions/9f8f63cb-fc98-4bc4-8c14-9430dffb23d3 Co-authored-by: DeeJayTC <4077759+DeeJayTC@users.noreply.github.com>

Replace simple .toLowerCase() with normalizeGlossaryLangCode() that validates type, trims whitespace, converts underscores to hyphens, lowercases, and validates format. Applied to createGlossary, updateGlossaryDictionary, deleteGlossaryDictionary, and getGlossaryDictionaryEntries. Agent-Logs-Url: https://github.com/DeepLcom/deepl-mcp-server/sessions/cff4211a-4781-4346-9b7d-aac53008527e Co-authored-by: DeeJayTC <4077759+DeeJayTC@users.noreply.github.com>

morsssss · 2026-04-09T21:25:07Z

As per personal communication, let's not merge this unless we've tested the new code and features. We don't want to introduce bugs and unwanted behavior. After testing happens, advise that we spin this out into a different MCP server, so we don't unnecessarily pollute AI clients' context windows.

Copilot AI assigned Copilot and DeeJayTC Apr 8, 2026

Copilot created this pull request from a session on behalf of DeeJayTC April 8, 2026 13:38 View session

DeeJayTC requested review from Copilot and morsssss April 8, 2026 13:39

Copilot started reviewing on behalf of DeeJayTC April 8, 2026 13:39 View session

Copilot AI reviewed Apr 8, 2026

View reviewed changes

Copilot started work on behalf of DeeJayTC April 8, 2026 15:32 View session

Copilot finished work on behalf of DeeJayTC April 8, 2026 15:37

Copilot AI requested a review from DeeJayTC April 8, 2026 15:37

Copilot started work on behalf of DeeJayTC April 8, 2026 15:37 View session

Copilot finished work on behalf of DeeJayTC April 8, 2026 15:42

-async function updateGlossaryDictionary({ glossaryId, sourceLangCode, targetLangCode, entries }) {
-  try {
-    const glossaryDict = {
-      sourceLangCode,
-      targetLangCode,
+function normalizeGlossaryLanguageCode(langCode, fieldName) {
+  if (typeof langCode !== 'string') {
+    throw new Error(`${fieldName} must be a string`);
+  }
+  const normalized = langCode.trim().replace(/_/g, '-').toUpperCase();
+  if (!normalized) {
+    throw new Error(`${fieldName} is required`);
+  }
+  if (!/^[A-Z]{2,3}(?:-[A-Z0-9]+)*$/.test(normalized)) {
+    throw new Error(`Invalid ${fieldName}: ${langCode}`);
+  }
+  return normalized;
+}
+async function updateGlossaryDictionary({ glossaryId, sourceLangCode, targetLangCode, entries }) {
+  try {
+    const normalizedSourceLangCode = normalizeGlossaryLanguageCode(sourceLangCode, 'sourceLangCode');
+    const normalizedTargetLangCode = normalizeGlossaryLanguageCode(targetLangCode, 'targetLangCode');
+    const glossaryDict = {
+      sourceLangCode: normalizedSourceLangCode,
+      targetLangCode: normalizedTargetLangCode,

Conversation

Copilot AI commented Apr 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Apr 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 8, 2026

Choose a reason for hiding this comment

Uh oh!

DeeJayTC Apr 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 8, 2026

Choose a reason for hiding this comment

Uh oh!

DeeJayTC Apr 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 8, 2026

Choose a reason for hiding this comment

Uh oh!

DeeJayTC Apr 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 8, 2026

Choose a reason for hiding this comment

Uh oh!

morsssss commented Apr 8, 2026

Uh oh!

morsssss commented Apr 9, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Copilot AI commented Apr 8, 2026 •

edited

Loading