> ## Documentation Index
> Fetch the complete documentation index at: https://composio-preview.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# LILT MCP

> LILT MCP connects agents to authorized translation and localization workflows exposed by LILT's hosted server.

![LILT MCP logo](https://mintcdn.com/composio-preview/ufpARLIsio16odJB/images/toolkits/lilt_mcp.svg?fit=max\&auto=format\&n=ufpARLIsio16odJB\&q=85\&s=3ddf9991bf2e7177d3d7e34d2e2aa647)

* **Category:** model context protocol
* **Auth:** DCR\_OAUTH
* **Composio-managed OAuth available?** No
* **Tools:** 9
* **Triggers:** 0
* **Slug:** `LILT_MCP`
* **Version:** 20261018\_00

## Tools

> [accordion] Check job status
>
> **Slug:** `LILT_MCP_CHECK_JOB_STATUS`
>
> Checks the status of a verified translation job.
>
> #### Input Parameters
>
> **job\_id** `integer` *required*
>
> The ID of the job to check.

> [accordion] Create trained model
>
> **Slug:** `LILT_MCP_CREATE_TRAINED_MODEL`
>
> Creates a new trained translation model for a specific language pair.
>
> Use this tool when a user wants to improve translation quality for a specific language pair by creating a translation model that can be trained with their own data. This is a prerequisite for using `translate_files_with_verification` for that language pair.
>
> #### Input Parameters
>
> **src\_lang** `string` *required*
>
> The source language formatted as an IETF BCP 47 language tag       (e.g., "en", "es").
>
> **trg\_lang** `string` *required*
>
> The target language formatted as an IETF BCP 47 language tag       (e.g., "fr", "de").

> [accordion] Download job
>
> **Slug:** `LILT_MCP_DOWNLOAD_JOB`
>
> Triggers a job export and returns a download link for the completed job.
>
> #### Input Parameters
>
> **job\_id** `integer` *required*
>
> The ID of the job to download.

> [accordion] Get credit balance information
>
> **Slug:** `LILT_MCP_GET_CREDIT_BALANCE_INFORMATION`
>
> Retrieves all available credit balances for the authenticated user.
>
> Instant credits refresh every month. You get a certain amount to use every month. Every month the credits refresh.
>
> Verified credits don't expire. Once they are purchased you will always have them.
>
> Use this tool to check the current balance for both instant and verified translation credits.
>
> Returns: dict\[str, int]: A dictionary representing the credits with the following keys:
>
> * instant\_translation(int): The number of instant translation credits
> * verified\_translation(int): The number of verified translation credits

> [accordion] Hello world
>
> **Slug:** `LILT_MCP_HELLO_WORLD`
>
> Returns a friendly hello world message.
>
> #### Output
>
> **result** `string` *required*

> [accordion] List resources
>
> **Slug:** `LILT_MCP_LIST_RESOURCES`
>
> Lists and filters jobs or models.
>
> #### Input Parameters
>
> **archived** `boolean`
>
> For jobs, whether to list archived jobs instead of active ones.       Defaults to False.
>
> **resource\_type** `string` *required*
>
> The type of resource to list. Can be "model" or "job".
>
> #### Output
>
> **result** `array` *required*

> [accordion] Translate files with verification
>
> **Slug:** `LILT_MCP_TRANSLATE_FILES_WITH_VERIFICATION`
>
> Create a verified translation job that will be assigned to professional LILT linguists for translation and review.
>
> IMPORTANT TWO-STEP WORKFLOW:
>
> STEP 1 - Get Quote (user\_confirmed=False, default):
>
> * Call this tool WITHOUT user\_confirmed or with user\_confirmed=False
> * Returns a quote showing credits needed, current balance, and detailed breakdown
> * NO job is created, NO credits are deducted
> * Show the user the quote and ask for explicit confirmation
>
> STEP 2 - Create Job (user\_confirmed=True):
>
> * Only after user explicitly confirms, call this tool again with user\_confirmed=True
> * Creates the job and deducts credits
> * Returns the created job details
>
> Credit Handling:
>
> * If insufficient\_credits=True: Inform user they don't have enough credits and suggest purchasing more at [https://mcp.lilt.com/home](https://mcp.lilt.com/home)
> * If credits are sufficient but low (10 or fewer): Warn user about low balance and suggest purchasing more credits
> * If credits are sufficient: Show quote details (credits needed, current balance, balance after) and ASK FOR EXPLICIT CONFIRMATION before proceeding to step 2
>
> If the tool throws a ValueError related to no trained models existing for a language pair, then tell this to the user and ask if they would like to create a trained model.
>
> #### Input Parameters
>
> **name** `string` *required*
>
> A 20 character name for the job summarized from the user's instructions.
>
> **file\_ids** `array` *required*
>
> A list file\_ids uploaded to LILT
>
> **src\_lang** `string` *required*
>
> The source language formatted as an IETF BCP 47 language tag.
>
> **trg\_langs** `array` *required*
>
> A list of target languages formatted as an IETF BCP 47 language tag.
>
> **user\_confirmed** `boolean`
>
> Set to True only after showing quote to user and getting            explicit confirmation. Default is False (quote mode).

> [accordion] Translate text
>
> **Slug:** `LILT_MCP_TRANSLATE_TEXT`
>
> Translates text using LILT's instant translate API.
>
> Use this tool when you need to translate text from one language to another using LILT's neural translation models.
>
> IMPORTANT: Check the 'insufficient\_credits' field in the response. If True, inform the user they don't have enough credits and suggest purchasing more instant translation credits at [https://mcp.lilt.com/home](https://mcp.lilt.com/home).
>
> If the user has sufficient credits but their balance is running low (5,000 credits or fewer), inform them about the remaining balance and suggest purchasing more credits to avoid interruptions.
>
> In the response object, if the "used\_trained\_model" is false, then tell the user that the translation was done with a LILT base model. If the user wants to improve the translation quality or match their brand voice they should create a trained model. Ask the user if they would like to create a trained model.
>
> #### Input Parameters
>
> **text** `string` *required*
>
> The text to translate
>
> **src\_lang** `string` *required*
>
> Source language code (e.g., "en", "es", "fr")
>
> **trg\_lang** `string` *required*
>
> Target language code (e.g., "en", "es", "fr")

> [accordion] Upload file
>
> **Slug:** `LILT_MCP_UPLOAD_FILE`
>
> Upload a file to LILT for translation.
>
> This tool handles file uploads. For non-plain-text file formats like PDF, DOCX, RTF, etc., it assumes the client has extracted the text content. To indicate that only the text is being used, the tool appends a `.txt` extension to the original filename (e.g., `report.docx` becomes `report.docx.txt`).
>
> IMPORTANT: You must inform the user of this behavior. When a user uploads a file like a PDF or DOCX, confirm by saying: "I am uploading the extracted text from your file \[original\_filename] for translation."
>
> #### Input Parameters
>
> **name** `string` *required*
>
> Original filename (e.g., "document.txt", "article.md")
>
> **content** `string` *required*
>
> The raw file content to be uploaded. For non-text files, this      should be the extracted text.
>
> **content\_type** `string`
>
> MIME type (defaults to "text/plain" if not specified)
