MCP App Store
by KeenEthics
General Translation icon

General Translation MCP Server for ChatGPT

By General Translation, Inc.

Overview

General Translation is an AI localization platform for software teams. This plugin lets ChatGPT and Codex work with your General Translation projects directly from a conversation. Translate text on the spot, or upload source files and queue translations into your project's locales. Check translation progress and retrieve translated files. Manage branches, file versions and tags. Build and edit glossaries, term translations and custom instructions. Publish or unpublish translations on the General Translation CDN. Translate Google Docs, Slides and Sheets through a connected Google Drive integration. Sign in with your General Translation account and approve the requested access. The plugin can access only the organizations and projects available to your account. Some tools modify or delete data, start translation jobs, or publish content publicly. Translation jobs may incur usage charges.

How to connect

  1. ChatGPT: Open the General Translation listing in the ChatGPT app directory and choose Connect. ChatGPT listing General Translation exposes 54 tools here.

Tools

assign_context_group

ChatGPT
Assigns a context group to a project at the lowest priority. If it's already assigned, nothing changes (created is false), and repeating it never moves a group that was reordered. Requires org:context:write.

create_branch

ChatGPT
Create a branch and return it. A name the project already uses is rejected, as is isDefault when the project already has a default branch; use update_branch to rename a branch or change the default. Non-default branches require a paid plan.

create_context_group

ChatGPT
Creates an empty, unassigned context group in the org. Each call creates a new one. Requires org:context:write.

create_project

ChatGPT
Create a project in an organization with a name and default locale, optionally enabling CDN delivery. Requires org:projects:create. Repeated calls create additional projects.

delete_context_custom_prompt

ChatGPT
Deletes a custom prompt. Requires org:context:write.

delete_context_glossary_entry

ChatGPT
Deletes a glossary term and all of its translations. Requires org:context:write.

delete_context_glossary_translation

ChatGPT
Deletes one locale's translation of a glossary term and keeps the term and its other translations. Requires org:context:write.

delete_context_group

ChatGPT
Deletes a context group along with all its glossary terms, custom prompts and project assignments. Requires org:context:write.

export_context_content

ChatGPT
Exports a context group's glossary terms and custom prompts as a page of a JSON or CSV transfer document; each page can be re-imported as one batch. Pass nextCursor as cursor until it is null. Pages aren't a snapshot, so changes made mid-export may be missed or included. Requires org:context:read.

generate_project_context

ChatGPT
Queue generation of glossaries and translation instructions for the supplied source files and locales, and return the queued job, or job: null when context is already current. Every file must exist in the project, or nothing is queued. Adds the target locales to the project's current locales when a job is queued. Set force to regenerate existing context; poll the job with get_job.

get_context_glossary_entry

ChatGPT
Gets a glossary term by ID: its keyword and definition. Translations are listed separately. Requires org:context:read.

get_context_group

ChatGPT
Gets a context group's ID and name. Requires org:context:read.

get_file_content

ChatGPT
Download one source file, or its translation when locale is given, as base64. A file too large to return inline has null data and a downloadUrl to fetch it from; media files always carry a downloadUrl. Reads the default branch unless branchId is given, and the branch head unless versionId is given. Without versionId, a translation is the newest completed version on the branch, and pending only while none is completed; to wait for a specific version, pass its versionId. Returns a query_files_content result: status ready with the content, or status pending with fileId, branchId, versionId, locale and null data when the translation is stored but not completed yet, for example during a layout pass. A missing file is an error, as is a translation while its translation job has not stored the translation yet, so wait for the job with get_job or query_jobs before reading it.

get_file_status

ChatGPT
Return a source file's metadata and the translation state of each locale. Reads the default branch unless branchId is given, and the branch head unless versionId is given.

get_google_drive_context

ChatGPT
List accessible GT projects with Google Drive integrations and their locale options. Optionally provide both fileId and hostApp for linked-file and translated-copy context. Users can use their linked Google account or an organization integration; API keys require an organization integration.

get_google_drive_translation_status

ChatGPT
Poll translation and layout progress for a Google Docs, Slides or Sheets file and get translated-copy URLs when available. A GT-managed translated copy resolves to its source and its target locale. Users can use their linked Google account or an organization integration; API keys require an organization integration.

get_job

ChatGPT
Return the status of a file translation or context generation job started by translate_files or generate_project_context. Poll until the status is completed or failed; a failed job reports its error.

get_project

ChatGPT
Return locale, approval and CDN delivery settings for a project. Use list_projects to find project IDs.

get_project_context_group

ChatGPT
Gets one assignment of a context group to a project: the group's ID, name and priority. Requires org:context:read.

import_context_content

ChatGPT
Merges one JSON or CSV transfer document of glossary terms and custom prompts (up to 1 MiB as a JSON string, so escaped quotes and line breaks count) into a context group as one atomic batch. The whole document is validated before anything is written. An absent field keeps its stored value. null clears a nullable field or removes a map entry; other fields reject null. In batches and imports, items match stored items by natural key, stored items absent from the batch are untouched, and nothing else is deleted. A blank CSV cell counts as absent. In CSV, a blank prompt locale means every locale, and the term, prompt name and prompt text can't be blank. For larger imports, send several documents; each commits on its own. Requires org:context:write.

list_branch_files

ChatGPT
Lists the files on a branch with their head version, in the order they were added to the branch. To find files removed locally, compare the listed file IDs with your own. Requires project:files:read. Omit cursor for the first page. When the response's nextCursor is a string, pass it as cursor to fetch the next page; when it is null, there are no more results.

list_branches

ChatGPT
List the project's branches ordered by ID, each marked with whether it is the default branch. Pass name to find branches by exact name, and isDefault to find the default branch. Use the returned branch IDs in file operations. Omit cursor for the first page. When the response's nextCursor is a string, pass it as cursor to fetch the next page; when it is null, there are no more results.

list_context_custom_prompts

ChatGPT
Lists a context group's custom prompts by ID. A null locale means the prompt applies to every locale. Requires org:context:read. Omit cursor for the first page. When the response's nextCursor is a string, pass it as cursor to fetch the next page; when it is null, there are no more results.

list_context_glossary

ChatGPT
Lists a context group's glossary terms by ID, without translations. keyword is optional and matches the stored keyword exactly (case-sensitive), so it returns at most one term. Requires org:context:read. Omit cursor for the first page. When the response's nextCursor is a string, pass it as cursor to fetch the next page; when it is null, there are no more results.

list_context_glossary_translations

ChatGPT
Lists a glossary term's translations by ID. Requires org:context:read. Omit cursor for the first page. When the response's nextCursor is a string, pass it as cursor to fetch the next page; when it is null, there are no more results.

list_context_groups

ChatGPT
Lists the org's context groups, ordered by ID. Requires org:context:read. Omit cursor for the first page. When the response's nextCursor is a string, pass it as cursor to fetch the next page; when it is null, there are no more results.

list_orgs

ChatGPT
List organizations you belong to. Use the returned IDs as orgId in organization tools. Omit cursor for the first page. When the response's nextCursor is a string, pass it as cursor to fetch the next page; when it is null, there are no more results.

list_project_context_groups

ChatGPT
Lists the context groups assigned to a project, in priority order. Requires org:context:read. Omit cursor for the first page. When the response's nextCursor is a string, pass it as cursor to fetch the next page; when it is null, there are no more results.

list_projects

ChatGPT
List projects you are a member of. Use the returned IDs as projectId in project tools. Omit cursor for the first page. When the response's nextCursor is a string, pass it as cursor to fetch the next page; when it is null, there are no more results.

move_files

ChatGPT
Move or rename 1–25 files by cloning each latest source file and its translations under a new file ID, and return one result per move. Each old and new file ID may appear once, and no file ID may be both an old and a new file ID: a batch that swaps, cycles or chains file IDs is rejected without changing anything.

overwrite_translations

ChatGPT
Replace 1–100 existing translations with full translated file content and return one result per translation. Each branch, file, version and locale may appear once; retrying failed translations with the same content is safe, but it re-saves them, which can repeat completed webhooks and revisions.

publish_files

ChatGPT
Publish or unpublish the translations of 1–100 file versions on the CDN and return one result per version, with a result for each translation. Each branch, file and version may appear once.

query_files

ChatGPT
Get source and translated file metadata, including approval, completion, and publication dates, for up to 100 selectors. Each selector reads the default branch unless branchId is given, the branch head unless versionId is given, and the source file unless locale is given; a selector repeated field for field is rejected. Returns one result per selector, in selector order: status found with the metadata (completed translations only), status pending when the translation is stored but not completed yet, for example during a layout pass, or status not_found, including while its translation job has not stored the translation yet, so wait for the job with get_job or query_jobs before reading it.

query_files_content

ChatGPT
Download up to 100 source and translated files as base64. A file too large to return inline has null data and a downloadUrl to fetch it from; media files always carry a downloadUrl. Each selector reads the default branch unless branchId is given, the branch head unless versionId is given, and the source file unless locale is given; a selector repeated field for field is rejected. Returns one result per selector, in selector order: status ready with the content, status pending when the translation is stored but not completed yet, for example during a layout pass (request it again later), or status not_found, including while its translation job has not stored the translation yet, so wait for the job with get_job or query_jobs before reading it. Without versionId, a translation is the newest completed version on the branch, and pending only while none is completed; to wait for a specific version, pass its versionId.

query_jobs

ChatGPT
Return the statuses of up to 100 file translation or context generation jobs by ID, such as the jobs returned by translate_files, one result per job in request order; a job not in the project is not_found and a repeated ID is rejected. Poll until every status is completed or failed; failed jobs report their error.

reorder_context_group

ChatGPT
Moves a context group assigned to a project to a new zero-based position; lower positions win for duplicate terms. Requires org:context:write.

set_context_glossary_translation

ChatGPT
Creates or replaces one locale's translation of a glossary term (outcome reports which). Keeps the term and its other translations. Requires org:context:write.

set_tag

ChatGPT
Create a tag pointing to source file versions, or replace an existing tag whole: its files become exactly files and its message becomes message, cleared when omitted. created reports which. Every file reference must resolve in the project, otherwise nothing is written.

submit_feedback

ChatGPT
Report friction with the General Translation MCP tools to the General Translation team. Use it when no tool could do what the user asked, a tool failed or returned something wrong, a tool or its docs were unclear, or you have a suggestion. Call at most once per task, after you finish or give up, then tell the user what you reported. Feedback goes to the General Translation team, not to the user's organization, and does not perform the requested fix.

translate_content

ChatGPT
Translate strings, structured entries, or markdown documents using project context and caching, and return the translations directly. The content is not saved to the platform as a file or translation job. To translate a file, use upload_source_files and translate_files instead. This may incur translation charges. Each key in translations holds its own result; inspect every result for success: false.

translate_files

ChatGPT
Queue up to 100 uploaded source files for background translation into the target locales (the project's current locales when omitted). Returns one job per file and locale that needs translating; files already translated into a locale get no job unless force is set, so jobs may be empty. Each file must be listed once and exist on its branch, or nothing is queued. Adds the target locales to the project's current locales; the project default locale is unchanged. This may incur translation charges. publish controls CDN publication. Poll the returned jobs with get_job or query_jobs.

translate_google_drive_file

ChatGPT
Sync a Google Docs, Slides or Sheets file, enqueue translations, and apply them to Drive copies on completion. Imports edits from existing translated copies. May incur translation charges and update existing copies. Inspect enqueued, failed, and errors for partial success, then poll get_google_drive_translation_status. A GT-managed translated copy resolves to its source and only its target locale. Users can use their linked Google account or an organization integration; API keys require an organization integration.

unassign_context_group

ChatGPT
Removes a context group's assignment to a project and keeps the group and its content. Does nothing if the group isn't assigned. Requires org:context:write.

update_branch

ChatGPT
Rename a branch or make it the default branch, and return it. Omitted fields are unchanged. Making a branch the default unsets the previous default. A name another branch uses is rejected. Use list_branches to find branch IDs.

update_context_custom_prompt

ChatGPT
Updates a custom prompt by ID. An absent field keeps its stored value. null clears a nullable field or removes a map entry; other fields reject null. In batches and imports, items match stored items by natural key, stored items absent from the batch are untouched, and nothing else is deleted. A blank CSV cell counts as absent. Fails if another prompt already has the resulting name and locale. Requires org:context:write.

update_context_glossary_entry

ChatGPT
Updates a glossary term's keyword or definition. An absent field keeps its stored value. null clears a nullable field or removes a map entry; other fields reject null. In batches and imports, items match stored items by natural key, stored items absent from the batch are untouched, and nothing else is deleted. A blank CSV cell counts as absent. The term ID and translations are kept when the keyword is renamed. Fails if another term already has the keyword. Requires org:context:write.

update_context_group

ChatGPT
Renames a context group or turns its automatic updates on or off. With automatic updates off, translation runs never add terms or context to the group. Requires org:context:write.

update_files

ChatGPT
Update 1–100 existing source files or translations, such as their preview URL, and return each after the update. A file without a locale is the source version; with one, its translation. An absent field keeps its stored value. null clears a nullable field or removes a map entry; other fields reject null. In batches and imports, items match stored items by natural key, stored items absent from the batch are untouched, and nothing else is deleted. A blank CSV cell counts as absent. Each file version and locale may appear once. Writes nothing when a source file version or translation does not exist, or when a source file version belongs to an integration; repeating a request is safe.

update_project

ChatGPT
Update the project default source locale or CDN publishing setting, and return the updated project. This tool does not publish or unpublish existing files. Omitted settings remain unchanged.

upload_source_files

ChatGPT
Upload 1–100 source files with base64-encoded content and return one result per file. Uploaded files are saved to the platform. They can then be translated with translate_files, or existing translations can be added with upload_translations. Rejects the whole request without saving anything when any file's content is not valid base64 or does not parse in its format. Existing source versions are saved again; retrying failed files with the same content and identifiers is safe. Does not change the project default locale. LOTTIE content must be a base64-encoded .lottie ZIP archive, and PNG, JPEG, and MP4 content must be the base64-encoded file.

upload_translations

ChatGPT
Import base64-encoded translations for 1–100 existing source files and return one result per translation. Rejects the whole request without saving anything when any translation's content is not valid base64 or does not parse in its format. May overwrite translations and publish according to the project CDN setting; does not change the project default locale. Approval follows project policy. Retrying failed translations with the same content is safe. Fails when translations were saved but the source files' locale list could not be updated; the error lists saved and failed files, and retrying the upload is safe but re-saves every translation, which can repeat completed webhooks and revisions.

upsert_assets

ChatGPT
Uploads 1–10 fonts (assetType: FONT, the only asset type today) for Lottie translations across the organization. A font is matched by family, weight and italic style (see family and style); uploading a matching font replaces it for the whole organization. The batch is checked before any font is written: an empty or oversized font, a font with no derivable family, or two fonts with the same identity reject the whole batch (400). Fonts are stored independently, so the result reports each font, in request order, as succeeded or as failed with an error, even when every font fails; retrying a failed font is safe. Requires org:assets:write.

upsert_context_custom_prompts

ChatGPT
Creates or updates 1–100 custom prompts, each matched by name and locale, in one atomic batch. An absent field keeps its stored value. null clears a nullable field or removes a map entry; other fields reject null. In batches and imports, items match stored items by natural key, stored items absent from the batch are untouched, and nothing else is deleted. A blank CSV cell counts as absent. Repeating a name and locale within one batch rejects the whole batch. Requires org:context:write.

upsert_context_glossary

ChatGPT
Creates or updates 1–100 glossary terms (send one term as a one-item entries array), matched by keyword, in one atomic batch. An absent field keeps its stored value. null clears a nullable field or removes a map entry; other fields reject null. In batches and imports, items match stored items by natural key, stored items absent from the batch are untouched, and nothing else is deleted. A blank CSV cell counts as absent. Repeating a keyword within one batch rejects the whole batch. Requires org:context:write.

Capabilities

Writes

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/general-translation"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/general-translation)