LILT

Translate text and files from your AI assistant using LILT's enterprise platform.

Add to Agentman Studio

Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.

Paste it into any MCP client. Setup docs

Resources

What the LILT connector does

The LILT MCP server lets Claude, ChatGPT or any MCP-compatible agent translate text and files through LILT's enterprise translation platform. Five tools cover machine translation, human-verified job submission, job status, download, and a balance reading. Sign-in is OAuth. One of those tools sends work to human translators and bills your contract.

Verified connector

Listed by Anthropic as a partner connector in its Connectors Directory.

Connection checked by Agentman on .

Anthropic states this reflects the level of review a connector received, not a security audit.

LILT tools (5)

  • translate_text
  • translate_files_with_verification
  • get_credit_balance_information
  • check_job_status
  • download_job

Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.

Limits

  • Tool calls spend contracted money, and no confirmation step is documented. LILT's web flow puts a quote in front of a human before submission; nothing states the MCP path does the same.
  • A verified submission commits human labour. LILT states verified translation has a human in the loop and cannot be immediate. Treat translate_files_with_verification as irreversible in effect, even if a job can be cancelled.
  • The workflow cannot be changed after job creation. LILT documents this explicitly, and adds that connector-created jobs take the organization's default workflow, with per-connector selection a future feature.
  • The directory record carries no permissions label. There is nothing to read before approving.
  • We could not verify tool behaviour ourselves. The endpoint returns 401 to an anonymous handshake, so no schemas and no safety annotations were readable. Classification here is inferred.
  • LILT documents none of the five tools. Setup instructions only. Parameters, defaults and error behaviour are unpublished.
  • "Credit" is undefined. The word appears nowhere in LILT's 327-page documentation index, so the balance tool's units are unspecified by the vendor.
  • No named OAuth scopes. scopes_supported is an empty array, so a user cannot grant read-only access and withhold the ability to spend.
  • Confidentiality of submitted content is undocumented for MCP. LILT's privacy policy covers personal information about users, not translation content.
  • PII redaction is opt-in and role-dependent. The org setting must be enabled; admin exports and connector delivery return unredacted content.
  • API rate limits apply to the underlying platform. LILT documents 300 requests and 2,500,000 characters per minute per organization on its translation endpoints from 30 June 2026, returning HTTP 429 — while stating connectors are out of scope for those limits (API Rate Limits, fetched 2026-08-19).
  • Instant quotes are unavailable for some inputs. LILT states they cover text-based inputs only, and are not produced for content needing extraction such as video.
  • Third-party models may be slower. LILT notes translations may take longer where models are configured with an external provider.
  • Prompts are unconfirmed. Anthropic's snapshot lists none, and the gated endpoint blocked our own prompts/list.

Frequently asked questions

Does calling the LILT MCP server cost money?

Yes, for translation work. LILT is a contracted enterprise platform billed per word at rates set in your agreement, and its quoting documentation states that every quote is an estimate and the final cost may vary. LILT publishes no per-tool price list and never uses the word credit anywhere in its 327-page documentation, so the balance figure's units are undocumented.

Does an MCP tool call send work to a human translator?

Yes, when verification is requested. LILT's developer documentation states plainly that Verified Translation has a human in the loop and that immediate turnaround is therefore not possible. A submitted job enters a workflow with named translator and reviewer stages, carries a contractual SLA due date, and produces a bill. The agent creates an obligation a person then fulfils.

Can an agent check the balance before spending it?

Yes, and that ordering is the connector's best design decision. The directory lists get_credit_balance_information alongside the two translation tools, so a model can read available funds before committing to a submission. Most metered connectors in this catalogue expose no such preflight. LILT documents neither the tool nor what its numbers mean, so treat the reading as unspecified.

Which LILT MCP endpoint should you actually use?

It depends which listing you followed, and the two disagree. Anthropic's directory names mcp.lilt.com/mcp; LILT's own MCP page names mcp-enterprise.lilt.com/mcp. Both hosts were live and both returned HTTP 401 to an anonymous request on 2026-08-19, and each advertises a completely different OAuth authorization server. Ask LILT which one your organization is provisioned against.

Does LILT train models on the content you submit?

Yes, on your own models by design. LILT's platform builds customer-specific Contextual AI models from confirmed segments in your Data Sources, which is the product rather than a side effect. Segments a linguist marks as PII are excluded from the Data Source and from AI training, but LILT documents that exclusion as an organization setting an admin must enable.

Do you need an admin to turn on the LILT MCP server?

Yes. LILT documents an organization-wide MCP toggle on a dedicated admin page, reached through Manage in the sidebar, that must be switched on before any user in the organization can connect an AI client. The same page lists every connected client, shows which users connected it, and lets an admin revoke individual connections.

What can the LILT MCP server not do?

It cannot deliver verified translation instantly, and it exposes no editing surface. LILT states verified work carries a human turnaround governed by your contractual SLA. The five listed tools cover submitting, checking and downloading only — nothing in the list edits segments, manages Data Sources, or assigns reviewers, all of which LILT's REST API does support.

Sources

  • LILT MCP Server documentation — https://support.lilt.com/kb/LILT-mcp (fetched 2026-08-19 as Markdown via the .md suffix; this is the documentation URL in Anthropic's directory) · retrieved 2026-08-19
  • LILT MCP Admin — https://support.lilt.com/kb/lilt-mcp-admin-page (fetched 2026-08-19) · retrieved 2026-08-19
  • LILT Verified Translation (knowledge base) — https://support.lilt.com/kb/Verified-Translation (fetched 2026-08-19) · retrieved 2026-08-19
  • LILT Verified Translation (developer guide) — https://support.lilt.com/developers/guides/translate-content/verified-translation (fetched 2026-08-19) · retrieved 2026-08-19
  • LILT AI Translation developer guide — https://support.lilt.com/developers/guides/translate-content/instant-translate (fetched 2026-08-19) · retrieved 2026-08-19
  • LILT Workflows — https://support.lilt.com/kb/lilt-workflows (fetched 2026-08-19) · retrieved 2026-08-19
  • LILT Quoting — https://support.lilt.com/kb/quoting-in-lilt (fetched 2026-08-19) · retrieved 2026-08-19
  • LILT PII Omitted Segments — https://support.lilt.com/kb/PII-Omitted-Segments (fetched 2026-08-19) · retrieved 2026-08-19
  • LILT AI Providers — https://support.lilt.com/kb/3rd-party-llm-support-for-the-contextual-ai-engine (fetched 2026-08-19) · retrieved 2026-08-19
  • LILT API Rate Limits — https://support.lilt.com/developers/guides/rate-limits (fetched 2026-08-19) · retrieved 2026-08-19
  • LILT Privacy Policy — https://lilt.com/legal/privacy (fetched 2026-08-19) · retrieved 2026-08-19
  • LILT docs machine index — https://support.lilt.com/llms.txt (fetched 2026-08-19; 327 pages, searched for "credit" and for all five tool names, zero hits for each) · retrieved 2026-08-19
  • support.lilt.com/robots.txt permits the docs path and carries Content-Signal: ai-train=yes, search=yes, ai-input=yes — synthesis is expressly permitted on all three axes (fetched 2026-08-19). lilt.com/robots.txt is Allow: / with tracking-parameter exclusions. The MCP host mcp.lilt.com/robots.txt is a blanket Disallow: /, which is a crawl rule on an API host and not a documentation source · retrieved 2026-08-19
  • Live RFC 9728 protected-resource descriptor — https://mcp.lilt.com/.well-known/oauth-protected-resource/mcp (fetched 2026-08-19; the root and path-insert forms both 404). scopes_supported is an empty array · retrieved 2026-08-19
  • Live RFC 8414 authorization-server metadata — https://mcp.lilt.com/.well-known/oauth-authorization-server (fetched 2026-08-19; issuer https://blissful-celebration-62.authkit.app) · retrieved 2026-08-19
  • Live anonymous probe — 2026-08-19: an anonymous POST to https://mcp.lilt.com/mcp returns HTTP 401 with www-authenticate: Bearer error="invalid_token", resource_metadata="https://mcp.lilt.com/.well-known/oauth-protected-resource/mcp" · retrieved 2026-08-19
  • Live probe of the vendor-documented endpoint — 2026-08-19: an anonymous POST to https://mcp-enterprise.lilt.com/mcp returns HTTP 401; its descriptor at the path-append form names itself as authorization server and omits scopes_supported · retrieved 2026-08-19
  • Anthropic connector directory — https://claude.ai/directory/88245f33-e3a7-468f-ae27-3f118bef198a (snapshot 2026-08-16; no permissions field present) · retrieved 2026-08-16

More Productivity connectors

  • 6sense

    Ask an AI agent about 6sense account insights, buying signals and keyword trends.

  • Adobe Workfront

    Find, create, update and delete Workfront work items from an AI agent.

Browse all MCP connectorsClaude and ChatGPT setup guideBrowse AI skills

Product names and logos belong to their owners. Agentman is not affiliated with a connector's maker unless the page says the connector is built and run by Agentman.

Use it in an agent

Put LILT to work.

Connect LILT once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.

Open in Agentman Studio