Claude DeepL connectorWhat Claude can do in your DeepL account.
The Claude DeepL connector exposes 14 tools. 7 read languages, glossaries and job status, 5 are filed by DeepL as write tools, from translating to fetching a finished file, and 2 carry no description. Below: what Claude translates, what DeepL keeps, and what needs an admin first.
Verified Trustpilot reviews · AI, automation & growth agency
What the Claude DeepL connector changes for multilingual teams
A support lead has to send the same incident notice to customers in German, Spanish and Japanese within the hour. Rather than copying the text into DeepL three times and checking the product terms by hand, she asks Claude. Claude sends the text to DeepL with the team's glossary and style rules applied, and returns the three versions in the conversation.
Translate with your own vocabulary. translate-text sends one or several texts to DeepL for up to 10 target languages at once, with optional formality, context, glossary and style rules. list-glossaries and get-glossary-dictionary-entries let Claude check which term pairs your team already fixed, so "customer account" comes out the way legal approved it.
Hand over a whole file. For a contract, a slide deck or a spreadsheet, upload-document starts a translation outside the conversation, get-document-status follows it, and download-document returns the translated file. DeepL moves the file bytes out of band, so the document content never passes through the model's working context.
Polish before sending. rephrase-text rewrites a passage in a chosen style or tone, and correct-text fixes grammar, spelling and punctuation, useful on a message drafted by a non-native speaker.
What the Claude sheet does not include: DeepL documents up to 28 tools, including ones that edit glossaries and style rules, but only 14 appear here, and none of them edits a glossary. File translation from Claude also depends on a network setting an organization Owner has to open. Nothing runs on its own either: no tool picks up a new file in a folder. For that, see the DeepL n8n integration or the DeepL Make integration.
Five words before you connect DeepL
- Connector
- The link you set up once between Claude and an account you already have, so Claude can work in it while it answers you.
- Tool
- One named action the connector gives Claude. Claude decides which ones to call while it answers; the directory lists each by name.
- Authorization
- DeepL's own sign-in and consent screen, where you hand Claude the access it will use. Granted once per person, and revocable.
- Approval
- The confirmation Claude waits for in the conversation before it carries on with something that would alter an account.
- MCP
- The shared standard that connectors are built on: it is what lets an assistant like Claude talk to an outside service such as DeepL.
Connect DeepL to Claude in three steps
- 01
Find DeepL in Claude
Open Claude's settings, go to Customize, then Connectors, and find DeepL. On a Team or Enterprise workspace, an Owner or Primary Owner must switch the connector on first, after which each member signs in separately.
- 02
Launch the connection
Click Connect on the DeepL row and sign in with your DeepL account in the window DeepL opens. If the link breaks later, choose Disconnect and connect again from the same list.
- 03
Approve DeepL's consent screen
Read DeepL's consent screen before approving it. It belongs to DeepL, not to Claude, and it sets what the access covers. You can take that access back later from Claude, or from your DeepL account.
The 14 tools, sorted by what they do
DeepL gives Claude 14 tools: 7 that read your account, 5 that change something in it, and 2 no official source describes.
Three groups: what Claude reads, what DeepL files as write tools, and two tools with no description. Names stay exactly as Claude shows them.
- 7 read
- 5 write
- 2 not documented
What Claude reads (7)
7 toolsSeven tools that look up languages, glossaries, instructions and job progress.
get-custom-instruction
Opens one custom instruction stored inside a style rule set, so Claude knows the extra guidance your team already gave DeepL before it asks for a translation.
get-document-status
Tells Claude where a document translation stands, so it knows whether the file is still being processed or ready to collect.
get-glossary-dictionary-entries
Pulls the term pairs of one glossary dictionary, source word against target word, so Claude can quote the approved wording.
get-glossary-info
Gives the metadata of one glossary, without spelling out every term it contains.
get-source-languages
Lists the languages DeepL accepts as input, so Claude does not promise a language pair that DeepL cannot handle in the first place.
get-target-languages
Gives Claude the list of languages DeepL can produce, the list it checks before promising you a version in a given language.
list-glossaries
Shows the glossaries available to you in DeepL, the starting point before Claude applies one or reads its entries.
Filed as write tools by DeepL (5)
5 toolsDeepL's own documentation labels these five as write tools: they produce new content or files. No source describes a Claude confirmation specific to them.
correct-text
Approval: see the ruleFixes grammar, spelling and punctuation in a passage and hands back the corrected version, keeping your wording otherwise.
download-document
Approval: see the ruleReturns the translated file once DeepL has finished. DeepL lists it among its write tools, even though it only fetches a file.
rephrase-text
Approval: see the ruleRewrites a passage in the style or tone you pick, handing back a new version beside the original.
translate-text
Approval: see the ruleSends one or several texts to DeepL for up to 10 target languages, with optional formality, context, glossary and style rules. DeepL classes it as a write tool.
upload-document
Approval: see the ruleLaunches the translation of a document or image outside the conversation. One upload can go into up to 10 target languages.
Not documented (2)
2 toolsThe directory shows these two names, but DeepL's documentation lists no tool under them. We will not guess from a name.
get-style-rule
This name shows up on the directory sheet, yet DeepL's documentation lists no tool called that; its style rule tools carry other names. No source describes it, so this page stops there.
list-style-rules
Listed on the directory sheet, but absent under this exact name from DeepL's tool reference. With no official text describing it, we do not read a purpose into it.
What Claude asks before acting
By default, Claude stops and asks for your go-ahead before each action it takes on an account for you. The request shows up in the conversation when it matters.
On a Team or Enterprise workspace, owners decide whether members may let certain actions through without a prompt every time. They can also restrict what a connector may do for the whole organization, for example keeping reads and closing writes, and no member overrides that. On the DeepL side, administrators can turn off AI connector access for their whole organization, and DeepL only shows the tools your subscription and team permissions allow. Claude works with your DeepL rights and nothing more: a glossary you cannot see stays out of reach.
Which plans include it
Among the 819 sheets in the official directory, not one shows availability per plan. That connector-by-connector answer is published nowhere: a real gap in the catalogue.
The public rule is the general one: remote connectors are open to all users on Claude, Cowork, Claude Desktop and mobile. On Team and Enterprise, an Owner or Primary Owner turns the connector on for the organization before members can sign in. What your DeepL subscription unlocks, glossaries or style rules for instance, is a separate matter set by DeepL. For the up-to-date state, check the connector's sheet in the official directory.
Where the Claude DeepL connector stops
A connector is not an automation. Claude calls these tools while it answers you: nothing translates by itself when a file lands in a shared folder.
Only DeepL documents this connector: no Claude help article covers it, so these limits come from DeepL's developer page and the directory sheet. The tool list is an observed floor, not a promise, since administrators can open actions no public sheet shows. The partner badge is not a security audit, and Anthropic states on every sheet that it does not choose a publisher's tools. On data, DeepL says text is not stored beyond the request and document sessions are deleted shortly after. For another publisher-only connector, see the Claude juro connector.
Need help connecting DeepL to Claude?
A person reads every message.
