DeepL - Atako

Let your agents translate text, check character usage, list supported languages and manage translation glossaries via the DeepL API.

Connection

Authentication: API key (Authentication key).
Required settings:
- Plan — Use “api-free” if your key ends with :fx (DeepL API Free), or “api” for DeepL API Pro.

Sign in at deepl.com → My Account → “API keys & limits” tab → API keys → create or copy your authentication key. If the key ends with “:fx” it is a DeepL API Free key: enter “api-free” in the Plan field; otherwise (DeepL API Pro) enter “api”.See DeepL’s documentation.

Read actions (5)

Action	Description
`get_glossary`	Retrieve the details of a single glossary, including the language pairs its dictionaries cover.
`get_glossary_entries`	Retrieve the entries of one glossary dictionary, returned as tab-separated values inside a JSON object. Arguments: glossary_id (string, required — from list_glossaries), source_lang and target_lang (lowercase two-letter codes, both required — they select the dictionary).
`get_usage`	Get the character and minute usage for the current billing period, with the matching account limits.
`list_glossaries`	List all glossaries with their meta-information (id, name, dictionaries), without their entries.
`list_languages`	List the languages a DeepL resource supports. Each entry says whether the language is usable as a source, as a target, or both. Argument: resource (“translate_text” \| “translate_document” \| “glossary” \| “voice” \| “write” \| “style_rules” \| “translation_memory”, required).

Write actions (2)

Action	Description
`create_glossary`	Create a glossary of preferred translations, usable afterwards via translate_text’s glossary_id. Arguments: name (string, required — the glossary name), dictionaries (array of objects, required — one object per language pair, each with source_lang and target_lang as lowercase two-letter codes, entries as a string holding the term pairs, and entries_format “tsv” (tab-separated, default) or “csv”). Example entries value in tsv: “Hallo\tHello” with one pair per line.
`translate_text`	Translate one or more plain-text strings into a target language. Billed per character. Arguments: text (array of strings, required — each string is translated independently, translations come back in the same order; UTF-8 plain text only, 128 KiB per request), target_lang (string, required — target language code, e.g. “FR”, “DE”, “EN-GB”; call list_languages to discover valid codes), source_lang (string, optional — source language code; omit it and DeepL detects the language), context (string, optional — extra context that influences the translation but is not translated and is not billed), formality (“default” \| “more” \| “less” \| “prefer_more” \| “prefer_less”, optional — only some target languages support it; the “prefer_” forms never fail on unsupported languages), glossary_id (string, optional — glossary to apply; requires source_lang, and the glossary must contain a dictionary for that language pair), preserve_formatting (boolean, optional — default false; keep the original formatting instead of correcting it), tag_handling (“xml” \| “html”, optional — treat the text as markup and translate only its text nodes).

Permissions

Every action above must be explicitly granted to an agent before it can be used. See Permissions for the grant model and Security for how credentials are protected.