Summary
Add an agent-invocable tool node for the DeepL API, from the partner-node priority list (DeepL is pri 2, assigned to @charliegillet).
Two tools:
deepl_translate — POST /v2/translate: translate a string or a batch of up to 50 strings; target_lang (regional variants supported), optional source_lang, formality, model_type (latency_optimized / quality_optimized / prefer_quality_optimized), context, preserve_formattingdeepl_write — POST /v2/write/rephrase: improve/rephrase text; writing_style XOR tone (the API rejects both together); supports DeepL Write's 11 target languages
Key mechanics:
- Auth via
Authorization: DeepL-Auth-Key <key> header; key from node config or ROCKETRIDE_DEEPL_KEY env var
- Free/Pro host auto-routing: keys ending
:fx route to api-free.deepl.com, others to api.deepl.com
- Error mapping: 456 = quota exceeded (DeepL-specific), 403 = auth, 429 = rate limit, generic non-2xx surfaces DeepL's own
message
- Follows the
tool_tavily pattern (error-dict returns, shared post_with_retry, no new dependencies)
Out of scope (future work): DeepL Voice API (WebSocket streaming; fits a future audio-lane node) and document translation endpoints (multipart upload + polling).
Summary
Add an agent-invocable tool node for the DeepL API, from the partner-node priority list (DeepL is pri 2, assigned to @charliegillet).
Two tools:
deepl_translate—POST /v2/translate: translate a string or a batch of up to 50 strings;target_lang(regional variants supported), optionalsource_lang,formality,model_type(latency_optimized/quality_optimized/prefer_quality_optimized),context,preserve_formattingdeepl_write—POST /v2/write/rephrase: improve/rephrase text;writing_styleXORtone(the API rejects both together); supports DeepL Write's 11 target languagesKey mechanics:
Authorization: DeepL-Auth-Key <key>header; key from node config orROCKETRIDE_DEEPL_KEYenv var:fxroute toapi-free.deepl.com, others toapi.deepl.commessagetool_tavilypattern (error-dict returns, sharedpost_with_retry, no new dependencies)Out of scope (future work): DeepL Voice API (WebSocket streaming; fits a future audio-lane node) and document translation endpoints (multipart upload + polling).