{"name":"Text Embeddings (Jina)","description":"Generate text embeddings via Jina AI (jina-embeddings-v3). Convert 1–96 input strings into dense vector representations for semantic search, similarity comparison, clustering, and RAG pipelines. Input: `texts` (array of 1-96 strings, ≤8192 chars each), `model` (defaults to jina-embeddings-v3), `dimensions` (32–1024, default 1024 — Matryoshka truncation), optional `input_type` (search_document / search_query / classification / clustering, default search_document) mapped to Jina task: retrieval.passage / retrieval.query / classification / separation. Output: `embeddings` (array of float vectors, one per input), `model` used, `dimensions`, and `tokenCount`. Multilingual (100+ languages). Max 8192 tokens per text — truncated upstream."}