Entity EnricherInstall

Summary

Enriches the JSON entity in each input item against a selected schema using the chosen LLM models, then returns the best model result or a fused result. Multiple selected models enable fusion; optional settings control languages, strategy, classification, attachments, web search, structured output, timeout, and result metadata.

Inputs

• Authentication (required) — Connect with an organization API key (default) or an Entity Enricher account via OAuth2.
• Connected To Name or ID — Shows the organization and credential linked to the configured connection.
• Schema Name or ID (required) — Schema that defines the enrichment output structure. The entity data comes from the input item’s JSON.
• Model Names or IDs (required) — Select one or more models. Selecting multiple models enables fused results; the Auto option resolves to one organization-selected model without fusion.
• Attachment IDs — Comma-separated IDs of previously uploaded attachments to use as source material.
• Upload Input Binary Files — Upload binary files from the input item as source material. When enabled, the node uploads every binary property by default.
• Binary Fields to Upload — Comma-separated binary property names to upload; leave empty to upload all binary files. Shown when binary uploads are enabled.
• Delete Uploaded Attachments After Enrichment — Delete attachments uploaded by this node after enrichment, including when enrichment fails. Defaults to true; referenced Attachment IDs are not deleted.
• Language Names or IDs — Languages for multilingual fields; defaults to English.
• Strategy Name or ID — Choose an enrichment strategy. Auto lets the server select based on the schema; multi-expertise runs parallel calls per domain.
• Classification Model Name or ID — Optional model for classifying the entity type before enrichment.
• Arbitration Model Name or ID — Optional model for resolving conflicts when merging multi-model results; empty uses rule-based merging.
• Web Search Name or ID — Enable built-in provider web search for selected models that support it; defaults to off.
• Advanced Options — Configure provider response-schema output (on by default) and strict structured output (off by default), subject to selected model support.
• Timeout (Ms) — Maximum wait for the enrichment stream; defaults to 300000 ms. On timeout, the job is cancelled.
• Include Per-Model Results — Also emit one output item for each model result; defaults to false.
• Include Enrichment Metadata — Include result and enrichment details such as record IDs, cost, tokens, and fusion information. Defaults to false.

Output shape

One primary output item per input item, plus optional per-model output items.

By default, the primary item’s JSON contains only the enriched data. With metadata enabled, it contains a result object and available status, record, cost, token, model, and fusion details. If no result arrives, the primary item contains success: false, an error message, and a null result. Input JSON must contain a value; its field names do not need to match the schema. Uploaded binaries are sent as attachments and, by default, deleted after enrichment.

Links

• n8n community nodes installation guide