Unify LLM icon

Unify LLM

LangChain-style orchestration and multi-provider intelligence via Unify LLM

Actions10

Orchestration → Generate Structured

AI-generated

Summary

Generate structured JSON output from a connected LLM using a provided JSON schema, with optional system prompt, temperature, and max token limits.

Inputs

  • prompt (required) — Prompt or query text
  • schemaJson — JSON Schema object for structured output
  • systemPrompt — Optional system instruction
  • temperature — Sampling temperature
  • maxTokens — Maximum number of output tokens

Output shape

resource Orchestration, operation Generate Structured, model connected-model, data parsed JSON matching schemaJson, schema input schema, guardedMode boolean default true, usage null, metadata raw model output object, request object with prompt, systemPrompt, temperature, maxTokens, and model

A connected AI model must be provided via the Chat Model input. The operation parses schemaJson, injects it as a prompt instruction or uses native LangChain structured output, validates the model's JSON response via parseStructuredOutput, and returns the parsed data along with the original schema and runtime metadata.

Examples

Example 1: Generate a structured user profile

Prompt: 'Extract name and email from this text: ...'; schemaJson: '{{type: {name: string, email: string}}}'; systemPrompt optional; temperature 0.7; maxTokens 512. Returns data with the extracted fields, the schema used, and metadata containing the raw model output and request details.

Links

Discussion