Automata ConsumerInstall

Summary

Searches Automata agents based on a user prompt or specified parameters, optionally validates relevance using an LLM, and executes selected agents with feedback in an n8n workflow.

Inputs

• Prompt (Optional) — A natural language prompt to analyze for extracting intents, categories, and tags if 'Use Analyze Prompt' is enabled.
• Use Analyze Prompt (required) — Enable to automatically extract intents, categories, and tags from the prompt via analyzePrompt.
• Intents (CSV) — Comma-separated intents used if 'Use Analyze Prompt' is disabled or as fallback.
• Categories (CSV) — Comma-separated categories used if 'Use Analyze Prompt' is disabled or as fallback.
• Tags (CSV) — Comma-separated tags or keywords used if 'Use Analyze Prompt' is disabled or as fallback.
• Location — Geographic string (City/State/Country) used if 'Use Analyze Prompt' is disabled or as fallback.
• Description — Optional natural language description sent to search when 'Use Analyze Prompt' is disabled.
• Search Limit (required) — Maximum number of agents to return from the search (1-50).
• Validate Relevance with LLM (required) — Enable to filter and select relevant agents/tasks using validateAgentRelevance (requires analysis).
• Execute Agents (required) — Enable to execute the selected agents using executeMultipleWithFeedback.
• Task — Fallback task used when executing agents if relevance validation is disabled.
• LLM Provider (required) — LLM provider to use (e.g., OpenAI, Claude, Gemini, DeepSeek, OpenRouter).
• LLM API Key (required) — API key for the chosen LLM provider.
• LLM Model (required) — Name of the LLM model to use (e.g., gpt-4o-mini).
• Temperature (required) — LLM temperature value between 0 and 1 used during analysis and validation.
• User Language (required) — Preferred language for responses from the SDK.
• Registry Environment (required) — Choice of Automata registry environment: Dev, Prod, or Custom URL.
• Registry URL — Custom Automata registry URL if 'Custom' environment is selected.
• Options — Advanced options including 'Return Interpretation Only' to output only the interpretation, omitting raw results.

Output shape

a single JSON output per input item containing analysis data, search results, relevant agents, execution results, and interpretation, depending on configuration.

The output fields vary based on node parameters: if 'Return Interpretation Only' is enabled, only the interpretation fragment is included; if 'Execute Agents' is disabled, no execution results are present. The node processes all input items individually, returning one output item per input.