Astrolabe icon

Astrolabe

Sovereign EU AI for the social economy — chat, embeddings and knowledge bases (OpenAI-compatible API)

Actions16

AI → Message a Model

AI-generated

Summary

Send an ordered set of system, user, and assistant messages to the selected Astrolabe chat model and return its completion. Optionally ground the answer in selected knowledge bases, with retrieval filters, date bounds, facet expansion, citation links, and static or agentic search settings.

Inputs

  • Model Name or ID (required) — Chat model to use; defaults to astrolabe-base. The model list is loaded from the Astrolabe catalogue.
  • Messages (required) — At least one message is required. Add messages in order, each with a system, user, or assistant role and text content.
  • Knowledge Base Names or IDs — Optional knowledge bases for retrieval-augmented answers. When selected, retrieval defaults to the last message and can be adjusted with the Knowledge Base Query, recent user turn count, result limit, date range, facet filter, and graph expansion options. You can choose static retrieval or let the model search in agentic mode; the max tool calls setting applies to agentic mode. Source-link citation is optional.
  • Options — Set response token limit (default 600), temperature (default 0.7), and text or JSON Object response format. Knowledge base-specific settings are used only when at least one knowledge base is selected.
  • Simplify Output — Defaults to true. Simplified output contains content, reasoning_content, finish_reason, astrolabe_sources, astrolabe_steps, and usage; when disabled, the full API response is returned.

Output shape

One n8n item per input item, containing the completion response.

With simplified output enabled, missing completion fields are represented by empty strings or null; astrolabe_sources and astrolabe_steps may be null when not returned. With simplification disabled, the full API response is returned. Errors stop execution unless Continue On Fail is enabled.

Links

Discussion