Overview

This is the first alpha release of datasette-llm, a plugin that lets developers assign different AI models to different purposes within Datasette applications. The key innovation is centralized model configuration - you can now specify that data enrichment uses one model while SQL assistance uses another, all from a single configuration point.

Changes by Version

0.1a1

  • Added register_llm_purposes() plugin hook for registering purpose strings that other plugins can use
  • Added get_purposes() function to retrieve all registered purpose strings
  • Added purpose-based model assignment - developers can now specify which AI model handles which task (e.g., 'data enrichment uses GPT-5.4-nano but SQL query assistance uses Sonnet 4.6')
  • Added centralized model configuration for Datasette plugins that depend on LLM functionality