lai

What problem does it solve?

It solves the challenge of building reliable AI-powered tooling for a Langium DSL by turning your language project into a descriptor-driven workflow that supports evaluation and continuous prompt refinement.

Core Features & Use Cases

• Descriptor generation and refinement: Create and iteratively improve a YAML language descriptor from your Langium project structure so prompt generation stays aligned with your grammar, services, and examples.
• System prompt synthesis: Generate a DSL-aware system prompt directly from the descriptor to guide LLM behavior during evaluation.
• Evaluation loop with diagnostics: Run evaluation suites against real Langium parsing/validation to measure outputs, compare runs, and diagnose failures so you can refine the descriptor and prompts.
• Workflow automation for common scenarios: Use it to bootstrap a new Langium AI project, regenerate descriptors/prompts from scratch, validate file/path correctness, and maintain evaluation history for prompt improvements.
• Use case: After you update your grammar or validation rules, regenerate the descriptor and system prompt, re-run evaluations, and use history/compare/stats to confirm that improvements increased pass rates and reduced diagnostics.

Quick Start

Tell an AI assistant: "Guide me through running lai init, generating a language descriptor, generating a system prompt, and running lai evaluate to measure and improve my Langium DSL prompt quality."