CLAPP: The CLASS LLM Agent for Pair Programming

2025-08-07

AI Summary

To address the high barrier to entry and low debugging efficiency that cosmologists face when using the CLASS Boltzmann solver, this paper introduces an interactive AI programming assistant tailored for numerical cosmology. Methodologically, it employs a multi-agent large language model (LLM) architecture integrating CLASS-specific knowledge retrieval via vector search, real-time Python code execution, and a Streamlit-based web interface—enabling natural-language querying, code generation, error diagnosis, and visualization. Its key contribution lies in being the first to deeply integrate semantically enhanced LLM reasoning with a domain-specific scientific computing environment, thereby delivering verifiable, conversational CLASS programming support. The system is open-sourced and publicly deployed; empirical evaluation demonstrates substantial reductions in user learning time and task completion latency, advancing the adoption of AI-augmented human–computer collaboration in computational cosmology.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Humans and AI: Intelligent User Interfaces
• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Assisted, interactive, and conversational search

Abstract

We introduce CLAPP (CLASS LLM Agent for Pair Programming), an interactive AI assistant designed to support researchers working with the Einstein-Boltzmann solver CLASS. CLAPP leverages large language models (LLMs) and domain-specific retrieval to provide conversational coding support for CLASS-answering questions, generating code, debugging errors, and producing plots. Its architecture combines multi-agent LLM orchestration, semantic search across CLASS documentation, and a live Python execution environment. Deployed as a user-friendly web application, CLAPP lowers the entry barrier for scientists unfamiliar with AI tools and enables more productive human-AI collaboration in computational and numerical cosmology. The app is available at https://classclapp.streamlit.app

Problem

• Provides conversational coding support for CLASS
• Generates code and debugs errors in CLASS
• Lowers entry barrier for scientists using AI tools

Innovation

• Leverages LLMs and domain-specific retrieval
• Combines multi-agent LLM orchestration
• Deploys as user-friendly web application