hitsz-ids/airda

Health v2 · maintenance only

Adoption not part of the score

1785 stars · 274 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

airda (Air Data Agent) is a Python-based multi-agent system for data analysis that understands natural language data requirements and generates SQL and Python code for querying, visualization, and machine learning tasks. It coordinates specialized agents (data lookup, SQL generation, code generation, visualization) through multi-turn dialogue with self-debugging capabilities.

Use cases

• generate SQL from natural language questions
• find the right tables across hundreds of database tables
• analyze business metrics with code generation
• visualize query results as charts and dashboards
• chat with a MySQL database in plain language
• auto-debug generated data analysis code
• build data APIs and metric dashboards from analysis results

When to choose

• you need natural-language text-to-SQL over MySQL data sources
• you want a multi-agent pipeline that plans and executes data analysis tasks
• your team needs business-metric-aware data retrieval across many tables
• you want a pip-installable CLI agent backed by MongoDB for knowledge storage

When to avoid

• you only support non-MySQL databases, since datasource support is currently MySQL-only
• you need production-grade chart generation or task planning, which are still unfinished roadmap items
• you want a fully self-contained tool without external dependencies like MongoDB and an OpenAI-compatible LLM endpoint
• you need a no-code GUI product rather than a developer-oriented library and CLI

Facets

library · maturity active

agent-framework rag llm-inference data-visualization nlp search-engine cli data-science analytics large-language-models databases data-visualization python cli cross-platform text-to-sql multi-agent data-analysis chatbot mongodb mysql embedding-model self-debugging ai-agents docker

Member repositories

For agents

markdown · JSON · MCP: product_card(name="hitsz-ids/airda")