AiiDA (Automated Interactive Infrastructure and Database for Computational Science) is a workflow engine that automates, stores, and manages computational simulations. It tracks every calculation as a node in a provenance graph — making your research reproducible and queryable.
Why AiiDA?
- Reproducibility: Every calculation, input, and output is stored and linked
- Automation: Define workflows once, run them many times
- Queryability: Search your results with powerful query tools
- Multi-code support: Works with VASP, CASTEP, QE, ASE, and more
🤖 Using AI Agents with AiiDA
One of the most powerful ways to use AiiDA is through an AI coding agent (like Claude Code, Cursor, or Copilot). Because AiiDA has a well-structured Python API and CLI (verdi), AI agents can:
- Write submission scripts — describe what you want in natural language, get working AiiDA Python
- Monitor and manage jobs — ask the agent to check job status, debug failures, resubmit
- Query results — “Find all failed CASTEP calculations from last week” → working
QueryBuildercode - Build workflows — agent can construct complex WorkChains from examples
- Troubleshoot — paste error output, get a fix
Tips for AI-Assisted AiiDA Work
- Share your profile and computer setup with the agent context (e.g.,
verdi profile list,verdi computer list) - Use the
verdiCLI for quick checks — agents can run shell commands easily - Use Python API for complex queries —
QueryBuilderinverdi shellor scripts - Keep example scripts — agents learn fast from working examples (see Usage)
- Reference the docs — point the agent to AiiDA documentation when it’s unsure
💡 Pro tip: If you’re stuck with AiiDA, ask an AI agent before spending hours debugging. AiiDA’s Python API is well-documented and AI agents are very good at writing correct queries and workflows.
Sections
- Setup — Installation, profile creation, computer & code setup
- Usage — Submitting jobs, querying data, managing groups
- Job Management — Monitoring, troubleshooting, and fixing calculations