Masterarbeit

This thesis develops an agentic AI system that analyses scientific Jupyter notebooks and identifies missing FAIR and reproducibility elements. Specialized agents will inspect metadata, dependencies, datasets, software versions, documentation, provenance and execution environments. Based on their findings, the system will generate and, where possible, automatically apply concrete improvements such as environment specifications, metadata, documentation and provenance information. The research will investigate how reliably agents can diagnose and remediate common FAIR and reproducibility problems in computational research artifacts. The system could use Jupyter4NFDI as a source of real-world notebooks and provide a foundation for future automated FAIRification services.