This contribution reflects on context engineering in digital scholarly editing, drawing on experiences from the Italian research landscape, including CNR-ILC, CLARIN-IT, Ca’ Foscari University of Venice, and the University of Pisa, with links to other Italian institutions.
Context engineering refers to the systematic design of the informational environment in which language models operate, including sources, constraints, terminologies, standards, and tools that guide their use in controlled philological workflows.
Three levels are discussed: workflows, where AI supports verifiable tasks such as HTR/ATR, correction, TEI encoding, and collation; infrastructures, where networks such as CLARIN, DARIAH, E-RIHS, and OPERAS provide shared data, standards, and services; and a theoretical level, where modal logics make explicit the epistemic, doxastic, and bouletic attitudes of human editors and AI agents within textual communities.
The aim is to identify transferable and assessable practices aligned with established standards in digital philology.