tuning-zep-ingestion

Procedure for an agent that must turn a dataset (database records, documents, email, chat messages, meeting transcripts, event logs, or a mix) into a domain-specific Zep Context Graph, with or without a gold question set. The procedure splits the data into a deterministic import and narrative episodes, shapes each episode, designs the ontology and the custom instructions, runs a versioned ingestion on a new graph, measures the graph shape, reads the debug logs and the Ingestion Traces, and changes one variable per run. Use when a user asks to "ingest my data into Zep", "improve extraction quality", "why does my graph have duplicate or untyped entities", "tune the ontology or the instructions", or "evaluate Zep on my dataset". Use together with the building-with-zep skill from the same plugin. Activate that skill first if it is not active yet.