What problem does it solve? Graph summaries like Mapper graphs and transition networks are often misused as substitutes for persistent homology or built with undocumented parameters, producing unreproducible and overclaimed results. This Skill enforces a disciplined procedure for defining, constructing, and reporting graph analyses in TDL research. ## Core Features & Use Cases - Graph Construction Discipline: Requires explicit node semantics, edge semantics, directionality, weighting, and construction thresholds before any metric is computed. - Question-Matched Metrics: Computes only the network metrics that answer a named scientific question, avoiding unfocused metric dumps. - Reproducibility Controls: Mandates recording layout seeds, thresholds, and component handling in a structured output record. - Use Case: When building a Mapper graph over a UMAP embedding of employment trajectories, the Skill routes the embedding choice through representation diagnostics first, then records all construction parameters so the graph result can support a paper claim. ## Quick Start Ask the assistant to construct a Mapper graph from trajectory clusters and record the full graph construction output record for a TDL paper.
