What problem does it solve? Spreadsheet exports rarely match the shape of an infrastructure source of truth: columns mix multiple object kinds, dropdown cells carry display labels instead of choice names, and reference columns need to match the target schema's human-friendly ID. This Skill turns raw CSV/TSV inputs into schema-conformant Infrahub object YAML and loads them onto a fresh branch, failing closed whenever a column has no schema home instead of silently dropping data. ## Core Features & Use Cases - Schema-driven column mapping: Introspects the live Infrahub schema (via MCP, infrahubctl, REST, or local files) and maps columns to attributes using a strict heuristic ladder, translating dropdown labels to choice names and coercing Boolean, Number, DateTime, and JSON values. - Denormalized sheet decomposition: Splits one big spreadsheet conflating devices, sites, and manufacturers into numbered per-kind files with correct load order, or nests repeated child rows as inline component children. - Branch-first safe loading: Emits files with provenance comments, self-checks against object format rules, verifies reference closure, then creates a branch, validates, and loads — never writing to the default branch. - Use Case: A network engineer receives an inventory.csv listing devices with their sites and manufacturers. The Skill splits it into 01_manufacturers.yml, 02_sites.yml, and 03_devices.yml, resolves dropdown labels like "Active" to "active", and loads everything onto a csv-import branch ready for review. ## Quick Start Import the attached inventory.csv into Infrahub by converting it to object YAML and loading it onto a new branch.
