What problem does it solve? New datasets arriving from collaborators or facilities often get dropped into projects without documentation, making them impossible to interpret or reuse later. This Skill registers each dataset with structured metadata, provenance, and validation so it becomes a durable, documented part of the analytical project. ## Core Features & Use Cases - Structured data placement: Organizes raw data into per-dataset directories under data/raw/ with UPPER_SNAKE_CASE documentation files, treating raw data as immutable. - Metadata generation: Creates schema.yaml, provenance.md, and summary_stats.md from templates, capturing source, acquisition date, column descriptions, known issues, and access restrictions. - Manifest and decision logging: Updates data/DATA_MANIFEST.md with a new entry and logs ingestion decisions and learnings to the .living/ directory. - Domain-aware validation: Consults installed domain conventions (e.g., bioinformatics, image-analysis) for format-specific QC requirements. - Use Case: A sequencing facility delivers FASTQ files for a new experiment. Use this Skill to place them in data/raw/, generate provenance and schema metadata, apply bioinformatics QC conventions, and register the dataset in the manifest. ## Quick Start Use the ingest skill to add these FASTQ files from the sequencing facility to my project with full metadata and provenance.
