go-fair-us/niaid-blueprint
Agent skills and MCP access for implementing the NIAID Blueprint for Digital Objects: FAIR assessment, dataset metadata intake and extraction, citation, validation, teaching, and model-influence disclosure.
Guide users through generating citation text and BibTeX entries aligned with the NIAID Blueprint for Digital Objects (Section 4). Supports original-data deposits, reused-data attribution, and repository-level citations with PID integration (DOI, RRID, ORCID). Use when the user wants citation examples, BibTeX for a dataset or software object, a "How to Cite" page draft, or runs /niaid-bp-citation.
Conducts a conversational interview to collect metadata for a schema.org Dataset and produces a valid JSON-LD document aligned with the NIAID Blueprint for Digital Objects. Covers all 16 Blueprint metadata elements across five natural groups: identity, provenance, content, access, and context. Use when a researcher or repository manager needs structured dataset metadata, or says "create metadata for my dataset", "generate JSON-LD", "describe my dataset", or "Blueprint metadata".
Conducts a structured NIAID Blueprint FAIR assessment interview across six phases (resource overview, metadata schema, persistent identifiers, APIs, citation, and outreach/training) and produces a prioritized gap report with actionable next steps. Use when a repository owner, data steward, or PI wants to evaluate Blueprint or FAIR alignment, prepare for NIAID Data Ecosystem Discovery Portal integration, or says "assess my repository", "Blueprint assessment", "FAIR gap analysis", or "how do we integrate with the NIAID Portal".
Fetches a web resource URL and extracts Blueprint-aligned schema.org JSON-LD metadata (Table 1 elements) by retrieving and analyzing the target page. Produces a JSON-LD record plus metadata notes — no user interview required. Use when the user provides a dataset or resource URL for automatic extraction, or says "/niaid-bp-metadata-extract", "extract metadata from URL", "generate JSON-LD from this page", "Blueprint metadata from URL", or "scrape metadata for this resource".
Guide authors through developing a complete Model Influence Statement using the exact questions, fields, branching, and output structures from the Model Influence Statement Generator (https://model-influence-statement-generator.netlify.app/). Use when the user wants to create, fill out, or export a model influence statement, disclose ML model use in research, acknowledge AI assistance, or runs /niaid-bp-model-influence.
Run SHACL 1.2 Rules (SRL) with eyeleng via Deno: materialize inferred triples from DATA + RULE programs. Use when the user has an .srl file, wants SHACL Rules reasoning, says "run this SRL", "materialize rules", "eyeleng", or /niaid-bp-shaclrules. Not for Dataset pySHACL shape validation (use niaid-bp-validation).
Teach the NIAID Blueprint for Digital Objects through stateful, multi-session lessons. Creates a persistent teaching workspace with MISSION.md, HTML lessons, learning records, glossary, and reference sheets. Grounds all claims in RESOURCES.md. Use when the user wants to learn or teach the Blueprint, says "teach me the Blueprint", "Blueprint course", or "FAIR digital objects training", or wants progressive lessons across sessions toward a concrete repository goal.
Validate a schema.org Dataset graph (JSON-LD or Turtle) against NIAID Blueprint SHACL shapes using pySHACL. Emits a severity-aware conformance verdict plus structured violation results. Use when the user provides Dataset JSON-LD or Turtle for SHACL checks, wants Blueprint required-field validation, or says "validate this metadata", "check my JSON-LD against SHACL", "run pyshacl", or /niaid-bp-validation.