| id | rare_disease_skills_index |
| name | Rare Disease Skills Index |
| description | Skills for rare disease case support: ontology-first normalization/retrieval
and the clinical genetics consult report format contract (structure + theme).
Load the relevant skill file when performing the matching task.
|
| tags | ["rare_disease","ontology","orphanet","omim","hpo","clinical_report"] |
Agent Skills for Rare Disease Case Support
Curated playbooks for the rare disease MDT workflow. Each agent loads the skill
file that matches its current task — do not inline these contracts into prompts.
Ontology-First Workflow
Deterministic normalization against the local Orphanet/OMIM/HPO layer before any
online evidence retrieval. Disease alias normalization, phenotype alignment,
xref resolution, and the SQLite query scripts live here.
Skill file: rd_ontology_first.md
When to use:
- Normalizing free-text phenotype or disease aliases (researcher,
ontology mode)
- Resolving ORPHA/OMIM/MONDO/HPO cross-IDs before retrieval (researcher,
ontology mode)
- Any candidate-name standardization step
Clinical Report Format
The authoritative format contract for the final clinical genetics consult report:
cover page, 4-level numbering, 9 sections, sign-off block, machine-readable JSON,
and the standalone HTML theme (embedded CSS, weasyprint PDF path).
Skill file: clinical_report_format.md
When to use:
- Producing the final report (leader, after audit)
- Any time a deliverable must follow the signed-report structure and theme
IMPORTANT: This file is the single source of truth for report layout and
styling. The leader must follow it exactly rather than improvising structure
or CSS.
Guardrails (apply across all rare disease skills)
- Ontology normalization comes before retrieval, never after.
- Do not fabricate xref IDs, PMIDs, DOIs, or database annotations.
- Do not present online evidence as ontology truth.
- The report is decision-support, never a definitive diagnosis.
