| name | literature-taxonomy-curation |
| description | Use when an agent must define viral genotype, subtype, clade, lineage, target/background groups, or PhyloGuide groups from cited literature before primer or tNGS panel design. |
Literature Taxonomy Curation
Use this skill before prepare-phylo or production primer design when group names affect the result. Do not infer genotype/subtype from sequence names alone.
Workflow
- Identify the exact species/virus, host, region/date scope, and design purpose.
- Search primary taxonomy/genotype papers, ICTV/NCBI/public-health references, and recent surveillance papers.
- Extract group names, representative accessions, distance thresholds or clade criteria, and citations.
- Write a taxonomy JSON with
species, optional target_group, references, and groups.
- Validate it:
python agentskill/scripts/curate_references.py validate-taxonomy \
--taxonomy-json taxonomy.json \
--output taxonomy_report.json
Optional PubMed search helper:
python agentskill/scripts/curate_references.py search-literature \
--species "Target virus" \
--extra-query "complete genome" \
--max-records 20 \
--output literature_candidates.json
Required Output
The downstream agent should receive:
taxonomy.json: curated groups and references.
taxonomy_report.json: validation status, warnings, and errors.
- A short rationale explaining why these groups are appropriate for the current species and use case.
If the species is missing or ambiguous, ask for clarification before assigning groups.