菜单
Literature: retrieve document-level scientific evidence from publications, clinical-trial registries, and patent sources. Use for PubMed records, reproducible searches, trial registrations, patent evidence, abstracts, study metadata, stable identifiers, and citation provenance.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Biological knowledge: query curated assertions and reference databases for genes, variants, pathways, regulation, cell types, reference atlases, perturbation-derived relationships, phenotypes, therapeutic targets, drugs, interactions, annotations, signatures, compounds, and predicted structures. Use for identifiers, marker sets, gene sets, associations, ontologies, database facts, and cross-database coverage.
Experimental data: discover and acquire raw or near-raw measurements, sample-level observations, and reusable datasets from sequencing, transcriptomics, single-cell, perturbation, proteomics, metabolomics, genomics, and structural-biology repositories. Use for study or sample metadata, matrices, reads, spectra, experimental structures, and downloadable research artifacts—not curated database assertions or reference marker collections.
Parse and analyze multiple sequence alignments using Biopython. Extract sequences, identify conserved regions, analyze gaps, work with annotations, and manipulate alignment data for downstream analysis. Use when parsing or manipulating multiple sequence alignments.
Calculate alignment statistics including sequence identity, conservation scores, substitution matrices, and similarity metrics. Use when comparing alignment quality, measuring sequence divergence, and analyzing evolutionary patterns.
Perform multiple sequence alignment using MAFFT, MUSCLE5, ClustalOmega, or T-Coffee. Guides tool and algorithm selection based on dataset size, sequence divergence, and downstream application. Use when aligning three or more homologous sequences for phylogenetics, conservation analysis, or evolutionary studies.
Perform pairwise sequence alignment using Biopython Bio.Align.PairwiseAligner. Use when comparing two sequences, finding optimal alignments, scoring similarity, and identifying local or global matches between DNA, RNA, or protein sequences.
| name | literature |
| description | Literature: retrieve document-level scientific evidence from publications, clinical-trial registries, and patent sources. Use for PubMed records, reproducible searches, trial registrations, patent evidence, abstracts, study metadata, stable identifiers, and citation provenance. |
| category | literature |
| license | Mixed |
| metadata | {"access-mode":"remote","collection":"literature"} |
Build traceable evidence from publications, clinical trials, and patents while preserving each
document type's identifiers, search fields, status semantics, and evidentiary role. The resource
tool exposes source metadata and instructions; the Agent constructs the task-specific search.
Define the question, document type, date range, organism or population, intervention or exposure, study design, identifiers, language constraints, and required metadata. This step is complete when the search protocol and unresolved constraints are explicit.
Browse the relevant document source:
resource({ action: "list", category: "literature/publications" })
This step is complete when kind is resource_metadata and every returned source has been
compared against the protocol.
Load the selected source instructions using its exact exposed name:
resource({ action: "read", name: "biomarker-pubmed" })
This step is complete when the exact query interface, fields, pagination, identifiers, and output schema and package-relative scripts, references, and assets are known.
Execute reproducible fielded queries through all result pages. Preserve raw responses, query syntax, retrieval date, and stable identifiers such as PMID, NCT, and patent publication numbers. This step is complete when every page has been retrieved or a concrete failure is recorded.
Deduplicate by stable identifiers and label every record by document type and evidentiary role. Emit citations exclusively from retrieved metadata. This step is complete when every retained record has traceable provenance and every selected source has a result, valid empty result, or concrete failure.
Write search logs, raw metadata, normalized tables, and citation exports to the Session workspace; use package directories as read-only sources.
