epiverse-overview - SKILL.md Agent Skill

name: epiverse-overview description: Master orchestrator for Epiverse-TRACE infectious disease epidemiology skills. Use this skill FIRST to understand the ecosystem and route tasks to the correct specialized skill.

Epiverse-TRACE Overview & Orchestrator

[!IMPORTANT] CRITICAL: The Introspection Protocol Before generating any analysis code using these packages, you MUST execute the following discovery commands in R to verify the API and available data. Do NOT rely on your training data, as package versions and functions may have changed.

List Functions & Reference: Run ls("package:pkg_name") and help(package = "pkg_name") to see what functions exist and read the index.

Read README: Run try(print(readLines(system.file("README.md", package = "pkg_name")))) to understand the package scope and basic usage.

Check Vignettes: Run vignette(package = "pkg_name") to list available vignettes. Then read relevant vignettes using vignette("vignette_name", package = "pkg_name") to see standard workflows.

Read Signatures: Run args(function_name) or str(function_name) on the specific functions you plan to use.

Check Documentation: Read the internal help file using tools::Rd2txt(utils::.getHelpFile(utils::help("function_name", package = "pkg_name"))).

Only after confirming the function signatures, reading the documentation, and understanding the workflows should you generate the final R code.

The Epiverse-TRACE Philosophy

Epiverse-TRACE provides a pipeline for outbreak analytics:

Data Intake → Reading data from HIS and cleaning it
Parameters → Retrieving epidemiological parameters from literature
Simulation → Generating synthetic data for testing
Analysis → Estimating severity, transmissibility, and other metrics
Visualisation → Creating standardised, publication-ready plots
Reporting → Synthesizing results into publication-ready documents

[!IMPORTANT] START WITH THE END IN MIND: The goal is always a single Quarto document containing the complete workflow. When the user asks for an analysis, create ONE .qmd file that includes all code chunks for data cleaning, parameter retrieval, analysis, and visualization, with narrative text explaining each step. Do not create separate R scripts - embed everything in the Quarto document.

Skill Routing Guide

User Goal	Relevant Packages	Target Skill
Clean/standardize data	`cleanepi`, `linelist`, `readepi`	`data-intake`
Get epidemiological parameters	`epiparameter`, `epiparameterDB`	`parameters`
Simulate outbreak/contacts	`simulist`, `epichains`	`simulation`
Estimate CFR/Rt/final size	`cfr`, `finalsize`, `epidemics`	`analysis`
Create plots with Epiverse theme	`tracetheme`, `ggplot2`	`visualisation`
Generate publication report	Quarto, `gt`, `knitr`	`reporting`

Skill Dependencies

data_intake → parameters → simulation/analysis → visualisation → REPORT (end goal)

Pipeline Flow: All skills ultimately feed into reporting, which is the final deliverable.

Common Task Mappings

User Goal	Primary Skill	Notes
Clean messy line list	data-intake	-
Estimate outbreak severity	analysis	Requires cleaned data from data-intake
Compare interventions	analysis	Use epidemics package
Create epidemic curve	visualisation	Use cleaned data
Generate test data	simulation	Use parameters for realism
Find disease parameters	parameters	-
Complete outbreak analysis	epiverse-overview	Chain all skills → reporting

General Conventions

Style: Use tidyverse style (snake_case, |>)
Paths: Use here::here() for all file paths
Reproducibility: Set seeds (set.seed()) for stochastic operations
Dates: Always use ISO 8601 (YYYY-MM-DD)
Visualisation: Use ggplot2 with tracetheme::theme_trace()

Decision Tree

┌─ User Request ─┐
│                │
├─ "Clean/Import data" ────────────► data-intake skill
│
├─ "Get parameters" ───────────────► parameters skill
│
├─ "Simulate outbreak" ────────────► simulation skill
│
├─ "Estimate CFR/Rt/Final Size" ──► analysis skill
│
├─ "Create plot" ──────────────────► visualisation skill
│
├─ "Generate report" ──────────────► reporting skill
│
└─ "Complete analysis" ────────────► Use pipeline (multiple skills) → REPORT

[!NOTE] The Default Goal: Unless the user explicitly says otherwise, assume they want a complete publication-ready report as the final deliverable.