Imported from CU-ESIIL/analytics-library (AGENTS.md). Install upstream with npx skills add CU-ESIIL/analytics-library. Copyright stays with the author.

Analytics Library Agent Guide

This repository hosts reusable analytics workflows for environmental data science. It is the companion to the ESIIL Data Library: the Data Library helps users get data, and the Analytics Library helps users analyze data.

Purpose

• Host analysis workflows, not datasets.
• Prefer examples that operate on small inline data or data retrieved from public, keyless sources.
• Link to Data Library entries when a workflow naturally starts from one of those datasets.
• Preserve existing useful content while improving reproducibility, structure, and discoverability.

Required Analytic Structure

Every analytic entry should use this structure:

# Analysis Name
## What this analysis does
## When to use it
## Inputs
## R example
## Python example
## Minimum viable output
## Interpretation
## Limitations
## Tags

Code Expectations

Every analytic must include both R and Python implementations.

Examples must be:

• Copy-paste runnable.
• Free of API keys, tokens, passwords, and secrets.
• Free of hidden local file paths.
• Small enough to run quickly on a normal laptop.
• Built around functions, not one-off notebook state.

Functions should:

• Take simple inputs such as data, aoi, time_range, or plain vectors/tables.
• Run without manual setup beyond installing common public packages.
• Produce a visible result such as a plot, map, or summary.
• Return a useful object for reuse.

Reproducibility Standards

• Favor clarity over cleverness.
• Use minimal dependencies and name them explicitly.
• Include small example data directly in the page when possible.
• Keep outputs deterministic by setting seeds where randomness is used.
• Avoid downloads that require accounts, API keys, cookies, or secrets.
• Avoid large generated files in the repository.

Tagging

Use Google-like discoverability. Include relevant tags from these families:

• Method tags: regression, clustering, classification, forecasting, spatial-statistics, machine-learning.
• Data type tags: raster, tabular, time-series, vector, remote-sensing.
• Domain tags: climate, ecology, fire, water, biodiversity.
• Workflow tags: R example, Python example, beginner, reproducible.
• Synonym tags: include common alternate names users may search for.

Repository Hygiene

• Do not turn this repository into a data host.
• Do not duplicate Data Library dataset documentation.
• Do not add AI, sustainability, or working-group lifecycle essays unless they directly support an analytic workflow.
• Do not remove existing content unless it is broken or superseded by a tested replacement.
• Before finishing, run:

mkdocs build
python scripts/check_analytics_library_health.py
python scripts/check_navigation.py