Skip to content

Latest commit

 

History

2,799 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

GitHub Copilot Customisation

This directory contains custom agents and always-on instructions for GitHub Copilot Chat in VS Code.


Always-on instructions

instructions/r-style.instructions.md — applied automatically to every *.R file. Enforces tidyverse idioms, script-style (top-to-bottom, minimal abstraction), and comment standards. No invocation needed.


Custom agents (agents/)

Agents appear in the Copilot Chat agent picker (the mode selector next to the chat input). Select one from the dropdown, or type its name after @ in the chat input.

How to invoke

@<agent-name> <your request>

Example:

@stats review R/process_burn_dates.R for assumption violations
@commenter document R/get_clouds_wilson.R
@coder_alpha implement a function that downloads MODIS tiles for a given month

To run all three competing coders on the same problem, send the same prompt to each:

@coder_alpha implement X
@coder_beta implement X
@coder_gamma implement X

Then pick the best solution or synthesise across them.


Agent reference

Review agents — read-only, raise issues for coders to fix

Agent Invoke when you want to…
@stats Check statistical assumptions, sample sizes, data leakage, uncertainty
@geo Metadata QA (CRS consistency, resolution mismatches, missing files) and spatial accuracy review (resampling methods, spatial joins, zonal statistics)
@science Get a peer-reviewer-style critique of analytical design and methods
@gh_actions Check that changes will run on GitHub Actions (interactive calls, server paths, secrets, cleanup, date ranges)
@targets Check target dependencies, dynamic branching, format=file correctness, tar_release round-trip, deployment guards

Coding agents — produce working R code

Agent Philosophy
@coder_alpha Correctness-first: readable, conventional, explicit sanity checks
@coder_beta Performance-first: memory-efficient, chunk-based, parallelism-aware
@coder_gamma Minimalist: fewest lines, no abstraction, inline everything

Documentation agent

Agent Invoke when you want to…
@commenter Review or add scientific comments, section headers, and unit annotations

Orchestrator

Agent Invoke when you want to…
@ci Run a full pre-commit sweep: lint + geo + stats + gh_actions + targets

Typical workflows

Before committing:

@ci

Writing new processing code:

  1. Draft with @coder_alpha, @coder_beta, and @coder_gamma in parallel.
  2. Pick or synthesise the best solution.
  3. Run @geo and @stats on the result.
  4. Run @commenter to fill in documentation.

Reviewing a script for scientific soundness:

@science review R/process_dynamic_data.R
@stats review R/process_dynamic_data.R
@geo review R/process_dynamic_data.R

About

Environmental data processing for www.emma.eco

Resources

Stars

9 stars

Watchers

4 watching

Forks

Releases

Packages

Used by

Contributors

Languages