Skip to content

Repository files navigation

CURESLab — Claude Code skills for remote sensing & research workflows

A working set of Claude Code skills built around computational environmental research: remote sensing and NPP modelling, life-cycle assessment, optimisation, MATLAB engineering, literature search, and knowledge-base management.

These are not tutorials. Every pitfall documented here was found in production code that ran without error and produced plausible output.

Start here if you are new to the remote sensing pipeline

hermes-domain/skills/remote-sensing-agentic-workflow/references/loading_and_plotting.md

A step-by-step operational guide: probe the data, load only the window you need, mask the boundary and the water, regrid coarse meteorology with the right interpolator, save the rasters, and draw a map whose axes are actually correct. Each step ends with a check you can run.

It covers the things that are easy to get wrong and hard to notice:

  • dir() returning empty, so a fallback constant silently becomes every published number
  • geographicToIntrinsic returning [col, row] when you assumed [row, col]
  • poly2mask called per tile with a polygon that extends past the tile — a real case where built-up area came out 2.7x too large
  • inpolygon on a large raster (over 50 minutes, versus 0.06 s for scanline fill)
  • administrative boundaries that include territorial waters — measured at 50–53% of two real study areas
  • when TPS is the right interpolator and when it will stall your load
  • plot coordinates taken from a hardcoded table instead of the raster reference
  • max(x, 0) and sum(..., 'omitnan') quietly turning NaN into 0

Layout

Plugin Contents
hermes-domain/ Remote sensing, geospatial modelling, NPP/CASA, LCA, MILP, air-quality HPC, plotting
hermes-matlab/ MATLAB development, parallelisation, debugging, live scripts, MCP integration
hermes-research/ Academic writing, peer review simulation, deep research pipelines, Zotero
hermes-litsearch/ arXiv, OpenAlex, PubMed, Europe PMC literature search
hermes-notes/ Obsidian knowledge base: capture, compile, health check
hermes-infra/ MCP server setup
learned/ Earlier skill store, kept for reference
others Standalone utilities (anysearch, defuddle, markitdown, rag-engineering, windows-scripting-discipline, …)

Installing

Copy a plugin directory into your Claude Code skills folder:

# macOS / Linux
cp -r hermes-domain ~/.claude/skills/

# Windows
xcopy /E /I hermes-domain "%USERPROFILE%\.claude\skills\hermes-domain"

Each plugin carries its own .claude-plugin/plugin.json. Individual skills live under <plugin>/skills/<name>/SKILL.md and can be copied on their own if you prefer.

Notes

  • Paths in the examples are genericised (<NAS>, <PROJECT>). Substitute your own.
  • Scripts that call APIs requiring a contact address (Unpaywall, OpenAlex) read a placeholder email — set your own before use.
  • The MATLAB skills assume R2024a or newer; a few use R2025a+ features and say so.

Licence

MIT. See LICENSE.

About

Claude Code skills for computational environmental research: remote sensing & NPP modelling, LCA, MATLAB engineering, literature search, knowledge-base workflows

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages