What are Claude Code Skills?

Claude Code Skills are reusable prompt templates stored as markdown files in your project's .claude/skills/ directory. They let you codify best practices and common workflows into slash commands that Claude follows consistently.

How do I create a Claude Code Skill?

Create a markdown file in .claude/skills/ with instructions for Claude. The filename becomes the command name — for example, .claude/skills/code-review.md can be invoked with /code-review during a Claude Code session.

Can I share skills with my team?

Yes. Skills stored in .claude/skills/ can be committed to version control so every team member has access to the same standardized workflows. You can also organize skills into subdirectories by category.

What is the difference between project and global skills?

Project skills (.claude/skills/) are specific to the current project and are committed to git. Global skills (~/.claude/skills/) are personal skills available across all your projects.

ctsm-skill

New

3GitHubDocumentationby earth-space-ai

Deep skill / knowledge package for the Community Terrestrial Systems Model (CTSM/CLM). SKILL.md routing hub plus reference/ deep-dive docs covering architecture, CESM case workflow, physics options, LILAC coupling, history output, debugging, and contributing.

Community PluginView Source

Overview

CTSM, Community Terrestrial Systems Model, Complete Guide

CTSM = Community Terrestrial Systems Model (includes the Community Land Model, CLM, of CESM)
Maintainers: NCAR/CGD CTSM software engineering and science teams
Source: https://github.com/ESCOMP/CTSM
Docs: https://escomp.github.io/CTSM/
Forum: https://bb.cgd.ucar.edu/cesm/forums/ctsm-clm-mosart-rtm.134/
Latest science version covered here: ctsm5.4.x (CLM6 physics default)

What CTSM does: Solves coupled land surface energy, water, carbon, and nitrogen budgets on a subgrid hierarchy (gridcell, landunit, column, patch). Represents vegetation phenology and biogeochemistry, snow and soil hydrology, glaciers, lakes, urban canyons, crops with irrigation and prognostic phenology, river routing (MOSART or mizuRoute), and optional ecosystem demography (FATES). Runs offline with CDEPS data atmosphere (I compsets), coupled inside CESM with CAM, or coupled to a host atmosphere model through LILAC.

Who this skill is for: Agents, students, and researchers who want to clone, build, run, modify, debug, and contribute to CTSM.

Quick Decision Tree

code

"What do I need?"
│
├─ First time, what is CTSM and how do I clone and build it?
│  └─ Read: reference/getting-started.md
│     (git clone, git-fleximod update, supported machines, py_env_create)
│
├─ I want to run a standard CTSM case (offline, I compset)
│  └─ Read: reference/running-cases.md
│     (create_newcase, case.setup, case.build, case.submit, xmlchange)
│
├─ I want to understand how the source tree is organized
│  └─ Read: reference/architecture.md
│     (src/biogeochem, src/biogeophys, src/main, src/dyn_subgrid, FATES, cpl)
│
├─ I want to pick a physics version, BGC vs SP vs FATES, or turn on crop / fire
│  └─ Read: reference/physics-and-options.md
│     (CLM6 vs CLM5 vs CLM4.5, BGC, SP, FATES, MOSART, urban, lake, glacier)
│
├─ I want to control history output, add fields, change frequency
│  └─ Read: reference/output-history-fields.md
│     (h0a/h0i tapes, hist_fincl, hist_nhtfrq, hist_mfilt, hist_dov2xy)
│
├─ I want to couple CTSM to my own atmosphere model (no CESM, no CAM)
│  └─ Read: reference/lilac-coupling.md
│     (Lightweight Infrastructure for Land Atmosphere Coupling, ESMF API)
│
├─ I want to make a surface dataset, subset to a site, or modify inputs
│  └─ Read: reference/python-tools.md
│     (mksurfdata_esmf, subset_data, fsurdat_modifier, mesh_maker, ctsm_pylib)
│
├─ I want to submit a pull request to ESCOMP/CTSM
│  └─ Read: reference/contributing-pr.md
│     (fork, feature branch, ChangeLog, ChangeSum, ExpectedTestFails, run_sys_tests)
│
└─ The model won't build, won't run, or gives wrong answers
   └─ Read: reference/debugging.md
      (build failures, externals desync, runtime crashes, restart issues, baselines)

What's Inside CTSM

code

CTSM/                                        ← top-level (this is $CTSMROOT)
├── README, README.md, README_GITFLEXIMOD.rst
├── WhatsNewInCTSM5.4.md                     ← release notes for ctsm5.4.x
├── CONTRIBUTING.md                          ← how to open issues and PRs
├── bin/git-fleximod                         ← submodule package manager
├── py_env_create                            ← creates ctsm_pylib conda env
├── ccs_config/                              ← CIME grids, compsets, machines
├── cime/                                    ← CIME (Common Infrastructure for Modeling Earth)
├── cime_config/
│   ├── config_compsets.xml                  ← I compsets, etc. (~130 compsets)
│   ├── config_component.xml                 ← CTSM XML variables (CLM_*)
│   ├── config_pes.xml                       ← processor layouts
│   ├── config_tests.xml                     ← CTSM-specific test types
│   ├── user_nl_clm                          ← starter user namelist (copied into each case)
│   ├── usermods_dirs/clm/                   ← named user-mod directories (e.g., NEON sites)
│   └── testdefs/ExpectedTestFails.xml       ← known-failing tests
├── components/
│   ├── cmeps/                               ← NUOPC mediator (default driver)
│   ├── cdeps/                               ← data models (DATM, DICE, DOCN, DROF, DWAV)
│   ├── cism/                                ← Community Ice Sheet Model
│   ├── mosart/                              ← Model for Scale Adaptive River Transport
│   ├── mizuroute/                           ← reach-based river transport
│   └── rtm/                                 ← legacy River Transport Model
├── doc/
│   ├── ChangeLog, ChangeSum                 ← per-tag history (UpdateChangeLog.pl)
│   ├── IMPORTANT_NOTES.md                   ← unsupported and experimental options
│   ├── Quickstart.GUIDE.md                  ← tiny NUOPC quick start
│   ├── README.NUOPC_driver.md               ← MCT to NUOPC migration notes
│   ├── source/                              ← Sphinx docs source
│   └── design/                              ← software design docs
├── bld/
│   ├── build-namelist                       ← Perl driver
│   ├── CLMBuildNamelist.pm                  ← namelist generator
│   └── namelist_files/
│       ├── namelist_definition_ctsm.xml     ← every CLM namelist variable
│       └── namelist_defaults_ctsm.xml       ← per-physics defaults
├── lilac/                                   ← Lightweight Infrastructure for Land Atmosphere Coupling
│   ├── src/                                 ← LILAC Fortran source
│   ├── atm_driver/                          ← bundled atm driver for testing LILAC
│   ├── build_ctsm                           ← script to build CTSM as a static library
│   ├── make_runtime_inputs                  ← script to stage runtime inputs
│   └── docs/
├── src/                                     ← CTSM Fortran source (~370 .F90 files)
│   ├── main/                                ← clm_driver, clm_initializeMod, controlMod, histFileMod
│   ├── biogeochem/                          ← CN cycle, fire, FUN, MEGAN, satellite phenology
│   ├── biogeophys/                          ← canopy, soil, snow, urban, lake, photosynthesis
│   ├── soilbiogeochem/                      ← decomposition, MIMICS
│   ├── dyn_subgrid/                         ← landuse change, harvest, dynamic PFT
│   ├── init_interp/                         ← online interpolation of initial conditions
│   ├── fates/                               ← FATES submodule (NGEET/fates)
│   ├── cpl/                                 ← driver coupling caps (nuopc, lilac, share_esmf)
│   ├── utils/                               ← clm_time_manager, restUtilMod, MatrixMod
│   ├── self_tests/                          ← internal tests run from a CTSM system test
│   ├── unit_test_shr/, unit_test_stubs/     ← pFUnit test scaffolding
├── tools/
│   ├── mksurfdata_esmf/                     ← parallel ESMF-based surface dataset generator
│   ├── site_and_regional/                   ← subset_data, mesh_maker, NEON and PLUMBER2 wrappers
│   ├── modify_input_files/                  ← fsurdat_modifier, mesh_mask_modifier
│   ├── crop_calendars/                      ← GGCMI sowing/harvest date processor
│   ├── mkmapgrids/                          ← legacy SCRIP grid scripts
│   └── contrib/                             ← user-contributed pre/post scripts
├── python/
│   ├── ctsm/                                ← Python package: subset_data.py, mesh_maker.py, etc.
│   └── conda_env_ctsm_py.yml                ← ctsm_pylib spec
├── share/                                   ← CESM shared code
└── libraries/                               ← PIO (deprecated copy)

For most users: create a case with create_newcase, edit user_nl_clm and xmlchange settings, build with case.build, submit with case.submit. Source code edits live under src/ (physics) and bld/namelist_files/ (namelist plumbing).

Critical Rules

**Always run ./bin/git-fleximod update after a fresh clone or after

checking out a new tag.** CTSM uses git-fleximod (not plain git submodules) to manage CIME, CMEPS, CDEPS, CISM, MOSART, mizuRoute, RTM, FATES, ccs_config. Without this step the build fails with missing components.

``bash git clone https://github.com/ESCOMP/CTSM.git my_ctsm_sandbox cd my_ctsm_sandbox ./bin/git-fleximod update ``

Switching tags requires a re-update. After git checkout ctsm5.4.002,

rerun ./bin/git-fleximod update. The .gitmodules file pins each submodule to a fxtag, and your sandbox can drift if you forget.

Never edit `cime_config/user_nl_clm` directly for a case. That file is

the template copied into each new case directory. Edit the per-case user_nl_clm inside the case directory, not the repo copy.

CMEPS NUOPC is the only driver as of ctsm5.4. The MCT driver was

removed. Older docs that mention rpointer.drv, cpl.log, or DATM_CLMNCEP_YR_* apply to MCT. The current names are rpointer.cpl, med.log plus drv.log, and DATM_YR_*.

History tapes are now split into `hXa` and `hXi` files (ctsm5.4+).

h0 becomes h0a (averaged or non-instantaneous) and h0i (instantaneous). Post-processors that hard-code h0 filenames must be updated. See reference/output-history-fields.md.

Run cases under a scratch filesystem, not inside the source tree.

CIME creates $CASEROOT (case directory) and $RUNDIR/$EXEROOT (run and build) by default in user-specific scratch paths defined in ccs_config/cesm/machines/config_machines.xml. Do not override these to point inside the git checkout.

`mksurfdata_esmf` is currently only fully validated on Derecho. For

other machines use tools/site_and_regional/subset_data to derive regional and single-point surface datasets from existing global ones. See https://github.com/ESCOMP/CTSM/issues/2341.

For PRs touching FATES, update both repos. FATES lives in src/fates/

as a separate ESCOMP-managed submodule pinned to an fxtag in .gitmodules. Coordinate FATES tag bumps with the FATES team.

Quick Start (CTSM offline, I compset, on a CIME-supported machine)

bash

# 1. Clone and pull submodules
git clone https://github.com/ESCOMP/CTSM.git my_ctsm_sandbox
cd my_ctsm_sandbox
./bin/git-fleximod update

# 2. Create a case (CLM6 BGC with crop, 1850 control, 1 degree)
cd cime/scripts
./create_newcase --case ~/cases/i2000ClmBgcCrop \
                 --res f09_t232 \
                 --compset I2000Clm60BgcCrop \
                 --mach derecho

# 3. Configure, edit namelists, build, submit
cd ~/cases/i2000ClmBgcCrop
./case.setup
$EDITOR user_nl_clm                  # add hist_fincl, finidat overrides, etc.
./xmlchange STOP_OPTION=nyears,STOP_N=5
./case.build
./case.submit

Output appears under $RUNDIR (see ./xmlquery RUNDIR). History files are named <case>.clm2.h0a.YYYY-MM.nc and <case>.clm2.h0i.YYYY-MM-DD-SSSSS.nc.

Full walkthrough in reference/getting-started.md and reference/running-cases.md.

Reference Documents

Document	What's inside
`reference/getting-started.md`	Repos, `git clone`, `git-fleximod update`, supported machines (Derecho, Casper, Izumi, Perlmutter), `py_env_create`, `ctsm_pylib`, common clone pitfalls
`reference/architecture.md`	Top-level layout, `src/biogeochem`, `src/biogeophys`, `src/main`, `src/dyn_subgrid`, `src/utils`, `src/cpl`, FATES submodule, subgrid hierarchy (gridcell, landunit, column, patch)
`reference/running-cases.md`	`create_newcase`, compset naming convention, `case.setup`, `case.build`, `case.submit`, `xmlchange`, `xmlquery`, `user_nl_clm`, `CaseDocs/`, resubmit, branch and hybrid runs, the `README.CHECKLIST.new_case` checklist
`reference/physics-and-options.md`	CLM6 vs CLM5 vs CLM4.5, `CLM_PHYSICS_VERSION`, `CLM_BLDNML_OPTS` (`-bgc sp`, `-bgc bgc`, `-bgc fates`, `-crop`), `LND_TUNING_MODE`, FATES options, MOSART vs mizuRoute, urban (TBD/HD/MD, LCZ), lake, glacier (CISM), fire (Li2014/2016/2021/2024)
`reference/output-history-fields.md`	History file architecture, h0a/h0i split, `hist_fincl1..10` and `hist_fexcl*`, `hist_nhtfrq`, `hist_mfilt`, `hist_dov2xy` for 1D vector PFT/column output, single-point and PLUMBER2/NEON cases
`reference/lilac-coupling.md`	LILAC purpose, ESMF API, `lilac/build_ctsm`, `lilac/make_runtime_inputs`, `atm_driver` test, integration into a host atmosphere
`reference/python-tools.md`	`tools/mksurfdata_esmf` (gen_mksurfdata_build, gen_mksurfdata_namelist, gen_mksurfdata_jobscript), `tools/site_and_regional/subset_data` (point and region), `tools/modify_input_files/fsurdat_modifier`, `mesh_maker`, `ctsm_pylib` Python environment
`reference/debugging.md`	Build failures (missing externals, ESMF mismatch), runtime crashes (mesh / fsurdat mismatch, finidat resolution), `lnd.log` and `cesm.log`, restart pointer mismatches, `CLM_FORCE_COLDSTART`, baseline comparisons
`reference/contributing-pr.md`	Fork ESCOMP/CTSM, feature branch off master, `[email protected]`, opening an issue first, PR review, `doc/UpdateChangeLog.pl`, `ChangeSum`, `ExpectedTestFails.xml`, `run_sys_tests`, integrator workflow

Install & Usage

Create the skills directory

mkdir -p .claude/skills

Download the skill file

mkdir -p .claude/skills && curl -o .claude/skills/ctsm-skill.md https://raw.githubusercontent.com/earth-space-ai/ctsm-skill/main/SKILL.md

Invoke in Claude Code

/ctsm-skill

View source on GitHub

Frequently Asked Questions

What is ctsm-skill?

How to install ctsm-skill?

To install ctsm-skill, create the .claude/skills directory in your project, then run the curl command to download the skill file. Once installed, invoke it in Claude Code with /ctsm-skill.

What is ctsm-skill best for?

ctsm-skill is a community categorized under Documentation. Created by earth-space-ai.