conda-manager

April 9, 2026 · View on GitHub

A Claude Code plugin for managing, auditing, and optimising Conda environments. Use it to inventory your environments, spot duplicates and broken installs, validate GPU compute libraries, back up specs, and get AI-driven suggestions for a cleaner environment portfolio.

Commands

CommandDescription
/setup-condaVerify Conda is installed and properly configured; guide through setup if needed
/verify-condaDeep verification of the Conda installation, PATH, shell init, and dependency solver
/list-envsList all environments with sizes, Python versions, key packages, and inferred purposes
/check-env-healthHealth-check all environments for broken deps, conflicts, and bloat
/validate-env <name>Activate and test a specific environment end-to-end, including GPU access
/compare-envs <env1> <env2> [...]Side-by-side comparison of two or more environments
/find-duplicatesIdentify near-identical environments to consolidate and reclaim disk space
/suggest-envsRecommend new baseline or specialised environments based on your workflows
/cleanup-brokenFind and safely remove corrupted or non-functional environments
/backup-envs <name|all>Export environment specs to YAML files for safe recreation

GPU support

Commands that deal with GPU-accelerated packages detect available hardware automatically:

  • NVIDIA GPUs: checks CUDA version and PyTorch CUDA build
  • AMD GPUs: checks ROCm/HIP version and PyTorch ROCm build

Installation

claude plugin install danielrosehill/conda-manager-plugin

Or add to your project .claude/settings.json:

{
  "plugins": ["danielrosehill/conda-manager-plugin"]
}

Example usage

/list-envs
/check-env-health
/find-duplicates
/validate-env ml-base
/compare-envs base ml-gpu data-science
/backup-envs all
/suggest-envs
/cleanup-broken

License

MIT — Daniel Rosehill