mlops-mlflow — Claude Code skill: track ML experiments with MLflow
A skill that gives Claude working instructions for MLflow: logging runs, versioning models in the registry, and reproducing experiments.
npx claude-code-templates@latest --skill ai-research/mlops-mlflowWhat mlops-mlflow does
The skill treats MLflow as an ML lifecycle platform and wires it into the agent's workflow. It ships Python examples: opening a run with mlflow.start_run(), recording parameters, metrics and artifacts, saving models from PyTorch, scikit-learn, Keras and HuggingFace Transformers.
Autologging gets its own section. A single mlflow.autolog() call, or a framework-specific one like mlflow.sklearn.autolog(), captures hyperparameters, metrics and the model, plus training duration. Model registry coverage includes versioning, stage transitions, and registering a model at log time with registered_model_name.
The examples assume mlflow, sqlalchemy and boto3 are already installed — skip them and the registry and artifact-storage snippets will not run. The local UI starts with mlflow ui and serves at http://localhost:5000.
What it helps with
- record parameters, metrics and artifacts for every training run
- compare model versions and their scores across runs
- register a model and move it between stages
- reproduce an experiment from a saved configuration
- turn on autologging for scikit-learn or PyTorch instead of logging by hand
Example prompts for the agent
- Wrap the training loop in
mlflow.start_run()and log learning_rate and batch_size - Enable
mlflow.sklearn.autolog()and fit a RandomForestClassifier on the train split - Register the model as
my-classifierwithregistered_model_namewhile logging
How to install mlops-mlflow
1claude-code-templates
npx claude-code-templates@latest --skill ai-research/mlops-mlflow2Skills CLI (Vercel)
npx skills add davila7/claude-code-templates --skill mlops-mlflow -a claude-code3Manually: put the files in place
~/.claude/skills/mlops-mlflow/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/mlops-mlflow/
Manually: put the files in place
.agents/skills/mlops-mlflow/SKILL.mdFor all projects use ~/.agents/skills/
Manually: put the files in place
.cursor/skills/mlops-mlflow/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install davila7/claude-code-templates mlops-mlflow2Manually: put the files in place
.github/skills/mlops-mlflow/SKILL.mdPlatform CLI
gemini skills install https://github.com/davila7/claude-code-templates --path cli-tool/components/skills/ai-research/mlops-mlflowManually: put the files in place
.devin/skills/mlops-mlflow/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about mlops-mlflow
How do I install mlops-mlflow in Claude Code?
Use the templates command: npx claude-code-templates@latest --skill ai-research/mlops-mlflow. Or run npx skills add davila7/claude-code-templates --skill mlops-mlflow -a claude-code. The skill file lands at ~/.claude/skills/mlops-mlflow/SKILL.md.
What does mlops-mlflow do?
It hands the agent MLflow guidance: experiment tracking, parameter and metric logging, a model registry with versions and stages, and autologging for common frameworks.
What has to be installed first?
The mlflow, sqlalchemy and boto3 packages. Without them the registry and remote artifact-storage examples will not run.
