Cmf
Use when working with the Common Metadata Framework (CMF) — initializing CMF in a project, instrumenting ML pipeline code, syncing metadata and artifacts with a CMF Server, querying lineage and artifacts, or setting up the CMF MCP server for AI assistant integration. Routes to the appropriate specialized skill based on the task.From its SKILL.md
npx -y skills add HewlettPackard/cmf --skill cmfAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
SKILL.md
2.3 KB, as published. Nobody here has run it
Route to the appropriate sub-skill based on the user's task. If the task is ambiguous, ask one clarifying question before routing.
Quickstart (default path)
If the user wants to add CMF to their project for the first time, or the request is general/unclear, route here:
cmf-init — Install cmflib, configure a storage backend, and create the mlmd metadata store.
Routing Table
| Task | Sub-skill |
|---|---|
Install CMF, configure storage backend (local, S3, MinIO, SSH, OSDF), run cmf init | cmf-init |
Add Cmf() calls to existing ML pipeline code — contexts, executions, datasets, models, metrics | cmf-instrument |
| Push or pull metadata and artifacts using the CMF CLI | cmf-sync |
Query pipeline history, artifact lineage, and execution metadata using CmfQuery | cmf-query |
| Deploy CMF Server with Docker Compose, or connect to an existing shared server | cmf-server |
| Set up the CMF MCP server so an AI assistant can query CMF metadata | cmf-mcp |
Key Concepts
- Pipeline — top-level grouping of stages, identified by
pipeline_namepassed toCmf() - Context — a stage type (e.g.
"train","evaluate"), created withcreate_context() - Execution — one run of a stage, created with
create_execution(); hyperparameters go here - Artifact — dataset, model, or metrics file logged as input or output of an execution
- mlmd — the local SQLite file that stores all metadata; pushed to CMF Server for collaboration
Docs: Getting Started · cmflib API · CLI Reference · MCP Server