ML Pipeline Creation — reproducible pipelines from data to deployment gates
A platform-neutral method for designing reproducible ML pipelines — stage DAGs with explicit artifact contracts, idempotency, schema validation, evaluation/promotion gates, observability, rollback, and incremental testing
- What
- A platform-neutral method for designing reproducible ML pipelines — stage DAGs with explicit artifact contracts, idempotency, schema validation, evaluation/promotion gates, observability, rollback, and incremental testing
- Cost
- Free
- Needs
- an existing ML project with data sources and a target environment; an orchestration or CI system you already use (Airflow, Kubeflow, or plain CI runners)
- Install
- Copy the installer prompt below into your Muse — your agent does the rest.
Curated by Skill Harbor — @seb1n's ml-pipeline-creation skill, listed here with credit to its creator: a platform-neutral method for an agent to design, implement and validate reproducible machine-learning pipelines — starting from required inputs (business objective, measurable acceptance criteria, data ownership and refresh cadence, target environments, compute/cost/compliance constraints), then an output contract (stage dependency graph, versioned pipeline definition, explicit input/output schemas, data/model/environment versioning rules, evaluation and promotion gates with failure behavior, observability/retry/backfill/rollback procedures, and a verification record). The workflow stresses inspecting the environment before picking tools ("don't introduce a platform merely to demonstrate one"), modeling the DAG as idempotent stages with declared artifacts, pinning dependencies and seeds, quality gates that fail closed, retries only for transient failures, and incremental testing — including forcing one stage to fail to verify downstream stages never execute silently. It also documents edge cases (streaming data, non-deterministic training, large backfills, schema drift, partial promotion). Honest caveats: methodology guidance, not a ready-made orchestrator — your team supplies the actual platform (Airflow, Kubeflow, CI runners); safety boundaries are spelled out (no production data in dev without authorization, no production promotion without explicit approval) but the agent can't enforce them for you — review every plan. MIT licensed. Skill Harbor never reviews the code, review it yourself before use. Discovered via skills.sh.
Version:
Install
Prerequisites: an existing ML project with data sources and a target environment; an orchestration or CI system you already use (Airflow, Kubeflow, or plain CI runners) Install "ML Pipeline Creation — reproducible pipelines from data to deployment gates" for me. It gives my agent @seb1n's platform-neutral method for designing reproducible ML pipelines: required-inputs checklist, output contract (stage DAG, artifact schemas, versioning rules, evaluation/promotion gates, observability/rollback procedures), a workflow that prefers existing orchestrators, idempotency and fail-closed quality gates, incremental testing, plus safety boundaries and edge-case handling. MIT licensed. Repository: https://github.com/seb1n/awesome-ai-agent-skills/blob/main/ai-ml-operations/ml-pipeline-creation/SKILL.md 1. Fetch the SKILL.md file (and any helper files) from the repository path into a temporary folder and summarize what it does in one or two sentences. 2. Safety check: review the SKILL.md and scripts for anything suspicious (unexpected network calls, shell commands, credential harvesting). This is a methodology skill — any executable action beyond documentation is a red flag. This repo should contain zero secrets in code. Verify that holds here; STOP on any red flag and tell me. 3. Install it as a skill: copy SKILL.md and its helper files into the agent's skills directory, in a folder named "ml-pipeline-creation". 4. Verify with no network calls: frontmatter valid, files in place. 5. Report what was installed, where, and what I still need to do myself (e.g. pick my orchestrator, define my acceptance criteria and promotion gates, and review every pipeline plan myself — the agent cannot enforce the safety boundaries for me). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.
Questions
How do I install a build?
Every product page includes a copy-paste install prompt. Paste it into your Muse and it sets the build up for you — no manual configuration.
Where does my money go?
Straight to the seller. Skill Harbor never processes payments: checkout happens on the seller’s own page, usually Stripe.
What does the ✓ next to a creator’s name mean?
It means we confirmed the identity of the person behind the listing. It says nothing about the code itself — always check a build before installing it.