agentleFS
mikeparcewskimikeparcewski/wicked-garden/skills/data/SKILL.md

wicked-garden-data

Skill9 starsChanged 7 months agoMITView on GitHub

Data engineering, analysis, and ML toolkit. One skill, routed sub-actions: analyze (interactive EDA on a CSV/Excel/data file), profile / validate / quality (schema-level data engineering ops), ml review / ml pipeline (model review + training-pipeline design), pipeline design / pipeline review (ETL architecture), ontology (recommend Schema.org, Dublin Core, DCAT, FOAF, GoodRelations, SKOS, or a custom shape for a dataset), shapes (a live source's data shapes, proposed for review).

  • Installs packages

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 4 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every sub-action runs inline: parse the arguments, pre-read the data where noted, load the Tier-3 rubric from refs/, apply it, and emit …

    line 28
  2. Distinctive:

    If none of these is available, or Python 3 is missing, skip the script-backed step, say so, and follow the manual alternative where one is …

    line 37
  3. Distinctive:

    Read first rows of the file to capture column names / types / nulls / sample.

    line 68
  4. Very specific:

    Read refs/analyze.md (relative to this skill's base directory) — the EDA rubric, quality/warehouse/ml modes, insight pattern, and output …

    line 69
  5. Distinctive:

    Read the data file head/tail to capture columns / types / nulls / sample.

    line 80
  6. Very specific:

    Read refs/data.md — the profile, validate, and quality rubrics with output formats and quality thresholds.

    line 81
  7. Very specific:

    Read refs/ml.md — the model review checklist, pipeline design template, deployment readiness checklist, and MLOps standards.

    line 110
  8. Very specific:

    Read refs/pipeline.md — the design checklist, review rubric with P1/P2/P3 findings, pattern selection, and engineering standards.

    line 123
  9. Very specific:

    Read refs/shapes.md — what to read (the source's catalog through knowledge.source_catalog, approved shapes, code and rules.

    line 151

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Data Domain — data engineering, analysis, and ML

Every sub-action runs inline: parse the arguments, pre-read the data where noted, load the Tier-3 rubric from refs/, apply it, and emit structured markdown with tables and prioritized findings. For delegated or parallel worker execution, hand off to the wicked-garden-data-engineer worker skill instead.

Hand-off — open the wicked-garden-data-engineer skill with the sub-action and its arguments as the argument; on Claude Code this is the Skill tool, on any other seat open the named skill from your catalog and carry it out inline, then continue here.

Runtime

Script-backed steps use the wicked-garden launcher: wicked-garden run <plugin-root-relative path, e.g. scripts/…> [args]. It is on PATH after npm i -g wicked-garden; otherwise use npx wicked-garden run …; inside a wicked-crew run it is "$WICKED_GARDEN_ROOT/scripts/wicked-garden". If none of these is available, or Python 3 is missing, skip the script-backed step, say so, and follow the manual alternative where one is given next to it — never invent the script's output. Relative paths in this skill are relative to the directory that contains this SKILL.md.

Routing

Sub-actionUse forRubric
analyze <file-path> [--focus stats|quality|warehouse|ml] [--context <file>] [--refresh] [--scenarios]one-off exploration of a CSV/Excel/data filerefs/analyze.md
profile <path>dataset structure + quality profilerefs/data.md
validate --schema <schema> --data <path>schema validationrefs/data.md
quality <path>quality report (completeness, uniqueness, validity)refs/data.md
ml review <path>ML model reviewrefs/ml.md
ml pipeline --type <classification|regression|ranking>training-pipeline designrefs/ml.md
pipeline design --source <src> --target <tgt> [--frequency <freq>]ETL pipeline designrefs/pipeline.md
pipeline review <path>ETL pipeline reviewrefs/pipeline.md
ontology <file-path>ontology recommendation for a datasetinline (§ Ontology)
shapes <source>propose a live source's data shapes (joins, describes, governed by) for reviewrefs/shapes.md

Detailed templates and examples: refs/analysis-templates.md, refs/ml-templates.md, refs/pipeline-templates.md, refs/data-examples.md.

Analyze — interactive data analysis

Interactive analysis on a CSV/Excel/data file. Use for one-off data exploration. NOT for schema-level checks (use profile / validate / quality) or pipeline review (use pipeline review).

  1. Parse <file-path>, --focus (stats|quality|warehouse|ml, default stats), --context, --refresh, --scenarios.
  2. Read first rows of the file to capture column names / types / nulls / sample.
  3. Read refs/analyze.md (relative to this skill's base directory) — the EDA rubric, quality/warehouse/ml modes, insight pattern, and output format.
  4. Apply the rubric directly for the chosen --focus mode and emit the analysis.

Profile / Validate / Quality — core data engineering ops

Schema-level engineering ops on a dataset. NOT for interactive exploration (use analyze) or ML pipeline review (use ml).

  1. Parse the sub-action (profile|validate|quality) and its args (<path>, --schema for validate).
  2. Read the data file head/tail to capture columns / types / nulls / sample.
  3. Read refs/data.md — the profile, validate, and quality rubrics with output formats and quality thresholds.
  4. Apply the rubric for the requested sub-action and emit structured markdown with tables and prioritized findings.

Optional scripted paths (deterministic profiling/validation):

wicked-garden run scripts/data/data_profiler.py \
  --input data.csv --output profile.json
wicked-garden run scripts/data/schema_validator.py \
  --schema schemas/expected.json \
  --data data/actual.csv

For files >1GB, use the analyze sub-action for efficient SQL-based profiling via DuckDB.

ML — model review and training-pipeline design

ML model review and training-pipeline design. NOT for ETL pipeline design (use pipeline) or data profiling (use profile).

  1. Parse the sub-action (review|pipeline) and args (<path> for review, --type for pipeline).
  2. For review, read model files at <path>.
  3. Read refs/ml.md — the model review checklist, pipeline design template, deployment readiness checklist, and MLOps standards.
  4. Apply the rubric for the requested sub-action and emit structured markdown.

Pipeline — data pipeline design and review

Data pipeline design and review. NOT for ML training pipelines (use ml pipeline) or one-off file analysis (use analyze).

  1. Parse the sub-action (design|review) and args. For review, read pipeline files at <path>. For design, capture --source, --target, --frequency.
  2. Read refs/pipeline.md — the design checklist, review rubric with P1/P2/P3 findings, pattern selection, and engineering standards.
  3. Apply the rubric for the requested sub-action and emit structured markdown.

Ontology — recommend ontologies for a dataset

Sample a dataset (CSV/Excel/Parquet/JSON) and recommend matching public ontologies (Schema.org, Dublin Core, DCAT, FOAF, GoodRelations, SKOS) or a custom shape. Use this for ontology mapping. NOT for interactive analysis (use analyze) or quality reports (use quality).

  1. Arg parse — extract file-path from the arguments.
  2. Run recommender:

    wicked-garden run scripts/_run.py scripts/data/ontology_recommender.py "${file_path}"
  3. Present the script's match table, column-mapping suggestions, and any custom-ontology fallback inline.

Shapes — propose a live source's data shapes for review

What a registered source's tables ARE: what each joins, which code describes it, which rule governs it. Shapes are PROPOSALS into the one review queue (a person approves each in Understand → Knowledge map), never facts you print or knowledge you write. NOT for a file on disk (use ontology or profile).

  1. Read refs/shapes.md — what to read (the source's catalog through knowledge.source_catalog, approved shapes, code and rules; never a row), the rules (never a value; never declare deps; your own repo's code) and how to propose.
  2. Build each shape with wicked-garden run scripts/data/shape_proposal.py <spec.json> --catalog <catalog.json> and send what it prints to the estate shim on stdin (wicked-garden run scripts/_estate_client.py call -). Without the launcher, build the request by hand as the ref shows.
  3. Report how many shapes wait for review, and where.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.