Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Get Started Free →

Data Preprocessing Pipeline · foryourhealth111-pixel | DecimalAI

foryourhealth111-pixel/preprocessing-data-with-automated-pipelines

NewSkillScoreMaintained

Design and implement repeatable preprocessing pipelines for cleaning, encoding, transforming, and validating ML input data.

Databy foryourhealth111-pixel

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Data Engineer pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/foryourhealth111-pixel-preprocessing-data-with-automated-pipelines/SKILL.md

decimalai skills pull foryourhealth111-pixel-preprocessing-data-with-automated-pipelines --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/foryourhealth111-pixel-preprocessing-data-with-automated-pipelines/SKILL.md https://app.decimal.ai/s/foryourhealth111-pixel-preprocessing-data-with-automated-pipelines/SKILL.md

Raw file: /s/foryourhealth111-pixel-preprocessing-data-with-automated-pipelines/SKILL.md ↗ · machine-readable summary at /s/foryourhealth111-pixel-preprocessing-data-with-automated-pipelines.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 46d ago

—+1% tokens0% turns77% of whole cases pass with skill (17/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-03	✗→✓	▲ Improved	198%	0%
case-04	✗→✓	▲ Improved	-28%	0%
case-20	✗→✓	▲ Improved	-15%	0%
case-06	✓→✗	▼ Worse	12%	0%
case-07	✓→✗	▼ Worse	11%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

foryourhealth111-pixel / vibe-skills

bundled/skills/preprocessing-data-with-automated-pipelines/SKILL.md

View on GitHub ↗

License MITImported 8/3/2026Version v1Bundled 8 files · scripts

Skill contents

scripts/

references/

assets/

SKILL.md

Data Preprocessing Pipeline

Positioning

Use this skill as the direct owner for ML input-preparation pipelines.

It covers preprocessing-heavy tasks where the requested deliverable is a repeatable pipeline for cleaning, encoding, transforming, and validating input data.

When to Use

Use this skill when:

Prepare raw data for machine learning models.
Automate data cleaning and transformation processes.
Implement a robust ETL (Extract, Transform, Load) pipeline.

Not For / Boundaries

Whole-task ML ownership: use scikit-learn or ml-pipeline-workflow
Leakage and prediction-time auditing: use ml-data-leakage-guard
Grouped scientific preprocessing with stronger methodological constraints: use scientific-data-preprocessing

Typical Outputs

A preprocessing pipeline plan or implementation sketch
Clear sequencing for clean, encode, transform, and validate steps
Notes that identify where leakage review, training, or evaluation should be run next

Related Skills

ml-data-leakage-guard before trusting fitted preprocessing steps
splitting-datasets when the next narrow problem is partition strategy

Illustrative Want to see what foryourhealth111-pixel/preprocessing-data-with-automated-pipelines would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/13/2026, 05:55 UTC · 22 cases

Measured on 21 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

-20%

40% vs 60%

Agent turns

0%

1.0 vs 1.0

Tokens

+1%

2,413.24 vs 2,400.76

fail → pass · 3pass → pass · 14fail → fail · 3pass → fail · 2

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-03 We are processing single-cell RNA sequencing count matrices that require methodological batch-effect correction with strict scientific domain constraints and replicate grouping rules across medical trial cohorts. Design the ingestion script.	fail→pass	47,970	20,927	-56%	1	1	0%	1,423	4,237	+198%	0	0	—
case-01 We have a clean dataset ready in PostgreSQL and need an automated script to perform cross-validation grid search, fit a Random Forest model, and export the trained model binary. Provide the end-to-end Python implementation for this task.	fail→fail	14,654	12,983	-11%	1	1	0%	3,198	2,896	-9%	0	0	—
case-02 We already fitted a fitted TargetEncoder and StandardScaler pipeline on our entire dataset, but we suspect target leakage across cross-validation folds. Perform a detailed post-hoc audit to identify temporal and target feature leakage in our fitted artifacts.	pass→pass	27,805	24,138	-13%	1	1	0%	5,310	4,803	-10%	0	0	—
case-04 We are building an input pipeline for raw customer interaction logs with missing null values, raw strings, and skewed continuous counters before sending to an XGBoost model. Outline the canonical step order for processing raw features.	fail→pass	16,585	11,437	-31%	1	1	0%	3,004	2,162	-28%	0	0	—
case-05 When ingesting raw telemetry data containing missing sensor readings, unstripped text fields, and high-cardinality categorical IDs, team members disagree on whether to run one-hot encoding first or handle missing value imputation and whitespace stripping first. Specify the required sequence.	pass→pass	11,796	9,524	-19%	1	1	0%	2,021	1,730	-14%	0	0	—
case-06 For an automated housing price predictor pipeline, raw data contains string columns like 'neighborhood' and 'building_type' alongside raw numeric square footage. Should continuous scaling happen before or after categorical string encoding?	pass→fail	14,419	11,560	-20%	1	1	0%	1,945	2,184	+12%	0	0	—
case-07 In an automated tabular preprocessing pipeline, at what point should schema integrity, distribution checks, and missing-value zero-tolerance assertions be executed relative to cleaning, encoding, and scaling?	pass→fail	15,572	16,209	+4%	1	1	0%	2,738	3,026	+11%	0	0	—
case-08 After designing an automated preprocessing pipeline that fits encoders and scalers on training data, what critical downstream review must be highlighted before deploying these fitted transformers into model training?	pass→pass	11,481	8,231	-28%	1	1	0%	1,756	1,505	-14%	0	0	—
case-09 We have designed the stages for cleaning and encoding incoming raw user records. Before fitting transformers or scaling continuous features, what partition decision must be established to prevent data leakage?	pass→pass	6,719	7,658	+14%	1	1	0%	1,052	1,550	+47%	0	0	—
case-10 Create an architectural plan for an automated ETL pipeline transforming raw e-commerce transaction logs into ML-ready tensor arrays. Outline the major stages required for input preparation.	pass→pass	17,918	16,503	-8%	1	1	0%	3,028	3,073	+1%	0	0	—
case-11 In a financial transaction anomaly detection pipeline, raw transaction amounts are right-skewed and contain nulls. Many engineers want to scale values before cleaning nulls. Provide the correct ordering of steps for this continuous variable.	pass→pass	12,757	14,229	+12%	1	1	0%	2,117	2,508	+18%	0	0	—
case-12 Our project team needs a repeatable workflow artifact to handle raw CSV files containing user profiles, clean raw text, convert categorical fields into numeric representation, normalize continuous columns, and verify output constraints. What specific artifact should be delivered?	pass→pass	10,197	10,808	+6%	1	1	0%	1,655	2,367	+43%	0	0	—
case-13 When preparing ad-click log feeds for real-time model inference, where should post-transformation shape and null-check validation gates be placed relative to string encoding and scaling?	pass→pass	15,127	11,233	-26%	1	1	0%	2,506	2,068	-17%	0	0	—
case-14 A data engineering team is tasked with building a pipeline that converts raw PostgreSQL tables containing mixed data types into clean, encoded matrices ready for model consumption. Outline the core pipeline steps.	pass→pass	15,080	14,967	-1%	1	1	0%	2,441	2,673	+10%	0	0	—
case-15 In an automated feature pipeline for hospital admission prediction, categorical fields like 'primary_diagnosis' must be encoded into numeric formats. In what stage of the pipeline architecture should encoding rules be defined?	fail→fail	11,573	9,318	-19%	1	1	0%	1,827	1,794	-2%	0	0	—
case-16 You are writing the documentation summary for a newly authored input data preprocessing pipeline for tabular credit risk data. What operational notes should be included regarding downstream model training?	pass→pass	16,207	13,335	-18%	1	1	0%	2,587	2,469	-5%	0	0	—
case-17 A client requests a solution that takes raw unstructured CSV dumps, standardizes headers, imputes missing values, one-hot encodes ordinal columns, scales numerical ranges, and asserts matrix non-emptiness. Define the core functional components.	pass→pass	14,148	12,401	-12%	1	1	0%	2,251	2,745	+22%	0	0	—
case-18 In an automated pipeline processing raw vehicle telemetry (speed, fuel level, engine temp), raw values contain occasional -999 noise markers and varying scales. How should noise removal and feature scaling be ordered?	pass→pass	10,910	11,026	+1%	1	1	0%	1,793	2,212	+23%	0	0	—
case-19 In an automated preprocessing pipeline receiving streaming IoT sensor logs, what is the role of the validation step after cleaning, encoding, and transforming raw records?	pass→pass	12,124	9,731	-20%	1	1	0%	1,913	1,840	-4%	0	0	—
case-20 We need to construct a repeatable python preprocessing sequence that takes uncleaned multi-type tabular data and produces clean numerical arrays for tree-based modeling. Detail the required execution stages.	fail→pass	22,945	17,094	-26%	1	1	0%	4,411	3,738	-15%	0	0	—
case-21 A junior data scientist wants to apply standard normalization directly to a raw dataframe containing string state codes and unparsed date strings. Explain why this sequence is invalid and state the correct sequence.	pass→pass	9,751	11,354	+16%	1	1	0%	1,765	2,291	+30%	0	0	—
case-22 After defining an automated pipeline that outputs clean, encoded, transformed, and validated feature arrays, what specific follow-up step should be noted for team members responsible for model deployment?	fail→fail	6,821	4,966	-27%	1	1	0%	1,098	1,044	-5%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of -20 percentage points is the difference between those two pass rates over the 21 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 60%
Pass rate with the skill: 40%
Cases attempted: 22
Cases counted toward lift: 21
Cases that regressed: 3
Skill version measured: v1
Run completed: 8/13/2026, 05:55 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-04

Prompt

We are building an input pipeline for raw customer interaction logs with missing null values, raw strings, and skewed continuous counters before sending to an XGBoost model. Outline the canonical step order for processing raw features.

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signalsno signals yet

BenchmarkRun a verified benchmark with positive lift

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+5% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

No signals yet — run a benchmark or collect evaluated traces to earn a score.

Verified on gemini-3.6-flash · re-tested 45 days ago

Other measured skills in the registry, with their headline benchmark lift.