ccc-data
Load ONE skill. Get the entire data domain. From pipeline design to ML models to dashboards and vector search.
Sub-Skills
| # |
Skill |
Command |
Description |
| 1 |
data-pipeline |
/data-pipeline |
Design ETL/ELT pipelines with Airflow, dbt, Dagster |
| 2 |
sql-optimization |
/sql-optimization |
Optimize SQL queries, indexes, and execution plans |
| 3 |
data-visualization |
/data-visualization |
Create charts and dashboards with D3, Chart.js, Tremor, Recharts |
| 4 |
machine-learning |
/machine-learning |
ML model development with scikit-learn, PyTorch, TensorFlow |
| 5 |
data-quality |
/data-quality |
Data validation, schema enforcement, quality monitoring |
| 6 |
analytics-setup |
/analytics-setup |
Analytics implementation with PostHog, Mixpanel, GA4 |
| 7 |
reporting |
/reporting |
Automated report generation and scheduling |
| 8 |
vector-search |
/vector-search |
Vector database setup and semantic search with Pinecone, pgvector, Qdrant |
How To Use
Step 1: Tell me what data problem you're solving — pipeline, query, visualization, ML, or analytics.
Step 2: I'll confirm your data stack, volume expectations, and output requirements before routing.
Step 3: The specialist skill handles the work. You get full data engineering and science coverage without loading 8 separate skills.
Routing Matrix
| Your Intent |
Route To |
Don't Confuse With |
| "Build an ETL pipeline" / "Move data from A to B" / "dbt" |
data-pipeline |
data-quality (validation, not movement) |
| "This query is slow" / "Optimize my SQL" / "Add indexes" |
sql-optimization |
data-pipeline (data movement, not query tuning) |
| "Build a dashboard" / "Chart this data" / "D3 visualization" |
data-visualization |
reporting (scheduled docs, not interactive charts) |
| "Train a model" / "ML prediction" / "Classification" |
machine-learning |
analytics-setup (tracking, not prediction) |
| "Validate my data" / "Schema enforcement" / "Data freshness" |
data-quality |
data-pipeline (movement, not validation) |
| "Set up PostHog" / "Track user events" / "GA4" |
analytics-setup |
data-visualization (display, not collection) |
| "Weekly report" / "Automated PDF report" / "Scheduled digest" |
reporting |
data-visualization (interactive, not scheduled) |
| "Semantic search" / "Vector embeddings" / "pgvector" |
vector-search |
machine-learning (broader ML, not search-specific) |
Campaign Templates
Analytics Platform
analytics-setup -> implement event tracking (PostHog, Mixpanel, or GA4)data-pipeline -> set up ETL to move analytics data to warehousedata-quality -> validate incoming event data and enforce schemasdata-visualization -> build dashboards for key metricsreporting -> schedule automated weekly/monthly reports- Deliver: complete analytics platform from event capture to automated reporting
Data Warehouse Build
data-pipeline -> design ETL/ELT pipelines with dbt transformationssql-optimization -> optimize warehouse queries and materialized viewsdata-quality -> implement data contracts and quality monitoringdata-visualization -> build executive dashboardsreporting -> automate stakeholder reports- Deliver: production data warehouse with optimized queries, quality gates, and dashboards
ML Feature Pipeline
data-pipeline -> build feature engineering pipelinedata-quality -> validate feature data integrity and distributionsmachine-learning -> train and evaluate modelssql-optimization -> optimize feature store queriesvector-search -> set up embedding storage if using embeddings- Deliver: ML pipeline from raw data to trained model with quality gates
Semantic Search System
vector-search -> set up vector database (Pinecone, pgvector, or Qdrant)data-pipeline -> build embedding generation and indexing pipelinedata-quality -> monitor embedding quality and index freshnesssql-optimization -> optimize hybrid search queries (vector + SQL filters)- Deliver: production semantic search with automated indexing and quality monitoring
Data Stack Selection
What's your data stack?
|
+-- Need a data warehouse?
| +-- Already on Postgres? -> pgvector + dbt
| +-- Enterprise scale? -> Snowflake/BigQuery + dbt
| +-- Cost-sensitive? -> DuckDB + dbt
|
+-- Need analytics?
| +-- Open source + self-hosted? -> PostHog
| +-- Product analytics focus? -> Mixpanel
| +-- Marketing analytics? -> GA4
|
+-- Need ML?
| +-- Tabular data? -> scikit-learn
| +-- Deep learning? -> PyTorch
| +-- Production ML? -> PyTorch + ONNX
|
+-- Need vector search?
+-- Already on Postgres? -> pgvector
+-- Managed service? -> Pinecone
+-- Self-hosted + performant? -> Qdrant
Context Strategy
This CCC domain uses on-demand loading. Sub-skills have disable-model-invocation: true so they only load when explicitly invoked, keeping your context lean.
1---2name: ccc-data3description: CCC domain — complete data ecosystem — 8 skills in one. Data pipelines, SQL optimization, visualization, machine learning, data quality, analytics, reporting, and vector search.4---56# ccc-data78> Load ONE skill. Get the entire data domain. From pipeline design to ML models to dashboards and vector search.910## Sub-Skills1112| # | Skill | Command | Description |13|---|-------|---------|-------------|14| 1 | data-pipeline | `/data-pipeline` | Design ETL/ELT pipelines with Airflow, dbt, Dagster |15| 2 | sql-optimization | `/sql-optimization` | Optimize SQL queries, indexes, and execution plans |16| 3 | data-visualization | `/data-visualization` | Create charts and dashboards with D3, Chart.js, Tremor, Recharts |17| 4 | machine-learning | `/machine-learning` | ML model development with scikit-learn, PyTorch, TensorFlow |18| 5 | data-quality | `/data-quality` | Data validation, schema enforcement, quality monitoring |19| 6 | analytics-setup | `/analytics-setup` | Analytics implementation with PostHog, Mixpanel, GA4 |20| 7 | reporting | `/reporting` | Automated report generation and scheduling |21| 8 | vector-search | `/vector-search` | Vector database setup and semantic search with Pinecone, pgvector, Qdrant |2223## How To Use2425**Step 1:** Tell me what data problem you're solving — pipeline, query, visualization, ML, or analytics.2627**Step 2:** I'll confirm your data stack, volume expectations, and output requirements before routing.2829**Step 3:** The specialist skill handles the work. You get full data engineering and science coverage without loading 8 separate skills.3031## Routing Matrix3233| Your Intent | Route To | Don't Confuse With |34|-------------|----------|--------------------|35| "Build an ETL pipeline" / "Move data from A to B" / "dbt" | `data-pipeline` | `data-quality` (validation, not movement) |36| "This query is slow" / "Optimize my SQL" / "Add indexes" | `sql-optimization` | `data-pipeline` (data movement, not query tuning) |37| "Build a dashboard" / "Chart this data" / "D3 visualization" | `data-visualization` | `reporting` (scheduled docs, not interactive charts) |38| "Train a model" / "ML prediction" / "Classification" | `machine-learning` | `analytics-setup` (tracking, not prediction) |39| "Validate my data" / "Schema enforcement" / "Data freshness" | `data-quality` | `data-pipeline` (movement, not validation) |40| "Set up PostHog" / "Track user events" / "GA4" | `analytics-setup` | `data-visualization` (display, not collection) |41| "Weekly report" / "Automated PDF report" / "Scheduled digest" | `reporting` | `data-visualization` (interactive, not scheduled) |42| "Semantic search" / "Vector embeddings" / "pgvector" | `vector-search` | `machine-learning` (broader ML, not search-specific) |4344## Campaign Templates4546### Analytics Platform471. `analytics-setup` -> implement event tracking (PostHog, Mixpanel, or GA4)482. `data-pipeline` -> set up ETL to move analytics data to warehouse493. `data-quality` -> validate incoming event data and enforce schemas504. `data-visualization` -> build dashboards for key metrics515. `reporting` -> schedule automated weekly/monthly reports526. Deliver: complete analytics platform from event capture to automated reporting5354### Data Warehouse Build551. `data-pipeline` -> design ETL/ELT pipelines with dbt transformations562. `sql-optimization` -> optimize warehouse queries and materialized views573. `data-quality` -> implement data contracts and quality monitoring584. `data-visualization` -> build executive dashboards595. `reporting` -> automate stakeholder reports606. Deliver: production data warehouse with optimized queries, quality gates, and dashboards6162### ML Feature Pipeline631. `data-pipeline` -> build feature engineering pipeline642. `data-quality` -> validate feature data integrity and distributions653. `machine-learning` -> train and evaluate models664. `sql-optimization` -> optimize feature store queries675. `vector-search` -> set up embedding storage if using embeddings686. Deliver: ML pipeline from raw data to trained model with quality gates6970### Semantic Search System711. `vector-search` -> set up vector database (Pinecone, pgvector, or Qdrant)722. `data-pipeline` -> build embedding generation and indexing pipeline733. `data-quality` -> monitor embedding quality and index freshness744. `sql-optimization` -> optimize hybrid search queries (vector + SQL filters)755. Deliver: production semantic search with automated indexing and quality monitoring7677## Data Stack Selection7879```80What's your data stack?81|82+-- Need a data warehouse?83| +-- Already on Postgres? -> pgvector + dbt84| +-- Enterprise scale? -> Snowflake/BigQuery + dbt85| +-- Cost-sensitive? -> DuckDB + dbt86|87+-- Need analytics?88| +-- Open source + self-hosted? -> PostHog89| +-- Product analytics focus? -> Mixpanel90| +-- Marketing analytics? -> GA491|92+-- Need ML?93| +-- Tabular data? -> scikit-learn94| +-- Deep learning? -> PyTorch95| +-- Production ML? -> PyTorch + ONNX96|97+-- Need vector search?98 +-- Already on Postgres? -> pgvector99 +-- Managed service? -> Pinecone100 +-- Self-hosted + performant? -> Qdrant101```102103## Context Strategy104105This CCC domain uses on-demand loading. Sub-skills have `disable-model-invocation: true` so they only load when explicitly invoked, keeping your context lean.
KevinZai