explore-data

Name: explore-data
Author: codeclawd

16stars6forks

How to install explore-data?

codeclawd/fable-mode/explore-data

$npx -y skills add codeclawd/fable-mode --skill explore-data

Installs into the current project.

Summary

Profile and explore a dataset to understand its shape, quality, and patterns. Use when encountering a new table or file, checking null rates and column distributions, spotting data quality issues like duplicates or suspicious values, or deciding which dimensions and metrics to an

Files · 1

View on GitHub

SKILL.md

1	# /explore-data - Profile and Explore a Dataset
2
3
4	Generate a comprehensive data profile for a table or uploaded file. Understand its shape, quality, and patterns before diving into analysis.
5
6	## Usage
7
8	```
9	/explore-data <table_name or file>
10	```
11
12	## Workflow
13
14	### 1. Access the Data
15
16	If a data warehouse MCP server is connected:
17
18	1. Resolve the table name (handle schema prefixes, suggest matches if ambiguous)
19	2. Query table metadata: column names, types, descriptions if available
20	3. Run profiling queries against the live data
21
22	If a file is provided (CSV, Excel, Parquet, JSON):
23
24	1. Read the file and load into a working dataset
25	2. Infer column types from the data
26
27	If neither:
28
29	1. Ask the user to provide a table name (with their warehouse connected) or upload a file
30	2. If they describe a table schema, provide guidance on what profiling queries to run
31
32	### 2. Understand Structure
33
34	Before analyzing any data, understand its structure:
35
36	Table-level questions:
37	- How many rows and columns?
38	- What is the grain (one row per what)?
39	- What is the primary key? Is it unique?
40	- When was the data last updated?
41	- How far back does the data go?
42
43	Column classification — categorize each column as one of:
44	- Identifier: Unique keys, foreign keys, entity IDs
45	- Dimension: Categorical attributes for grouping/filtering (status, type, region, category)
46	- Metric: Quantitative values for measurement (revenue, count, duration, score)
47	- Temporal: Dates and timestamps (created_at, updated_at, event_date)
48	- Text: Free-form text fields (description, notes, name)
49	- Boolean: True/false flags
50	- Structural: JSON, arrays, nested structures
51
52	### 3. Generate Data Profile
53
54	Run the following profiling checks:
55
56	Table-level metrics:
57	- Total row count
58	- Column count and types breakdown
59	- Approximate table size (if available from metadata)
60	- Date range coverage (min/max of date columns)
61
62	All columns:
63	- Null count and null rate
64	- Distinct count and cardinality ratio (distinct / total)
65	- Most common values (top 5-10 with frequencies)
66	- Least common values (bottom 5 to spot anomalies)
67
68	Numeric columns (metrics):
69	```
70	min, max, mean, median (p50)
71	standard deviation
72	percentiles: p1, p5, p25, p75, p95, p99
73	zero count
74	negative count (if unexpected)
75	```
76
77	String columns (dimensions, text):
78	```
79	min length, max length, avg length
80	empty string count
81	pattern analysis (do values follow a format?)
82	case consistency (all upper, all lower, mixed?)
83	leading/trailing whitespace count
84	```
85
86	Date/timestamp columns:
87	```
88	min date, max date
89	null dates
90	future dates (if unexpected)
91	distribution by month/week
92	gaps in time series
93	```
94
95	Boolean columns:
96	```
97	true count, false count, null count
98	true rate
99	```
100
101	Present the profile as a clean summary table, grouped by column type (dimensions, metrics, dates, IDs).
102
103	### 4. Identify Data Quality Issues
104
105	Apply the quality assessment framework below. Flag potential problems:
106
107	- High null rates: Columns with >5% nulls (warn), >20% nulls (alert)
108	- Low cardinality surprises: Columns that should be high-cardinality but aren't (e.g., a "user_id" with only 50 distinct values)
109	- High cardinality surprises: Columns that should be categorical but have too many distinct values
110	- Suspicious values: Negative amounts where only positive expected, future dates in historical data, obviously placeholder values (e.g., "N/A", "TBD", "test", "999999")
111	- Duplicate detection: Check if there's a natural key and whether it has duplicates
112	- Distribution skew: Extremely skewed numeric distributions that could affect averages
113	- Encoding issues: Mixed case in categorical fields, trailing whitespace, inconsistent formats
114
115	### 5. Discover Relationships and Patterns
116
117	After profiling individual columns:
118
119	- Foreign key candidates: ID columns that might link to other tables
120	- Hierarchies: Columns that form natural drill-down paths (country > state > city)
121	- Correlations: Numeric columns that move together
122	- Derived columns: Columns that appear to be computed from others
123	- Redundant columns: Columns with identical or near-identical information
124
125	### 6. Suggest Interesting Dimensions and Metrics
126
127	Based on the column profile, recommend:
128
129	- Best dimension columns for slicing data (categorical columns with reasonable cardinality, 3-50 values)
130	- Key metric columns for measurement (numeric columns with meaningful distributions)
131	- Time columns suitable for trend analysis
132	- Natural groupings or hierarchies apparent in the data
133	- Potential join keys linking to other tables (ID columns, foreign key

People who liked this also viewed

6 picks

Type

explore-data

16stars6forks

View on GitHub

How to install explore-data?

codeclawd/fable-mode/explore-data

$npx -y skills add codeclawd/fable-mode --skill explore-data

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# /explore-data - Profile and Explore a Dataset
2
3
4	Generate a comprehensive data profile for a table or uploaded file. Understand its shape, quality, and patterns before diving into analysis.
5
6	## Usage
7
8	```
9	/explore-data <table_name or file>
10	```
11
12	## Workflow
13
14	### 1. Access the Data
15
16	If a data warehouse MCP server is connected:
17
18	1. Resolve the table name (handle schema prefixes, suggest matches if ambiguous)
19	2. Query table metadata: column names, types, descriptions if available
20	3. Run profiling queries against the live data
21
22	If a file is provided (CSV, Excel, Parquet, JSON):
23
24	1. Read the file and load into a working dataset
25	2. Infer column types from the data
26
27	If neither:
28
29	1. Ask the user to provide a table name (with their warehouse connected) or upload a file
30	2. If they describe a table schema, provide guidance on what profiling queries to run
31
32	### 2. Understand Structure
33
34	Before analyzing any data, understand its structure:
35
36	Table-level questions:
37	- How many rows and columns?
38	- What is the grain (one row per what)?
39	- What is the primary key? Is it unique?
40	- When was the data last updated?
41	- How far back does the data go?
42
43	Column classification — categorize each column as one of:
44	- Identifier: Unique keys, foreign keys, entity IDs
45	- Dimension: Categorical attributes for grouping/filtering (status, type, region, category)
46	- Metric: Quantitative values for measurement (revenue, count, duration, score)
47	- Temporal: Dates and timestamps (created_at, updated_at, event_date)
48	- Text: Free-form text fields (description, notes, name)
49	- Boolean: True/false flags
50	- Structural: JSON, arrays, nested structures
51
52	### 3. Generate Data Profile
53
54	Run the following profiling checks:
55
56	Table-level metrics:
57	- Total row count
58	- Column count and types breakdown
59	- Approximate table size (if available from metadata)
60	- Date range coverage (min/max of date columns)
61
62	All columns:
63	- Null count and null rate
64	- Distinct count and cardinality ratio (distinct / total)
65	- Most common values (top 5-10 with frequencies)
66	- Least common values (bottom 5 to spot anomalies)
67
68	Numeric columns (metrics):
69	```
70	min, max, mean, median (p50)
71	standard deviation
72	percentiles: p1, p5, p25, p75, p95, p99
73	zero count
74	negative count (if unexpected)
75	```
76
77	String columns (dimensions, text):
78	```
79	min length, max length, avg length
80	empty string count
81	pattern analysis (do values follow a format?)
82	case consistency (all upper, all lower, mixed?)
83	leading/trailing whitespace count
84	```
85
86	Date/timestamp columns:
87	```
88	min date, max date
89	null dates
90	future dates (if unexpected)
91	distribution by month/week
92	gaps in time series
93	```
94
95	Boolean columns:
96	```
97	true count, false count, null count
98	true rate
99	```
100
101	Present the profile as a clean summary table, grouped by column type (dimensions, metrics, dates, IDs).
102
103	### 4. Identify Data Quality Issues
104
105	Apply the quality assessment framework below. Flag potential problems:
106
107	- High null rates: Columns with >5% nulls (warn), >20% nulls (alert)
108	- Low cardinality surprises: Columns that should be high-cardinality but aren't (e.g., a "user_id" with only 50 distinct values)
109	- High cardinality surprises: Columns that should be categorical but have too many distinct values
110	- Suspicious values: Negative amounts where only positive expected, future dates in historical data, obviously placeholder values (e.g., "N/A", "TBD", "test", "999999")
111	- Duplicate detection: Check if there's a natural key and whether it has duplicates
112	- Distribution skew: Extremely skewed numeric distributions that could affect averages
113	- Encoding issues: Mixed case in categorical fields, trailing whitespace, inconsistent formats
114
115	### 5. Discover Relationships and Patterns
116
117	After profiling individual columns:
118
119	- Foreign key candidates: ID columns that might link to other tables
120	- Hierarchies: Columns that form natural drill-down paths (country > state > city)
121	- Correlations: Numeric columns that move together
122	- Derived columns: Columns that appear to be computed from others
123	- Redundant columns: Columns with identical or near-identical information
124
125	### 6. Suggest Interesting Dimensions and Metrics
126
127	Based on the column profile, recommend:
128
129	- Best dimension columns for slicing data (categorical columns with reasonable cardinality, 3-50 values)
130	- Key metric columns for measurement (numeric columns with meaningful distributions)
131	- Time columns suitable for trend analysis
132	- Natural groupings or hierarchies apparent in the data
133	- Potential join keys linking to other tables (ID columns, foreign key

People who liked this also viewed

6 picks

Type