research

name	research
description	Analyze data, investigate datasets, debug with data, and explore system behavior. Use when analyzing, running SQL queries, data profiling, investigating patterns, building statistics, exploring CSV/JSON data, or debugging with data evidence.
allowed-tools	Read, Bash, Grep, Glob
model	sonnet
user-invocable	true

Data Research Protocol

Principle: Data first, code second.

Jumping to conclusions without examining the data leads to building solutions for problems that don't exist. Every assumption needs verification with actual data.

Workflow

1. LOAD — Verify data is accessible (DB connection, file exists, API responds)
2. SCHEMA — Examine structure, types, relationships before analyzing
3. PROFILE — Check for risks: nulls, duplicates, outliers, wrong types
4. HYPOTHESIS — Formulate what you expect to find (and what would disprove it)
5. EXPERIMENT — Test ONE hypothesis at a time with a focused query
6. DOCUMENT — Record findings with evidence, not opinions

Schema Analysis

PostgreSQL

-- Schema inspection
SELECT column_name, data_type, is_nullable
FROM information_schema.columns WHERE table_name = 'target';

-- Data profiling
SELECT count(*), count(DISTINCT column_name),
       count(*) FILTER (WHERE column_name IS NULL) as nulls
FROM target;

-- Distribution
SELECT column_name, count(*) FROM target GROUP BY 1 ORDER BY 2 DESC LIMIT 20;

TypeScript

console.log(`Records: ${data.length}`);
console.log(`Keys: ${Object.keys(data[0] || {})}`);
const nullCount = data.filter(item => item.field == null).length;
const uniqueCount = new Set(data.map(item => item.field)).size;

Risk Profiling

Risk	How to detect
Missing data	Count nulls per column
Duplicates	Compare total vs distinct count
Wrong types	Check actual types vs expected
Outliers	Look at min/max, percentiles

Mini-Experiment Protocol

EXPERIMENT: [Description]
HYPOTHESIS: [What we expect]
METHOD: [Query or code]
RESULT: [Actual output]
STATUS: CONFIRMED / DISPROVED

Rules: one question per experiment, fast execution, logged results, compared with expectations.

Cognitive Bias Prevention

• Analyze ALL data, not just the first N records (survivorship bias)
• Actively look for evidence that DISPROVES your hypothesis (confirmation bias)
• Check edge cases and outliers, not just the median (anchoring bias)

Safety

• Read-only database queries (SELECT only)
• Always include LIMIT on exploratory queries
• Never modify production data during research