seed-data-strategist

What this is for

Empty databases are useless for testing. Teams waste days writing seed data that misses edge cases, duplicates production patterns incorrectly, or creates unrealistic distributions. This skill reads a schema and generates a comprehensive seed data strategy: which entities need what variety, what edge values matter, what cardinality to aim for.

The dominant failure mode is the seed data that covers only the happy path, leaving error paths and edge cases untested until production.

PROTECTION RULE - never ~/.claude/

Read-only skill. Guard required if write mode added later.

## What You Must Do When Invoked

Step 1

1. If -help is passed, print the ## Usage block below and stop.

Step 2

2. Confirm -ProjectDir is provided and the path exists.

Step 3

3. Run: scripts/seed-analyze.ps1 -ProjectDir "<path>"

Step 4

4. LLM reads the JSON output. For each entity:

Step 5

• What seed variants cover meaningful test scenarios?

Step 6

• List specific instances (e.g. "Customer with active subscription",

Step 7

"Customer with expired card", "Customer with maxed credit limit").

Step 8

• Ensure all FK chains have at least one complete path.

Step 9

• What edge values for nullable/enum/unique fields?

Step 10

5. Confidence: proven (from schema constraints), likely (inferred from

Step 11

column names), suspected (domain assumption).

Step 12

6. Write seed-strategy-report.md to the working directory.

Usage

/seed-data                         # interactive, prompts for directory
/seed-data <dir>                   # scan project directory
/seed-data -help                   # show usage

Returns JSON with entities[]: each entry {table, columns[], fkDependencies[], statusFields[], nullableFields[], uniqueFields[], enumValues[]} plus counts: {scannedFiles, totalEntities, totalColumns}.

Report Format

seed-strategy-report.md with:

• Executive summary (total entities, total columns, FK chains)
• Per-entity seed strategy (what variants, what edge values)
• FK chain coverage (which paths are complete, which are missing)
• Edge case recommendations (nulls, enums, boundaries)
• Confidence column for every finding
• Open questions (suspected, needs human review)