Etl data simulation
Self-correcting knowledge corpus for Claude Code — 9 stable shape clusters, bias-correction pipeline baked into contribution flow. 47 papers, 45 techniques, 1.1k skills.
npx -y skills add kjuhwa/skills-hub --skill etl-data-simulationAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Deterministic mock ETL data generation with realistic row counts, schemas, and job run histories
SKILL.md
1.8 KB, as published. Nobody here has run it
etl-data-simulation
ETL demo apps need three layers of synthetic data: schema definitions (tables with typed columns), row samples (10-50 rows per table showing realistic values, not lorem ipsum), and run history (timestamped job executions with durations, statuses, row counts). Use a seeded PRNG (mulberry32 with a fixed seed like 0x5E7ED) so every page reload shows identical data — this is critical for screenshots, demos, and debugging. Generate schemas for canonical domains users recognize instantly: orders, customers, products, events, sessions. Include a mix of PK/FK relationships, nullable columns, and at least one JSON/array column per schema to exercise edge cases.
For job run history, generate 30-90 days of runs with realistic patterns: daily batch at 02:00 UTC, occasional failures (5-10% rate), duration that scales with simulated row count, and late-arriving data spikes on Mondays. For etl-rule-builder specifically, every rule must have a "before/after" row-pair preview — generate 5 representative input rows and compute the output rows in real-time as the user edits the rule, showing NULL handling, type coercion, and cast failures explicitly.
Lineage apps need cross-table dependency graphs: 15-30 tables with an average of 2-3 upstream dependencies, creating a realistic DAG depth of 4-6 levels. Avoid cycles deterministically by generating tables in topological order and only adding edges from later-index to earlier-index tables. Store the whole simulated dataset in a single /src/mock/etlFixtures.ts file so it's trivially replaceable with a real API adapter later.