Dataset reconciler
Skill nicholashidalgo/claude-skillforge/data-engineering/dataset-reconciler
npx -y skills add nicholashidalgo/claude-skillforge --skill dataset-reconcilerAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 1 stars1 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
Compare two datasets by key, isolate missing rows and field-level differences, and summarize reconciliation exceptions clearly.
SKILL.md
1.6 KB, as published. Nobody here has run it
Runtime Configuration
version: "1.0.0"
gotcha_pack: "sql-data-gotcha-pack"
gotcha_pack_version: "1.0.0"
gotcha_enforcement: "block_on_high"
Purpose
Compare two datasets and explain the differences clearly.
Workflow
- Define the comparison key and expected grain.
- Standardize data types and key formatting.
- Identify rows only in left, only in right, and in both.
- Compare important numeric and text fields.
- Bucket exceptions by issue type.
- Summarize count and amount deltas.
Output format
- Comparison setup
- Reconciliation summary
- Exception categories
- Python script
- Next action
Gotcha Enforcement
Every reconciliation script must satisfy these rules before output. HIGH violations block output. MEDIUM violations appear in Exception summary with an explanation.
| ID | Sev | Check |
|---|---|---|
| G003 | HIGH | Every aggregation documents NA/null behavior; sums must match treatment on both sides |
| G007 | HIGH | Reconciliation uses an independent access path; not re-running the same transform |
| G012 | HIGH | Confirm grain alignment, period alignment, and filter parity before comparing totals |
| G015 | MEDIUM | A net-zero variance triggers a mandatory segment-level breakdown before declaring clean |