just-vibeData engineering · inspect by default
/data-reconcile
Compare source and destination records and explain discrepancies
Use to compare corresponding datasets; db-integrity checks database invariants.
Make it your own.
In Claude Code, use the slash command and add your context. In Codex, select data-reconcile from the just-vibe skill picker, then send the same brief.
Version 0.11.0 also supports /jv data-reconcile, /just-vibe data-reconcile and /jv:data-reconcile in Claude. See shortcut setup and context examples.
/just-vibe:data-reconcile Compare these aligned snapshots by key, not just row counts./just-vibe:data-reconcile Reconcile exports with equal row counts but missing and duplicated IDs./just-vibe:data-reconcile Compare misaligned snapshots without calling timing differences data loss.What the agent does
- Align snapshots or time windows and key grain.
- Compare membership before values, normalizing only explicitly documented transformations.
- Sample discrepancies safely and explain likely causes.
Inputs
- source/destination snapshots, keys, transformations, tolerances, and comparison budget.
Optional context: scope, references, constraints, successCriteria, environment, mode, budget.
Scope
- Reads
- Missing, duplicated, changed, or aggregated discrepancies; no automatic repair.
- Writes
- No source changes in inspect/plan. Save only requested planning artifacts. data-pipeline or fix applies an accepted change.
- Mode
- Inspect; source/destination snapshots, keys, transformations, tolerances, and comparison budget.
- Prerequisites
- Data source/version, schema/semantics, transformation code, permitted sampling scope, and storage/compute budget. Prefer aggregates and redacted samples; never upload datasets to external services implicitly. Record time zones and snapshot identity for reproducibility.
Expected output
- Reconciliation report for the snapshot pair with keyed mismatch categories, denominators, evidence, repair candidates and evidence limits.
How the work is checked
- Equal row counts do not hide different records; allowed rounding/timing differences are distinguished from loss.
When to stop or clarify
- Misaligned snapshots prevent a definitive mismatch claim. No row-level data exposure beyond necessary authorized evidence.
Handling missing context
- Infer
- Inspect schema, source snapshot, transformation code, grain, time zones and permitted sample scope.
- Assume
- Use bounded synthetic or supplied samples when full data is unavailable; keep unknown values distinct from zero.
- Ask
- Resolve ambiguous entity/grain/time semantics before reconciliation or backfill; obtain missing data/compute limits only for the dependent scan or execution.
Technical guidance
- Evidence
- Align source/destination snapshots, key grain, time window, lag, normalization and delete semantics.
- Method
- Compare key membership then per-field values with explicit tolerances; isolate legitimate lag from corruption.
- Pitfall
- Equal counts or totals can hide missing and duplicated rows that cancel out.
- Check
- Use a fixture with equal totals but different membership and verify the report locates discrepancies without exposing sensitive values.
Situational decisions
When counts match but keys or values differ: Quantify each mismatch class and retain redacted examples rather than declaring parity.
The coding agent follows this workflow using its available tools. Installation does not grant service access or guarantee an outcome. Read the compatibility notes.