1. Diagnose
Find row-shape, transport, header, whitespace, NULL, and mixed-type hazards before SQL is generated.
Inspect row shape and inferred types, quote identifiers safely, preview conservative PostgreSQL or SQLite SQL, and export only a clean parse.
Structural errors and a contract FAIL block SQL generation. A contract PASS or WARN keeps export available. All table and column identifiers are double-quoted; text literals escape single quotes.
Analyze a CSV to see conservative type inference.
Paste a small local contract to compare expected columns, types, nullability, uniqueness, and allowed values. Use the header column,type,required,nullable,unique,allowed_values; separate allowed values with |. PASS and WARN permit export; FAIL visibly blocks the SQL preview and download until the contract or CSV is corrected.
Turn the analysis into a local Markdown handoff with the findings, inferred schema, transport notes, and a review checklist. It never includes a network destination.
Future Pro hypothesis: a bounded Import Readiness Pack could add destination-column mapping, normalization decisions, larger local batches, and a human-reviewed plan. Pricing and payment are not active.
A WARN does not certify an import. It explains a bounded risk while leaving the preview available; a FAIL blocks the preview until the reported contract or structural issue is addressed.
Find row-shape, transport, header, whitespace, NULL, and mixed-type hazards before SQL is generated.
Review conservative PostgreSQL or SQLite types and quoted identifiers without silently repairing the source.
Compare a local destination contract with the inferred source shape and see explicit mismatches.
Download SQL and a readable report so a teammate can review assumptions before an import.
This is a local diagnostic aid, not a database connection, migration service, schema guarantee, or production-import approval.