Repository navigation
v0.1.2
What's Changed
New Features
- Added
data stackfor assembling multiple local CSV, TSV, JSONL, or JSON files and directories into one.csv,.tsv, or.jsontable. - Added interactive
data stacksource discovery, mixed file/directory selection, generated default output paths, dry-run plan saves, and reviewed write boundaries. - Added replayable
data-stack-planartifacts plusdata stack replayfor rerunning reviewed stack plans. - Added stack diagnostics for schema drift, duplicate rows, candidate unique keys, selected duplicate policy, and replay fingerprint changes.
- Added optional reviewed Codex assistance for stack dry-runs and interactive checkpoints, including headerless column names, union exclusions, unique-key choices, duplicate policy choices, and schema-drift explanations.
Improvements
- Added
--schema-mode strict|union-by-name|autoso stack runs can keep strict matching by default, explicitly widen by name, or use deterministic automatic schema analysis. - Added JSON and JSONL stack support alongside CSV and TSV stack inputs.
- Added exact
--exclude-columnshandling for union-by-name stack outputs. - Refined interactive stack source discovery, status previews, artifact retention prompts, and replay tips.
- Updated the dependency baseline, including the
@xmldom/xmldomoverride used to clear the OOXML dependency security concern.
Bug Fixes
- Hardened dry-run artifact and output-path collision validation before stack writes.
- Rejected duplicate JSON output keys and unreplayable Codex report payloads instead of writing ambiguous artifacts.
- Preserved replayability for Codex exclusion and schema-mode recommendations after review.
- Kept excluded keys out of derived stack plans and remapped headerless unique keys onto reviewed output columns.
- Restored data-stack type imports after the TypeScript module split.
Changelog
Full Changelog: v0.1.1...v0.1.2