feat(commands): add /ds-data-review + /ds-data-mode — data-layer coverage (#23)#38
Merged
Merged
Conversation
…rage (gleicon#23) Two companion commands closing the data-layer gap, split so each does one job: - /ds-data-review (action): store-agnostic data-correctness/integrity/migration- safety audit. Owns the unowned question — is the data correct, consistent, and well-modeled? Lines drawn by consequence vs perf/security/bug-review. - /ds-data-mode (mode): build-time data-engineering discipline (idempotency, late/out-of-order data, schema drift, replay/backfill safety, DQ assertions). Tool-agnostic; stacks with tiger-style + test modes; complements the review. Registered in README, docs/commands.md (Reviews + Engineering modes), and docs/recipes.md (modes stack, pre-PR gate, new pipeline recipe, which-command-when).
…ETL audit From the /ds-grill-me pass on PR gleicon#38: - Add /ds-data-review as a conditional pass 5 in /ds-quality-gate-mode (it's a review, so it belongs in the gate's review-family spine, unlike the plan commands). Gated: runs only when the change touches schema/queries/ transactions/migrations, else skipped with a note. Six-pass → seven-pass across the command file, docs/commands.md, and README. - Add --pipelines to /ds-data-review: an additive sixth check area auditing ETL/pipeline code (idempotency, replay/backfill safety, late/out-of-order data, dedup, schema drift) — the after-the-fact audit counterpart to the /ds-data-mode build mode. Default scope (operational store) unchanged. - Weave /ds-data-review into the Reviews layered-gate intro as a conditional node. - Cross-link the mode <-> review --pipelines pairing in both command files, docs/commands.md, and the recipe.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #23.
Two companion commands closing the data-layer gap, split so each does a single job — one audits the operational store after the fact, the other shapes how pipelines get built.
/ds-data-review— action (correctness family)Store-agnostic (relational and NoSQL) audit owning the unowned question: is the data correct, consistent, and well-modeled? Checks schema & integrity, query-result correctness, transactions & consistency, and migration safety. Lines drawn by consequence, not feature:
/ds-perf-plan; query returning wrong/duplicate data → here/ds-security-review(assumes parameterized queries)/ds-bug-reviewOutput is a prioritized findings list (critical / wrong-results / integrity-gap / hardening); changes nothing. Spine: no finding without a concrete data hazard reachable in this schema/queries.
--pipelinesflag adds a sixth check area — data-pipeline / ETL correctness (idempotency, replay/backfill safety, late & out-of-order data, dedup, schema drift). Off by default (the locked default scope is the operational store); when on, it's the after-the-fact audit counterpart to the/ds-data-modebuild mode./ds-data-mode— mode (build-time complement)Tool-agnostic data-engineering discipline (batch + streaming) countering the naive ETL shape LLMs default to. Encodes constraints up front: idempotent transforms, late/out-of-order data (event-time windowing + watermarks), dedup, schema-drift contracts, replay/backfill safety, boundary DQ assertions that fail the run, E/T/L separation. Stacks with
/ds-tiger-style-mode+/ds-test-mode; its audit counterpart is/ds-data-review(store) and/ds-data-review --pipelines(pipeline code).Quality gate integration
/ds-data-reviewjoins/ds-quality-gate-modeas a conditional pass 5 (security → bug → data → code-quality), running only when the change touches schema/queries/transactions/migrations, else skipped with a note. It's a review, so it belongs in the gate's review-family spine (unlike the plan commands #21/#22, which stay out). Six-pass → seven-pass.Registration
docs/commands.md—/ds-data-reviewunder Reviews (+ layered-gate intro node),/ds-data-modeunder Engineering modes, quality-gate reference updateddocs/recipes.md— modes stack, pre-PR gate step, new "Building a data pipeline" recipe, two "Which command, when" rowsVerification
npm test→ 36 passed, 0 failed; integrity check sees 40 commands (was 38)bash install.sh --dry-runinstalls both files cleanly across Claude / OpenCode / Codex targetsDrafted from the locked proposals attached to #23; refined via
/ds-grill-me(quality-gate integration +--pipelinesaudit added in review).