[alberttwong]

The deck listed 21 of 27 Flight SQL routes as candidates without retained tool-level execution evidence. Add repeatable workflows for all 27 entries, seven pinned dependency profiles, per-tool result gating, and explicit RStudio/Tableau/Metabase application acceptance records.

The saved run records 28 automated passes, two strict Datasets expected failures, and three application passes. It includes logs, JUnit reports, resolved versions and output artifacts. Isolate model tools in separate interpreters after a reproducible native-runtime abort; declare Quarto's PyYAML dependency.

Future-run instructions and the manually dispatched self-hosted workflow use the qualified Redis Enterprise TLS/authenticated environment required by current main. Historical results remain explicitly labeled as server c019afc with standalone Redis OSS 8.6.2; they do not qualify the rebased Enterprise composition.

Validation:

• Fresh-environment automated sweep: 28 passed, 2 strict expected failures; all 27 scoped route entries have evidence.
• Real RStudio/Tableau UI checks and Metabase API reads/generated SQL/write Actions.
• Rebased branch: community Go unit tests, two report unit tests, Python compilation, actionlint and Git whitespace checks passed.
• Live profiles were not rerun against the rebased Enterprise composition; the new self-hosted workflow has not been dispatched.

Tracks #57; both Datasets failures are reproductions of upstream huggingface/datasets#8912. The one-shard workaround passes. No universal application, distributed deployment or Flight TLS/authentication certification is claimed.

Review artifacts: intent, specification, plan, validation. Maintainer review pending.

Remove the mirrored docs/data-engineering-adbc-support.md and update README, client-guide and Superset links to the canonical Redis ADBC inventory. The corresponding canonical validation update is redis/redis-adbc-driver#285.