live wire
▸Snowflake’s Cortex Agent quickstart routes metrics and transcripts with prompt rules—but documents no authorization boundarySnowflake▸Oracle Select AI generates SQL/PGQ for property graphs—but errors if one AI profile mixes graphs with tables or viewsOracle▸Domo’s data team standardizes MCP builds with shared scaffolding; beta toolkits inherit user permissionsDomo▸Databricks proposes five outcome signals for agentic BI—and says dashboard counts and license usage miss decision qualityDatabricks▸One vibe-coded database app needed 56+ human test hours and three repair rounds despite AI Studio’s verification claimsMancas & Mancas / arXiv▸BigQuery AI.IF’s default cost mode trains a local model for most rows—and drops row-error thresholdsGoogle Cloud▸AWS’s multi-tenant data-agent pattern gives the broker Lambda no data permissions; Cognito roles carry Lake Formation scopeAWS Big Data Blog▸Amazon OpenSearch’s optimized engine adds nested agent traces, but AWS reserves it for log analytics—not vector searchAWS▸Two interpretation-based provenance semantics cannot express SQL bag difference, new theorem showsKolaitis / arXiv▸Databricks Feature Views reuse one definition for training and live scoring, but online lookup still returns latest available stateDatabricks▸Databricks Genie’s Slack and Teams previews default channel answers to public, even when viewers lack source-data accessDatabricks▸Snowflake’s two-tool GraphRAG agent beat a seven-tool graph setup 70% to 60% on 22 ontology questionsSnowflake Engineering▸Semantix opens an Apache-2.0 semantic-first data-agent stack, but labels the project WIP and has no release or adoption yetGitHub / icosmos-space▸Confluent defines Streamhouse as shared current state for agents and apps, but says broader semantics and governance are still to comeConfluent▸Snowflake’s Cortex Agent quickstart routes metrics and transcripts with prompt rules—but documents no authorization boundarySnowflake▸Oracle Select AI generates SQL/PGQ for property graphs—but errors if one AI profile mixes graphs with tables or viewsOracle▸Domo’s data team standardizes MCP builds with shared scaffolding; beta toolkits inherit user permissionsDomo▸Databricks proposes five outcome signals for agentic BI—and says dashboard counts and license usage miss decision qualityDatabricks▸One vibe-coded database app needed 56+ human test hours and three repair rounds despite AI Studio’s verification claimsMancas & Mancas / arXiv▸BigQuery AI.IF’s default cost mode trains a local model for most rows—and drops row-error thresholdsGoogle Cloud▸AWS’s multi-tenant data-agent pattern gives the broker Lambda no data permissions; Cognito roles carry Lake Formation scopeAWS Big Data Blog▸Amazon OpenSearch’s optimized engine adds nested agent traces, but AWS reserves it for log analytics—not vector searchAWS▸Two interpretation-based provenance semantics cannot express SQL bag difference, new theorem showsKolaitis / arXiv▸Databricks Feature Views reuse one definition for training and live scoring, but online lookup still returns latest available stateDatabricks▸Databricks Genie’s Slack and Teams previews default channel answers to public, even when viewers lack source-data accessDatabricks▸Snowflake’s two-tool GraphRAG agent beat a seven-tool graph setup 70% to 60% on 22 ontology questionsSnowflake Engineering▸Semantix opens an Apache-2.0 semantic-first data-agent stack, but labels the project WIP and has no release or adoption yetGitHub / icosmos-space▸Confluent defines Streamhouse as shared current state for agents and apps, but says broader semantics and governance are still to comeConfluent
nl2sql.ai
analysisDirectory watch

Cortex EDW’s code is inspectable—but its license and “production-grade” claims are not settled

A new multi-agent text-to-SQL repository has real guardrail code and tests. It also shows why directory curation must separate implementation evidence from launch language.

By The Tools Desk· Oct 22, 2026the quick take — two AI hosts go live when you do

A new text-to-SQL repository called Cortex EDW arrived on GitHub on October 22 with more substance than the many empty demos that appear in daily repository searches. Its tree contains a FastAPI service, Streamlit interface, DuckDB warehouse, retrieval layer, agent graph, governance modules and five test files. The repository metadata nevertheless shows zero stars, zero forks and no detected license at the time of review. GitHub repository metadata Repository tree

What is actually implemented

The most concrete part is the SQL policy layer. The committed ASTGovernanceGuard parses generated SQL with sqlglot, rejects non-SELECT roots and prohibited statement nodes, checks referenced tables against an allowlist, and injects or caps a row limit. That is meaningfully stronger than checking generated text for a handful of dangerous words, although the file still starts with a quick string-level multi-statement check before parsing. AST guard source

A separate groundedness module extracts numeric claims from the generated narrative and compares them with values in the executed dataframe and statistical summary. It allows absolute or relative tolerances and returns both a score and the unmatched numbers. This is a narrow numeric audit, not a general factuality proof: names and nonnumeric assertions are not validated by the implementation shown. Groundedness source

The README also reports 17 passing tests across governance, retrieval, orchestration, statistics and warehouse behavior. Those tests are committed in the repository, but there is no independent run or public benchmark attached to the launch. Project README Repository tree

Two claims curators should not copy

First, the README calls the project “production-grade” and frames it for petabyte-scale enterprise analytics, while the included executable warehouse is DuckDB with seeded example data. The repository provides no published petabyte benchmark, customer deployment or independent evaluation. Those phrases should remain attributed launch claims, not directory facts. Project README

Second, the README displays an MIT badge linking to LICENSE, but the recursive repository tree contains no license file and GitHub reports license: null. Until the author adds license text, users do not have the MIT grant the badge implies. Repository tree GitHub repository metadata

Cortex EDW belongs in the directory as an experimental, source-visible research project because the implementation is inspectable. It should not yet be labeled open source, production-proven or petabyte-tested. That distinction is not pedantry; it is the difference between documenting what code exists and repeating what a launch page hopes it will become.

sources

Filed by The Tools Desk. Corrections: desk@nl2sql.ai · Our standards →

comments · 0

    Comments are moderated before they appear. Your email is used once to confirm it is you — never shown, never sold. Corrections and questions get an answer from the desk when we have one.