live wire
▸Snowflake agent evals can pin judge versions; unpinned metrics may roll forward and shift scoresSnowflake▸ThoughtSpot makes Spotter 3 default for new instances; it requires download rights and always shares query data with an LLMThoughtSpot documentation▸MotherDuck adds image classification and object location to prompt_jev(), returning typed SQL outputs with confidence scoresMotherDuck▸BigQuery conversational analytics adds Montréal and Toronto; mixed-region agents still default to globalGoogle Cloud Documentation▸Fabric enables Azure OpenAI Copilot access by default, but OpenAI-subprocessor and cross-region processing/storage switches remain offMicrosoft Learn▸Snowflake Cortex Agents now continue by default when named tools are inaccessible, emitting warnings instead of failing the runSnowflake documentation▸Databricks Genie One can use group memberships to tailor answers, while Unity Catalog remains the access-control boundaryDatabricks documentation▸Fabric data agents can query Power BI models with Read alone; Build and workspace access are unnecessaryMicrosoft Learn▸Fabric data-agent Git configs expose column flags, but selecting a table still includes every columnMicrosoft Learn▸Google Data Agent Kit offers remote Spark kernels, but notebook exports can still move results into local Pandas or ArrowGoogle Cloud Documentation▸Cube says production analytics agents need explicit partial-result states, caller-filtered model search and durable artifactsCube▸Thinking Machines names Bridgewater behind ReViSQL-K2.6, citing 92.97% SC-16 accuracy at 15% of frontier-model costThinking Machines Lab▸AWS says a mismatched data/tool/prompt release cut its Field Sales agent pass rate from 78% to 38% in pre-productionAWS Artificial Intelligence Blog▸NL2SQL Multi-Agent ships evaluation artifacts across BIRD, CSpider, Archer and KDD DataSpace—but no software licenseGitHub▸Snowflake agent evals can pin judge versions; unpinned metrics may roll forward and shift scoresSnowflake▸ThoughtSpot makes Spotter 3 default for new instances; it requires download rights and always shares query data with an LLMThoughtSpot documentation▸MotherDuck adds image classification and object location to prompt_jev(), returning typed SQL outputs with confidence scoresMotherDuck▸BigQuery conversational analytics adds Montréal and Toronto; mixed-region agents still default to globalGoogle Cloud Documentation▸Fabric enables Azure OpenAI Copilot access by default, but OpenAI-subprocessor and cross-region processing/storage switches remain offMicrosoft Learn▸Snowflake Cortex Agents now continue by default when named tools are inaccessible, emitting warnings instead of failing the runSnowflake documentation▸Databricks Genie One can use group memberships to tailor answers, while Unity Catalog remains the access-control boundaryDatabricks documentation▸Fabric data agents can query Power BI models with Read alone; Build and workspace access are unnecessaryMicrosoft Learn▸Fabric data-agent Git configs expose column flags, but selecting a table still includes every columnMicrosoft Learn▸Google Data Agent Kit offers remote Spark kernels, but notebook exports can still move results into local Pandas or ArrowGoogle Cloud Documentation▸Cube says production analytics agents need explicit partial-result states, caller-filtered model search and durable artifactsCube▸Thinking Machines names Bridgewater behind ReViSQL-K2.6, citing 92.97% SC-16 accuracy at 15% of frontier-model costThinking Machines Lab▸AWS says a mismatched data/tool/prompt release cut its Field Sales agent pass rate from 78% to 38% in pre-productionAWS Artificial Intelligence Blog▸NL2SQL Multi-Agent ships evaluation artifacts across BIRD, CSpider, Archer and KDD DataSpace—but no software licenseGitHub
nl2sql.ai
releaseRELEASE

Chat2DB 5.3.6’s quiet fixes matter more than its import features

The release adds CSV mapping and MySQL transaction inspection, but operators should prioritize its SSH-log redaction and connection-close API change.

Before-and-after of leaked SSH logs and a GET-to-POST API change.
Side by side: what changed
By The News Desk· Sep 11, 2026the quick take — two AI hosts go live when you do

Chat2DB 5.3.6 is presented as a usability release: import previews, column mapping, configurable CSV parsing, MySQL transaction inspection and better execution-history filtering. For teams that run the open-source database client against production systems, however, the two most consequential changes sit near the bottom of the fix list: one removes SSH credentials from error logs, and the other changes a state-changing API call from GET parameters to a POST JSON body.

Patch the logging path first

The v5.3.6 release notes say SSH connection-error logs no longer include passwords, private-key passphrases or key-file fields. The linked pull request explains the failure path more precisely: when SSH or port-forwarding failed, SSHManager.getSSHSession logged ssh.toString(), while Lombok’s generated representation of SSHInfo included password, passphrase and key-file values.

The patch replaces that generated representation with a manual toString() that omits those three fields while retaining diagnostic connection details such as host, port, user and forwarding information. This is a narrow change, but it changes the upgrade priority for any team that forwards Chat2DB logs into shared support bundles or centralized observability systems. The source does not claim that credentials were exploited; it establishes that connection failures could write them to logs before the patch.

One API client change is required

The same release changes the connection-close endpoint from a GET request with query parameters to POST /api/connection/close with a JSON request body. Chat2DB explicitly tells API clients that call the endpoint to update their requests.

The implementation review says the controller now accepts the existing validated close-request object and delegates to unchanged connection-pool logic. It also records the validation performed before merge: one controller test, 10 connection-pool tests, a 48-module backend package, frontend lint/prebuild/build checks and real HTTP and SQL Server session checks. In those checks, GET requests and invalid bodies did not close sessions, while a valid POST closed only the selected datasource; repeated close and reconnect remained usable. Desktop JCEF transport was inspected statically rather than tested in a desktop runtime, an explicit remaining boundary.

The visible features are still useful

For day-to-day database work, v5.3.6 adds CSV and Excel previews and column mapping on web and desktop, plus controls for encoding, delimiters, row ranges and date or decimal formats. MySQL 5.7 and 8.0 users can inspect active InnoDB transactions, duration, SQL, lock waits and blocking connections, with explicit messages for unavailable metadata or insufficient privileges. Execution history can be filtered by datasource, database and schema and searched by SQL keyword.

The practical rollout order is therefore straightforward: upgrade to stop exposing SSH secrets through the documented error path, update integrations that call the close endpoint, then validate the new import and transaction-inspection workflows under the same database privileges used in production. That prioritization follows the release’s verified behavioral changes rather than its headline feature list.

Filed by The News 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.