
Coding Agent Swarms, Part 5: Running the Fleet From Your Phone
The Last Mile Is the Operator The first four parts of this series built the substrate: foundation, fleet, multi-fleet …

DB Loadr v1.2.0 is live. This is a substantial release: five new database connectors bringing the total to 27, a completely redesigned query editor with syntax highlighting and autocomplete, a new AI query suggestion engine that actually understands your schema, and a table diff tool that makes data auditing genuinely fast. If you have been waiting for DB Loadr to feel production-ready as a daily driver, this is that release.
DB Loadr is a browser-based database studio that replaces the clutter of DataGrip, DBeaver, TablePlus, and pgAdmin with a single unified tool that runs inside the Calliope Workbench. 27 connectors, AI-assisted querying, centralized credentials through Zentinelle, and zero local installation required.
v1.2.0 adds native support for:
This brings the total connector count to 27. The full list now covers every major OLTP, OLAP, document, time-series, and cloud data warehouse in common use.
The query editor has been rebuilt from scratch. The new editor includes:
The previous editor was functional. This one is fast and genuinely pleasant to work in.
The AI query assistant now does schema-aware suggestion, not just natural language to SQL translation. Describe what you want, and the assistant generates the query, explains it line by line, and flags any potential performance concerns before you run it.
New in v1.2.0: the assistant can propose query optimizations for queries you have already written. Paste in a slow query, and it will suggest index hints, rewritten joins, or CTEs where appropriate. For teams doing heavy analytical work, this alone is worth the upgrade.
The new table diff tool lets you compare two tables side by side, across connections if needed. Useful for auditing data migrations, validating ETL outputs, and comparing production against staging. Differences are highlighted at the cell level. Export the diff as CSV for documentation.
The connections panel now shows live health status for all configured connections: latency, active query count, and last successful ping. Connections that fail a health check surface an alert immediately rather than failing silently when you try to run a query. For teams managing many data sources, this saves real time.
Organization-level connection templates are now fully supported. Admins can define connection configurations through Zentinelle and push them to all team members. Users get the connection pre-configured with no credentials visible. Rotating credentials is a one-step operation at the admin level.
Query results can now be exported directly to Parquet in addition to CSV and JSON. For data engineering workflows that feed into downstream pipelines, this eliminates a conversion step. Export also now supports streaming for large result sets rather than buffering the full result in memory.
Runs in the Calliope Workbench browser environment. Free for personal use. Enterprise credential management and audit trails available with Zentinelle.
v1.3.0 will introduce a visual query builder for users who prefer drag-and-drop table joins, a data pipeline view for scheduling and monitoring ETL jobs, and expanded Parquet and Arrow format support for data engineering workflows.

The Last Mile Is the Operator The first four parts of this series built the substrate: foundation, fleet, multi-fleet …

A Short Story About Why the Stack Has the Shape It Does Every platform has an origin story. Most of them are forgotten …