A native desktop tool for one job: find out why your query is slow.
SQL Visual Explorer runs your query, parses its execution plan, and highlights the bottleneck — sequential scans, blown row estimates, hash joins spilling to disk, sorts that don't fit in memory.
It is not a general-purpose database client and does not try to replace DBeaver or DataGrip.
It does one thing: turn an EXPLAIN ANALYZE output into an answer.
- Five databases, one interface — PostgreSQL, MySQL, MariaDB, SQLite, SQL Server
- Query execution — resizable results grid, column sorting, CSV/JSON export
- Plan visualization — tree, node graph, and flame-graph views, colour-coded by cost
- Automatic issue detection — ten rules covering scans, joins, sorts, estimates, and index gaps
- Plan comparison — run two variants side by side and get a cost/time/node-count diff with a winner
- HTML reports — export a plan or a comparison as a self-contained HTML file
- Snippets — save, tag, and search reusable SQL, recallable from a
Ctrl+/popup - Query history — every run is recorded locally; reopen the SQL or reload the saved plan
- AI Advisor (optional) — send the query and detected issues to any OpenAI-compatible endpoint for rewrite suggestions
- Read-only by design —
Analyzerefuses mutating statements and multi-statement batches - Credentials in the OS keychain — never in plaintext config
| Rule | Fires when |
|---|---|
| Sequential scan | A large row set is scanned without an index |
| Missing index hint | A sequential scan carries a filter that an index could serve |
| Large nested loop | A nested-loop join is driven by too many outer rows |
| High cost node | A single node dominates total plan cost |
| Row estimate mismatch | Planner estimate and actual rows diverge sharply |
| Large sort | A sort processes a large row set |
| Hash join spill | The hash table exceeds work_mem and spills to disk in multiple batches |
| Bitmap heap scan | Informational — bitmap access path in use |
| Function in filter | A function call wraps a column and blocks index use |
| Implicit cast | A type cast in a filter blocks index use |
Grab a self-contained binary from the Releases page —
no .NET runtime required. Builds are published for win-x64, osx-x64, osx-arm64, and linux-x64.
On macOS, first launch requires Right click → Open (the binary is unsigned).
Requires the .NET 10 SDK.
git clone https://github.com/Komortes/SQL-Visual-Explorer.git
cd SQL-Visual-Explorer
dotnet restore SQLVisualExplorer.sln
dotnet run --project src/SQLVisualExplorer.Desktop/SQLVisualExplorer.Desktop.csproj- Connections → add a connection; the password goes to the OS keychain, everything else to a local SQLite file.
- Editor → write a query and press
Ctrl+Enterto run it. - Press
Ctrl+Shift+Enterto runEXPLAIN ANALYZEand jump to the plan. - Click a node to inspect its cost, rows, timing, and any issues attached to it.
- Compare → paste an optimized variant and confirm it's actually faster.
| Shortcut | Action |
|---|---|
Ctrl+Enter |
Run query |
Ctrl+Shift+Enter |
Explain analyze |
Ctrl+Shift+F |
Format SQL |
Ctrl+/ |
Open snippet picker |
Ctrl+C |
Copy selected result rows |
| Database | Driver | Plan source |
|---|---|---|
| PostgreSQL | Npgsql | EXPLAIN (ANALYZE, BUFFERS, FORMAT JSON) |
| MySQL / MariaDB | MySqlConnector | EXPLAIN FORMAT=JSON, EXPLAIN ANALYZE |
| SQLite | Microsoft.Data.Sqlite | EXPLAIN QUERY PLAN |
| SQL Server | Microsoft.Data.SqlClient | SHOWPLAN_XML, STATISTICS XML |
| What | Location |
|---|---|
| Connections, history, snippets | %LOCALAPPDATA% / ~/.local/share → SQLVisualExplorer/sqlvisualexplorer.db |
| Passwords & API keys (macOS) | Keychain, service sql-visual-explorer |
| Passwords & API keys (Windows) | DPAPI-encrypted files under %APPDATA%\sql-visual-explorer\secrets |
| Passwords & API keys (Linux) | Secret Service via secret-tool (GNOME Keyring, KWallet, …) |
Nothing leaves the machine unless you explicitly enable the AI Advisor.
Off by default. Open the advisor settings from the gear button next to AI Advisor on the plan toolbar and point it at any OpenAI-compatible endpoint:
| Setting | Default |
|---|---|
| Endpoint | https://api.openai.com/v1 |
| Model | gpt-4o-mini |
| API key | required — stored in the OS keychain |
When enabled, the advisor receives the SQL text, the database type, and the list of detected issues, and returns a summary, concrete suggestions, and an optional rewritten query.
src/
SQLVisualExplorer.Desktop # Avalonia entry point
SQLVisualExplorer.UI # views, controls, view models
SQLVisualExplorer.Application # service interfaces and use cases
SQLVisualExplorer.Infrastructure # drivers, plan parsers, persistence, secrets
SQLVisualExplorer.Domain # models and enums
tests/
SQLVisualExplorer.Infrastructure.Tests
Dependencies flow inward: Desktop → UI → Application ← Infrastructure, with Domain at the centre.
Adding a database means implementing IDatabaseDriver and IExplainParser — nothing above
Infrastructure needs to change.
Built with Avalonia 11, AvaloniaEdit, CommunityToolkit.Mvvm, and EF Core 10.
# build
dotnet build SQLVisualExplorer.sln --no-restore /m:1
# test
DOTNET_ROLL_FORWARD=Major dotnet test SQLVisualExplorer.sln --no-restore /m:1/m:1 keeps solution builds deterministic in the current local environment.
Warnings are errors (TreatWarningsAsErrors), and nullable reference types are enabled solution-wide.
EF Core migrations
Local EF tooling is pinned through dotnet-tools.json.
dotnet tool restore
DOTNET_ROLL_FORWARD=Major dotnet tool run dotnet-ef migrations add <MigrationName> \
--project src/SQLVisualExplorer.Infrastructure/SQLVisualExplorer.Infrastructure.csproj \
--startup-project src/SQLVisualExplorer.Infrastructure/SQLVisualExplorer.Infrastructure.csproj \
--output-dir Database/Migrations
DOTNET_ROLL_FORWARD=Major dotnet tool run dotnet-ef database update \
--project src/SQLVisualExplorer.Infrastructure/SQLVisualExplorer.Infrastructure.csproj \
--startup-project src/SQLVisualExplorer.Infrastructure/SQLVisualExplorer.Infrastructure.csprojSee CONTRIBUTING.md. Bug reports and feature ideas are welcome via issues.
MIT © Komortes