Features
35 features in one native app for exploring, querying, and running every database you have, with an AI chat and an MCP server so your agents can work alongside you.
1.0Data
The grid never gets in the way. A million-row table opens as fast as a ten-row one, and every edit waits for your review before it lands.
Rows paint immediately while the total count fills in the background, so a million-row table opens as fast as a ten-row one.
Change cells, add rows, and delete rows. Everything is staged and marked in the grid until you apply it.
Pick a column, match case, whole word, or a regex, and see every cell that will change before you replace it.
Filter by a value or exclude it, set NULL, expand, duplicate, or copy a row as JSON, CSV, Markdown, plain text, or an INSERT statement.
Every table tab can show its rows as a grid, JSON, a record card, plain text, a chart, an ER diagram, or a map. Switch per tab, or set a default.
Values that don't fit a cell get a viewer instead of being cut off.
Watch a table in real time. Inserts, updates, and deletes stream into the grid as they happen on the database.
Find in database searches across tables and objects as you type, so you never hunt through the sidebar.
2.0Query
A schema-aware editor that saves everything you run, notebooks for the analysis you hand to someone else, and codegen that turns the schema back into code.
Schema-aware autocomplete, one keystroke to run, and everything you run saved automatically.
Mix SQL cells and Markdown in one .sqlnb file. Run cells on their own and keep the results next to the notes. Good for an analysis you hand to someone else.
Write ORM-style queries, preview the SQL they generate, and run them against the connected database.
Read the live schema back out as Prisma or Drizzle source. It introspects once, so switching between the two is instant, even on a large schema.
3.0Schema
Browse every object, walk foreign keys row by row, draw the schema, and see exactly what changed between any two snapshots.
Entity-relationship diagrams drawn from the live schema, and five more ways to read the same structure.
See it liveTables, views, materialized views, and foreign tables, with live row counts in the sidebar.
Every table, view, function, trigger, enum, sequence, and index in one browsable list, with owners, estimated rows, and data and index sizes.
Walk foreign keys as an expandable tree, drilling from one row into everything it references. Row counts stream in without blocking you.
Snapshot the schema and see exactly which columns, indexes, and constraints were added or removed between any two snapshots. Then generate the migration SQL that gets from one to the other.
Compare two query results or table snapshots row by row. Added, removed, and modified rows are highlighted, including changes inside JSON and array columns.
4.0Visualize
Turn any query result into a chart, draw every PostGIS layer on a map, and pin the views you keep to a dashboard.
Turn any query result into a bar, line, area, step, or pie chart, or a geographic choropleth map.
Every PostGIS layer drawn on a map, clustered when there are too many points to draw one by one. The basemap ships with the app, so the default view makes no network requests.
Pin charts and saved views to a dashboard you can come back to any time.
5.0Operate
See who can read or write what, what the server is doing right now, and every statement that ran. Back up, restore, and browse production read-only.
See who can read or write what. Roles, grants, and privileges in one place.
A read-only check of the database that ranks what's wrong, worst first. It never reads a row of your data and never runs DDL.
What the server is doing right now: version, uptime, connections, replication, cache hit rates, sessions, and locks.
Export a full SQL dump, schema and data, with toggles for what to include and per-table selection. Restore from a .sql file with results per statement.
Every statement the app has run, with its duration and outcome, including the ones it ran on your behalf. Nothing the UI does is invisible.
Lock any connection so writes are blocked entirely. Browse production without the risk.
Connect through a bastion host, or let Stroke find what's already running: Docker containers, local Postgres and MySQL, and the SQLite files your ORM config points at.
6.0AI and agents
Ask in plain language and the AI chat runs the queries. The built-in MCP server opens the same door to the agents you already use.
{ "mcpServers": { "stroke": { "url": "http://127.0.0.1:4319/mcp" } }}An assistant with real database access: it runs queries, reads the schema, explains what it found, and draws charts inline. Destructive statements always ask first.
Start it in Settings, copy the config, and Claude, Cursor, or any MCP client can query your database. A stable token means you configure it once.
7.0Keyboard and workflow
A command palette, a shortcut for every core view, split panes, and an optional Vim mode. Every feature is reachable from the keyboard.
Press Cmd/Ctrl+K and type a table, a feature, or a shortcut to jump there.
A shortcut for every core view, drawn as keycaps wherever it appears.
Drag a tab to either edge to split the window. Compare two tables, or keep a query beside the rows it returns.
Add new panels, query tools, and integrations with first-party and community extensions.
Dark and light themes, and six grid styles: lines, bordered, striped, dotted, dots, and minimal.
Free to download. $9.99 once, whenever you decide to keep it.