TypeDB Studio

The web app at studio.typedb.com is served over HTTPS, so browsers may block connections to plain http:// servers. At the time of writing, Chrome, Chromium and Firefox allow http://localhost and http://127.0.0.1; other browsers, including Safari, may not. If your server has no TLS, use the desktop app.

The desktop app runs on:

The web version updates automatically.

The desktop app does not yet auto-update. Download the latest version from the downloads page and run it. You may optionally uninstall the previous version.

Select TypeDB edition below, and follow the instructions:

The Connect TypeDB server page (/connect) has two modes, selected at the top.

Use address and credentials:

Studio suggests up to 10 recently used addresses as you type. Remove a suggestion with its trash icon, or by highlighting it and pressing Delete.

Studio warns you under the Address field if you use port 1729 (the gRPC port), omit the port, or use http:// from a browser that may block it.

Use connection string: paste a connection string (see below). Both modes stay in sync, so you can paste a string and then edit individual fields.

Click Connect. Studio then opens the tool you last used.

The connection string format is:

Each address must include http:// or https://. The username and password must be URL-encoded. The database and name are optional.

For example, to connect to the mydb database on a local server:

The gear icon next to the field opens the Connection string builder, which assembles a string from separate fields and previews it.

TypeDB Cloud credentials files include a prebuilt connection string.

You can pre-fill the connection form with URL parameters:

Available parameters:

The password is never taken from the URL. If address and username match the server you are already connected to, Studio skips the form.

Studio remembers your last successful connection and reconnects to it on launch, reselecting the database you last used. If the server has exactly one database, Studio selects it automatically.

Studio does not reconnect if it is launched from a link whose address names a different server; it shows the pre-filled connection form instead.

To stop Studio reconnecting, use Sign out (see below).

The connection button on the left of the top bar shows <username>@<server name> and a status dot: green (connected), amber (connecting) or red (disconnected). Its menu shows:

Click a database in the menu to switch to it. Studio remembers your choice. Use Refresh to reload the list.

Click New database. The name may contain letters, digits, hyphens, and underscores only. You can optionally choose a sample dataset to load into it. The new database is selected immediately.

Click Load sample dataset, choose a dataset, and confirm the database name (defaults to the dataset’s name). Studio then:

If the schema fails to load, the data load is skipped. Check the query tabs' Log output for details.

Sample datasets currently include Bookstore, Cyber Threat Intelligence and Robotics. They are maintained in the typedb-examples repository, and Studio automatically picks the dataset version that matches your server.

Click the trash icon next to a database in the menu. Deleting a database is irreversible and permanently removes all of its data. You must type the database name to confirm.

The default. Each query runs in its own transaction. Studio detects the transaction type from the query (define etc. runs as schema; insert, put, update and delete run as write; anything else as read), and retries with another type if the server rejects it. Write and schema queries are committed automatically.

Switch to Manual mode from the transaction menu to run several queries in one transaction. Then:

Commit and Rollback appear only for write and schema transactions.

Changing the type or mode closes any open transaction immediately. Only one transaction can be open at a time within a single instance of TypeDB Studio.

Manual mode also gives the Data explorer and Graph explorer a fixed snapshot of your data. Your chosen mode is remembered between sessions.

Options in the transaction menu sets the transaction timeout (Timeout (seconds), default 300). It applies to all transactions, including auto mode.

Inspector in the transaction menu shows the current (or most recently closed) transaction: its status, type, and the times it was opened and closed, along with low-level metadata such as the transaction ID and timeout.

Without writing queries, using the Graph explorer:

From a query, using the Query tool:

See What gets drawn for how variables become nodes and edges.

Alternatively, put several queries in one query tab separated by end;. In auto mode they run in a single transaction that is committed only if all succeed. See Run multiple queries.

The query tool supports multiple query tabs, so you can keep several queries open at once.

Open tabs, their text, and each tab’s chosen output view are remembered between sessions. Results are not.

The query editor offers TypeQL highlighting, autocompletion, bracket matching, code folding, and multiple cursors.

Autocompletion appears as you type, or on Alt+Space / Ctrl+Space. It is schema-aware: it suggests type labels, the attributes a variable’s type owns after has, a relation’s roles inside links (…​), variables already used in the query, TypeQL keywords, and functions (with their signatures) defined in the schema or the editor.

Useful shortcuts (use ⌘ in place of Ctrl on macOS):

Click the run (▶) button at the top right of the editor, or press Ctrl+Enter. If the button is disabled, its tooltip explains why (e.g. no database selected, or no open transaction in manual mode).

The Limit dropdown next to the button caps the number of answers returned: from 10 to 100,000, default 100. Your choice is remembered. When a result reaches the limit, the Log says so.

A running query cannot be cancelled; the run button is disabled until it finishes.

Separate queries in one tab with end; to run them in sequence, in a single transaction:

Batches are logged as one history entry. The Log shows each query’s output under a ## Query N header. The Table view shows one table per query, for batches of up to 10 queries. Graph output and export are not available for batches.

Each run appears as a Run N tab above the results. A new run replaces the selected run tab, unless that tab is pinned, in which case it opens a new one. Right-click a run tab to close, pin, or rename it. Run tabs are not saved between sessions.

The Results pane offers four different views of results - you can choose between Log, Table, Graph and Raw.

Log contains textual output, similar to TypeDB Console, as well as the input log and error messages. It follows new output until you scroll up. Very large results are truncated in the log; view them in the Table instead.

Table shows each answer as a row:

Graph draws match (without fetch) and insert results as an interactive network. You can expand nodes, restyle, and export the graph. See Graph visualisation for all its capabilities.

Raw shows the raw HTTP response from TypeDB as JSON, and is useful for debugging.

Query results can be exported directly from the Results pane:

In CSV, entities and relations are written as type:iid, and attributes as type:value. Downloaded files are named after the run, database, and time.

The query tool’s left pane is the schema pane, a reference of what you can query. Right-click a type and choose Load instances to open a new query tab that fetches its instances and runs it.

The right pane lists every query and transaction operation, newest first, with its time, duration, and outcome (✓ or ×).

This pane mirrors the dedicated Query history tool.

Query-level errors are printed to the Log. These are typically errors returned directly from TypeDB, and will contain detailed error codes and explanations of the problem (e.g: illegal TypeQL semantics).

Transaction-level errors trigger an error notification. Click View error details on the notification to see and copy the full error.

Both types of errors can also be viewed and copied from the History pane.

Each code block is an editable TypeQL editor with Run and Copy buttons. After running a block, Save adds it to Saved queries.

Each run opens a Run N output tab under the block, with the same Log, Table, Graph and Raw views as the query tool, including graph exploration. Send to AI in the Log sends the output back to the chat. The Limit query results setting at the top of the page caps answers per run.

Agent mode keeps a separate, automatically titled conversation history for each database.

Conversations are stored locally (see Data storage). Only messages are saved, not query results.

To analyse output from the query tool, use Send to AI in its Log view. This starts a new conversation with the log.

Agent mode is powered by OpenAI’s models. Using it sends your database schema and chat messages (not your data or query results, unless you send them) to OpenAI, where they may be used for model training. Rate limits apply.

AI features are off by default and require explicit, one-time opt-in the first time you use them. To withdraw consent, click Disable AI features at the bottom of the Agent mode page. You can optionally delete your local chat history when doing so.

Tables show:

Multi-valued attributes are comma-separated. Long values are truncated, with the full value in a tooltip. Each attribute value has a copy button. The tab title shows the instance count.

Table controls:

Searching, sorting and pagination run server-side as TypeQL queries.

Selecting any row opens its Instance details.

Instance details show the same info as tables in much more detail, in collapsible sections:

Use the Explore buttons to jump to connected instances. Explorations leave behind a breadcrumb showing the chain you’ve taken. Click a breadcrumb to go back to that point.

Right-click a tab for Close, Close Others, Close to the Right, Close All, and Pin / Unpin. Middle-click closes a tab. Pinned tabs are kept by the bulk close actions.

Tabs are remembered per database, and restored (with fresh queries) when you reopen TypeDB Studio. Search, sort and page settings are not.

In manual transaction mode, the explorer reads from the open transaction, giving you a fixed snapshot of your data. If no transaction is open, it offers Open read transaction or Switch to auto mode. If the data was loaded in a different transaction, a banner offers to Refresh it.

The initial load is capped by the query tool’s Limit setting (default 100). There is no paging; grow the graph by exploring connections instead.

Right-click nodes, or use the side panel, to load relations, roleplayers and attributes on demand. See Explore connections for all the ways to expand the graph.

The Graph explorer also offers:

Each type gets one tab. Opening the same type again focuses its tab. Tabs close when you switch database, and are not restored between sessions.

The explorer works in both auto and manual transaction modes. In manual mode with no open transaction, it offers Open read transaction or Switch to auto mode.

Click Create user and enter a username and password. Only admins see this button.

Click Change password in that user’s row. Changing your own password reloads Studio and clears all session data: you are signed out, and in-memory state such as query history and run results is discarded. Reconnect with the new password.

Click Delete user in that user’s row. Deleting your own user also reloads Studio and clears all session data, and you will not be able to log in again.

The More actions (…) menu in the pane’s header offers:

Refresh reloads the schema. Your view settings are remembered.

For an empty database, the pane offers shortcuts to load a sample dataset, learn about schemas, or build with AI.

Graph output is available for match queries without fetch, and for insert queries. Schema queries, fetch queries, and multi-query batches cannot be drawn.

Each variable in the answers becomes a node:

The same instance appearing in several answers is drawn once. Only variables in the answers are drawn; variables removed with select are not. Type labels written in the query (e.g. $x isa person) are not drawn as nodes; use a type variable ($x isa $t) to include them.

Constraints between drawn variables become edges: has (owner to attribute), links (relation to roleplayer, labelled with the role name), isa, sub, owns, relates and plays.

Entity and relation nodes are labelled with their type and a display attribute (e.g. person: Alice), chosen automatically. You can change the display attribute per type in the side panel.

Labels hide automatically when zoomed out on large graphs.

Add more data to the graph without re-running a query, in either of these ways:

Right-click an entity or relation node for:

Each item has two scopes: here (this instance) and every '<type>' (every instance of its type in the graph). Loaded items are highlighted; clicking one again removes what it added.

The menu also sets the node type’s colour. For attribute nodes, colour is the only option.

Use the Explorer at the top of the side panel. Select a node to see its type, how many of that type are in the graph, and chips to load its links, attributes, and relations. In the Graph explorer, switch to here for the selected instance’s full details (see Graph explorer).

In the Elements tab, you can also right-click an entity or relation type chip to load links or attributes for every instance of that type.

The side panel has the Explorer at the top and three tabs below: Elements, Themes, and Customise. Use its menu to Dock to bottom or Dock to right.

Elements highlights parts of the graph by fading everything else. Nothing is removed. Toggle chips by kind (entity, relation, attribute, types, role type, value), by type, or by edge label (has, links, isa, sub, owns, relates, plays). Double-click a chip to show only that one.

Themes applies a built-in preset (The Classic, Cal Aesthetics, Fair and Square, To Boldly Go, Fifty Shades of Data, Size over Substance). Save current configuration stores your own preset, which you can later overwrite, edit, or delete.

Customise has two sub-tabs:

Reset all styles restores the defaults. Redraw graph re-runs the layout.

Export as PNG offers Download PNG (current view) or Download PNG (entire graph). Exports keep the current highlighting, use a solid background, and are capped at 8192 pixels per side.

The left sidebar links to every tool: Home, Query, Agent mode, Schema, Data, Graph, Users, Query history, and Saved queries. It also opens App info and Appearance, and links to TypeDB Cloud. Use the chevron to expand or collapse it.

The top bar holds the connection button, the database button, the transaction control, and:

You can switch between light mode, dark mode, and syncing with your operating system’s preference using the Appearance control in the sidebar. Your choice is remembered between sessions.

The App info dialog, opened from the sidebar, shows the current version of TypeDB Studio along with helpful links to get started.

Success messages appear briefly at the bottom right. Error notifications stay until dismissed; click View error details to see and copy the full error.