Editor with intellisense.
SQL completion based on the live catalog. Format the query with one click. Multi-tab editor so three queries can stay warm at once.
Pillar 02 · Shared queries
Stop losing operational SQL in Slack and laptops.
A folder per topic. A query per diagnostic. Variables for the values that change between runs. Versioning for the edits. Read-only execution through the agent. Every run audited.
Personal and team.
Personal queries are yours alone. Company queries are visible to the team. Folders organize by domain, on-call area, or customer.
{{customer_id}} built in.
Typed variables (string, number, date, datetime, json, guid, boolean). Required or optional. Defaults at the query level, overrides per user, team, or company.
Every save is a checkpoint.
Restore any previous version. The history shows who edited what and when, with a diff next to each version.
Read-only through the agent.
The cloud renders the SQL with your variables, validates the statement, dispatches to the agent. The agent runs it through the configured connection with a statement timeout.
Every run is a row.
Rendered SQL, variable values, who, when, on which connection, row count, truncation flag, duration. Filter by saved query, by user, by status.
From query to schedule
A saved query is one click away from being a schedule.
The natural next step after saving a query is asking the team how often they want it. Schedules turn a saved query into a recurring CSV or Excel that lands in Slack or email.
See schedulesSafety, in detail
Three controls before the database sees the SQL.
Statement classifier
Single statement. Read-oriented statements only. DML and DDL rejected at the API.
Read-only role and timeout
Use a read-only database login. The agent enforces the statement timeout during provider-side execution.
Plan caps
Concurrency, daily run quota, max rows, max bytes, max statement timeout. Each cap is enforced before the agent is contacted.
FAQ
Shared SQL queries for teams
Can I run a query that writes? +
No. DML and DDL are rejected before dispatch. Use a read-only database role for the connection. If you need to write, use your existing migration tool and let Taavik observe the result.
How are concurrent runs handled? +
Each tab is independent. Concurrency caps are per user and per company. Pro allows up to ten concurrent runs per user and thirty per company. The agent enforces a hardware-level cap per connection key.
What if my query takes too long? +
Statement timeout: fifteen seconds on Free, thirty seconds on Pro. You can also click Cancel and the agent cooperatively cancels the statement. The result returns with the rows received so far.
Where do query results live? +
In your browser for the current session. The cloud holds the result file in encrypted storage governed by your retention. Row contents are not persisted in the database, only audit metadata.
Can I share a query with the team? +
Save it under the company scope, the team sees it. Personal scope keeps it private. Variables ensure portability across users with different defaults.
Reuse it
Pull team queries out of pgAdmin and into the workspace.
Save the first diagnostic, share it, watch it become the first place the on-call looks next time.