Find the right starting point.
Look up scene objects, inspect members, scan data, and narrow the target before making a change.
SELECT · SCAN · METHODSUNIVERSAL QUERY LANGUAGE / UNITY
Explore your game’s data. Save checks you can run again.
Use UQL directly or connect your development agent to Unity.
01 / FROM A QUESTION TO A WORKFLOW
See how a few commands can help you find a value, check a change, or follow its history.
BUCKET $hits; VALUE $report = null;
SCAN FIELD health IN SCENE INTO $hits REPORT $report;
RECORD $report.Text; REGISTER health_scan;Command excerpt. Assumes a field named health exists; no matches is a valid result. Scan limits can leave partial coverage.
WHAT THIS WORKFLOW DOES
Search the loaded scene for health fields and retain a report of what was found. Follow a result into the values that matter.
Search a scene, object, or supported static scope.
Review discovered paths and the report’s coverage.
Save the report for the next step in your investigation.
VALUE $before = null; VALUE $after = null; VALUE $check = null;
BLOCK Test {
OBSERVE FIELDS hp = $boss.health INTO $before PHASE Before;
CALL PerformAction;
OBSERVE FIELDS hp = $boss.health INTO $after PHASE After;
CHECK health_changed: $after["hp"] != $before["hp"] INTO $check;
}Script excerpt. The surrounding script supplies $boss and a PerformAction block. CHECK reports an outcome; it does not stop the process automatically.
WHAT THIS WORKFLOW DOES
Freeze selected values at two points in a workflow, then check what changed. Get a Pass, Fail, or Inconclusive outcome that can be reviewed.
Freeze the fields relevant to this question.
Run the action you defined in your script.
A completed run alone does not mean the check passed.
UQLSCRIPT PositionEvidence;
ALIAS $character = [SCENE:Character];
TRACEOPT manual = true;
VALUE $positionTrace = null;
BLOCK Begin {
TRACE $character.transform.position INTO $positionTrace;
UPDATE TRACE $positionTrace PHASE BeforeMovement;
}Assumes an object named Character. Run Begin in the UQL host. This example samples once; it does not move the object or schedule movement.
WHAT THIS WORKFLOW DOES
Trace a scene object’s position, then take an explicit sample at a named point. Add later samples to compare changes through the same captured path.
Use a path rooted in the intended scene object.
Manual TRACE records at your chosen execution points.
Changes between samples can still be missed.
02 / BUILT AROUND THE INVESTIGATION
Find the data behind a bug, try a change, and keep the checks that help you catch it next time.
Look up scene objects, inspect members, scan data, and narrow the target before making a change.
SELECT · SCAN · METHODSSave a question as a script with named blocks. Capture selected values before and after an action, then compare them.
BLOCK · OBSERVE · CHECKRecord selected value changes with TRACE. Take explicit samples at meaningful points, including supported hook workflows.
TRACE · UPDATE TRACE · HOOKConnect a compatible development agent through the local MCP integration. It can find your Unity host, look up commands, run scripts, and read the results.
LOCAL MCP · UNITY HOSTChoose which scripts can run, edit values, capture images, or use events. Review the permissions before running a workflow.
PER-SOURCE GRANTSCollect logs, named reports, and supported PNG captures alongside your checks. Review the outcome and the limits of its evidence.
LOG · REGISTER · CAPTURE03 / WORK WITH YOUR AGENT
Choose what to investigate, define the steps in UQL, and review the results with your agent.
Explore the guides and commandsConfirm the intended project and a ready session.
Use the host’s help and the grants needed by the workflow.
Inspect reports and check outcomes, not just completion.
04 / A FIRST WORKFLOW
Your first workflow in four steps. For command details and more examples, open the UQL documentation ↗.
Use UQL’s Editor tools for interactive work, or a connected local agent through MCP. Confirm the project and session before running a command.
Read command help, then locate the object and member you actually need. Start with a small search, then expand it if you need more detail.
Put repeatable work in a .uql script with named BLOCK entry points. Enable the required grants, compile the group, and run the intended block.
Review logs, reports, and each check’s Pass, Fail, or Inconclusive outcome. Close permissions that are no longer needed.
Observations freeze selected values, not a whole object graph. Traces can miss changes between samples. Reading getters may execute code, and grants are operational controls rather than a security sandbox. Writes are not automatically rolled back.
05 / TWO TOOLS. ONE CONNECTED WORKFLOW.
UDB gives you the visual browser. UQL gives you queries and repeatable workflows.
For developers who want scripted investigation, reusable checks, and an agent connection to Unity.
Move between UDB’s visual inspection and UQL’s explicit workflows as the investigation changes.
This page is a development preview. UQL’s standalone release and its inclusion in the full UDB package are planned; download and purchase links will be added when available.
UQL is designed to work independently and is planned as a standalone product. An AI agent is optional: UQL also provides Editor workflows for direct use. Agent integration requires a compatible MCP client and a connected local host.
No. These are documentation-based examples, shown to explain the workflow. This website has no connection to your Unity project. Replace example targets and supply the required script context in your own UQL host.
No. It means the process finished. Read the individual CHECK outcomes and their reasons; a workflow can complete with a failed or inconclusive check.
FROM A QUESTION TO EVIDENCE.