Land your first recording in one session
You do not need to configure everything on day one. Start narrow, learn the surfaces, then deepen the Scenario over time.
Before you begin¶
Installation
Import TraceMind from the Unity Asset Store into your project. Unity resolves package dependencies on import.
A playable project
Something you can enter Play Mode with and reproduce a behavior worth inspecting.
About five minutes
Enough time to reproduce the issue once, cleanly, from start to symptom.
Where your data lives
TraceMind stores data per project under Library/TraceMind/. Your Assets layout can move without breaking the database path.
Open TraceMind¶
Open Tools, then DivinityCodes, then TraceMind. The left rail shows Home, Scenarios, Reports, and MindBoard. Bottom actions open About, Help, Preferences, and the DivinityCodes Roadmap.
Treat TraceMind like a first-class workspace. Dock it beside the Game view so recording and reproduction stay in one glance.
Your first investigation¶
Seven steps take you from an empty workspace to an explained bug. Watch the pass once, then open a step when you want the written detail.
Create a Scenario
From Home or Scenarios, choose New Scenario. Name it after the system, such as Player Health Sync or Loot Drop Timing. A Scenario is a reusable configuration, not a one-off capture, so the name should still make sense the next time this bug appears.
Choose a Profile
Pick a recording depth in the Scenario workspace. Start with Standard unless you already know the issue is tiny or enormous. The Profile controls how much detail each captured call carries, which is the main lever on Report size and Editor overhead.
Track what matters
Open the Tracking Editor and search for the types or members involved. Track the one type that owns the broken state first. Resist tracking everything until the evidence tells you something is missing, because a narrow first pass is faster to read and faster to record.
Add a Watch
In the member inspector, watch the property that mysteriously changes. Tracking captures call activity, while Watches follow values over time and unlock Causal Flow. Pin one or two high signal members first rather than every field on the type.
Enable relevant Probes
If the bug touches scene loads, object lifecycle, coroutines, animator transitions, or resource loading, enable the matching Probes. Each enabled category becomes a lane in Probe Map, which is where timing bugs between systems become visible.
Record the reproduction
Press Start Recording, enter Play Mode, reproduce the issue once cleanly, then press Stop Recording. A short intentional take beats a long fishing expedition every time, because everything you capture is something you later have to read.
Open the Report
Switch to Reports and open the new recording. Tour the Tree for structure, the Timeline Overview for ordering, Watch Studio for value history, then Causal Flow from a suspicious write. Use Add to MindBoard to keep anything you will need again.
Pick the right Profile¶
| Profile | Best for |
|---|---|
| Minimal | Enter and exit points plus durations, when you want a light footprint |
| Standard | Methods, exceptions, and probes for everyday bugs |
| Deep | Full capture plus snapshots when you need nested detail |
Selectivity is speed
Over-tracking makes recordings heavier and graphs noisier. Prefer a tight first pass, then widen only where evidence is missing.
Validate before you record¶
Use Scenario Validation when available. It reports whether recording is ready, ready with warnings, or unable to start, so you fix blockers before investing time in a long reproduction.
A minimal success checklist¶
- The Scenario name matches the system under investigation
- Tracking includes the owning type of the broken state
- At least one Watch points at the suspicious member
- The recording is short enough to explore comfortably
- You can point to a Report moment and explain what changed
Where to go next¶
If something fails to record or open, see Troubleshooting.