Enter with one purpose and leave before greed chooses for you

Before entering, choose one main purpose for the run. It can be learning a route, advancing a public quest, testing an agent or carrying out confirmed high-efficiency loot. Trying to complete every objective at once makes risk hard to judge. Bring only what supports that purpose and leave enough grid space to react. A light starting plan also makes failure easier to understand, because you can identify whether the problem was navigation, combat, inventory or timing.

During the run, treat information as a resource. Notice the selected map difficulty, remaining time, nearby threats and at least one plausible direction of retreat. Do not travel deeper only because the current room still contains unopened objects. If an objective item has been found, its practical value can exceed its vendor value per slot. If the bag is nearly full, compare confirmed value, footprint, stack quantity and purpose rather than sorting only by the largest price number.

Plan extraction before the final minute. A map relationship or exit marker helps navigation, but it does not guarantee that the same option is available under every condition. Keep enough time to change direction, avoid an unfavorable fight or abandon low-priority loot. When forced to choose, protecting a completed objective or a coherent loadout is often more useful for learning than gambling the entire run for one additional container.

After the run, review one failure or one successful decision. If time expired, decide whether the route was too long or the search phase lacked a stopping rule. If inventory became blocked, inspect item shapes and partial stacks. If combat ended the run, compare the agent's skills, usable gear and the monster variant rather than assuming one number explains the result. TamasDB guides offer a method, while the database supplies the versioned facts behind it. Recheck both when a test build changes.