Scan the scene before aiming
Agent J is a scene-clearing game: each room holds active targets and a marked safe position, and the round ends when every target is handled and the route to the next area opens. The input is scene-specific — a mouse aim, a drawing gesture, or a touch action — so the prompt matters as much as the aim.
Every target handled opens the next route
Clear the active targets in each scene and reach the position the level marks as safe. A scene finishes when every required action is done and the route to the next area opens up, so the round is a sequence of cleared rooms rather than a score chase.
Use the aiming and action input the current scene shows, which may be mouse clicks, drawing gestures, or touch taps. The on-screen prompt tells you which control each scene expects, so read it once and act accordingly.
Before acting, scan the whole scene and pick the target that sits closest or most exposed. Deal with that one first, then wait for the scene to update before choosing the next action, because the layout changes after each clear.
Scan the scene before the first aim
The environment usually signals which action is safe, so spend the opening seconds reading the layout. Note where targets sit relative to obstacles and which one is closest to your start point. Acting on the most visible target first keeps the scene simple.
Finish one target before switching
Completing one action changes the scene state, and new options often appear only after it finishes. If you switch early, the first target stays active while you try to read a second one. Let each action resolve, check what updated, then decide the next move.
Use obstacles as reference points
Objects in the scene mark where you are safe and which target becomes reachable next. When a move feels blocked, the answer is usually on the other side of an obstacle — adjust your position instead of repeating the same input, and the scene usually opens up.
Plan the retreat before committing
Before you commit to a scene, notice which direction stays clear. If an action goes wrong, you want a way to reposition without pushing through the middle. Reading that route in advance turns a stalled attempt into a quick correction.
Let the prompt name the input
Each scene displays the control it expects, so the gesture that worked in one room may not register in the next. Read the prompt once at load, match the input, and commit to one action at a time instead of assuming the previous control still applies.
Treat a cleared scene as your stop point
Each finished scene banks progress, which makes it the natural place to pause. The route to the next area is saved, so leaving between rooms costs nothing; leaving mid-room leaves a target active and a layout half-read.
Which target should go first
Which target should I handle first in a scene?
The one closest to you or most exposed, because it can interrupt your next move. Let that action finish, check how the scene updated, then pick the next target from the new layout.
Is rushing through the scene actions a good idea?
No. Committing to the next action before the previous one resolves usually leaves a target active and the read wrong. One action at a time, with a check of the updated scene between each, clears rooms faster.
Why does my input not match the scene?
The control is scene-specific, and the prompt at load shows which one is expected. If a gesture does nothing, look at the prompt again and match that input before judging the attempt.
What does the safe position mean for stopping?
The marked safe position is where a scene closes, and that is the neatest point to leave. Progress at the boundary is banked, so finishing the scene first keeps the next session's start clean.
