Images Showcase - Trapped inside a flooded car

July 2026

Tools GPT 5.5 
Technique: Agent Skill & Context Engineering
--------
Context/
-Agent.md
-Style.md
-Underwater.md
-Referencence Character (.png)
-Benchmark (.png)

Skill/
-Character-to-image.md
-QC-anatomy.md

Prompt;
[Scene][SITUATION] [CAMERA] [POSTURE] [EXPRESSION]
---------------------------------------------------------------------------------

As you can see, The arichitecture is 3 Layer cascaded as..
-Context : Contain Constrain, Guard Rail & Benchmark
-Skill : Contain Knowhow and SOP (Work Flow)
-Prompt : Trigger and give explict order

-----

Showcases