Find agent skills by outcome

location: The Blue Note Lounge

a stylized jazz club with art deco architecture

description: A charming

animated noir scene where a gruff detective questions a glamorous jazz singer in a stylized 1950s club.

The Midnight Melody Mystery

{

Null Values: If a field is not applicable

set it to null rather than omitting it

Micro-Details: You must note scratches

dust

accent_colors: [Color name 1

Color name 2]

// REPEAT for EVERY single object

no matter how small.

Granularity: Never say a crowd of people. Instead

list the crowd as a group object

scene_description: A comprehensive

objective paragraph describing the entire scene.

dominant_hex_estimates: [#RRGGBB

#RRGGBB]

You must return ONLY a single valid JSON object. Do not include markdown fencing (like ```json) or conversational filler before/after. Use the following schema structure

expanding arrays as needed to cover every detail:

Macro Sweep: Identify the scene type

global lighting

Micro Sweep: Scan for textures

imperfections

Relationship Sweep: Map the spatial and semantic connections between objects (e.g.

holding

Data

CORE DIRECTIVEDo not summarize. Do not offer high-level overviews unless nested within the global context. You must capture 100% of the visual data available in the image. If a detail exists in pixels

it must exist in your JSON output. You are not describing art; you are creating a database record of reality.

Before generating the final JSON

perform a silent Visual Sweep (do not output this):

accent_colors: [Color name 1

Color name 2]

dominant_hex_estimates: [#RRGGBB

#RRGGBB]

// REPEAT for EVERY single object

no matter how small.

Relationship Sweep: Map the spatial and semantic connections between objects (e.g.

holding

You must return ONLY a single valid JSON object. Do not include markdown fencing (like ```json) or conversational filler before/after. Use the following schema structure

expanding arrays as needed to cover every detail:

scene_description: A comprehensive

objective paragraph describing the entire scene.

Before generating the final JSON

perform a silent Visual Sweep (do not output this):

Macro Sweep: Identify the scene type

global lighting