All skills
thedivergentai avatar

/godot-agent-vision

@90634df

Agentic eyes for Godot work: cross-platform screenshot CLI (asset/region/window/screen) to budgeted WebP, TEMP editor viewport bridge under .gdskills/, and Taste Receptor Atlas v2 (~237 micro-receptors × 0–2: hierarchy, contrast, typography×41, affordance, space, color, icon, composition, light/FX, motion-tells, anti-slop, originality, identity). Use when agents must see UI, assets, or the editor — not for in-game screenshot Autoloads. Keywords: agent vision, screenshot, WebP, visual QA, taste receptors, typography, UI review, asset sheet, editor viewport, anti-slop.

Use this Skill: https://skilld.dev/gh/thedivergentai/gd-agentic-skills/godot-agent-vision

This session only. Nothing lands on disk.

referencescomposition-fx-receptors.md

≈1.2k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Composition / lighting / motion receptors (deep criteria)

Supports COMP-, LIGHT-, MOT-* in taste-receptors.md. Still WebP is primary; use a 2–8 frame sequence when MOT / hit-peak / shake decay is ambiguous.


Still vs sequence

Signal Still enough? Sequence when
Thirds, lines, depth, silhouette Yes Camera pan changes focal
Motivated light / rim lottery Usually Flicker / TOD blend
Bloom / DOF mush Yes Threshold ramps
Hit spark peak Partial Peak <90ms
Screen shake Partial (offset) Confirm decay + UI re-anchor
Squash / UI overshoot Mid-pose tell Confirm settle

Composition (COMP-*)

Power points vs HUD

Rule of thirds is a guide. COMP-POWER-CONFLICT is a hard readability fail: crosshair + telegraph + CTA competing at one third → 0.

Receptor PASS FAIL
COMP-THIRD-ACTION Subject near intersection / framed Dead center with no reason
COMP-THIRD-HUD Persistent HUD on edges HP sits on power point
COMP-CENTER-CLEAR ~60% center free Center chrome + combat
COMP-MASS-EDGE Heavy chrome at edges Edge + center both heavy
COMP-SAFE-EDGE Critical inside safe margins Clipped / bezel-adjacent

Leading lines & value

  • COMP-LEAD-ENV / LIGHT / UI — Tracks, shafts, rails aim at intent; competing rails → 0.
  • COMP-VALUE-GROUP / COMP-FOCAL-* — Subject pops in mental B/W; hue-only focal → 0.
  • COMP-DEPTH-LAYERS — FG/MG/BG separation; flat single plane → 0.
  • COMP-SIL-ACTION / THREAT — Limbs and telegraphs readable as fill.
  • COMP-NEG-SPACE / BALANCE / HORIZ / FRAME / SCAN / DIEG-META — Breath, intentional mass, level or owned dutch, env framing, predictable scan, meta ≠ HUD replacement.

Lighting / FX (LIGHT-*)

Juice layers must obey hierarchy — not replace it.

Motivated light

Receptor PASS FAIL
LIGHT-MOTIV Shared key vector Sourceless wash
LIGHT-KEY-FILL Form readable Crushed mud
LIGHT-RIM-OWN Rim supports sep Identical rim on all assets
LIGHT-PLAY-SEP Interactive layer pops Player melts into BG

Bloom / DOF / post

Receptor PASS FAIL
LIGHT-BLOOM-HIER Bloom on story beats only Bloom = hierarchy
LIGHT-BLOOM-ABUSE Edges preserved Detail mush
LIGHT-DOF-SUBJ / UI Subject sharp; HUD sharp Gameplay or UI blurred
LIGHT-POST-STACK ≤2 dominant posts AO+bloom+DOF+CA+grain scream

Particles / hit / vignette

Receptor PASS FAIL
LIGHT-PART-READ Sparks explain direction/material Semantic-free noise
LIGHT-PART-OCCLUDE FX clear of HP/ammo/crosshair Particles cover HUD
LIGHT-PART-OVER Impact ≤~12–18% viewport Screen-filling soft quads
LIGHT-FX-LANE Telegraph ≠ flourish hue Same lane
LIGHT-HIT-SPARK / FLASH / SCALE Consistent color; brief flash; L/M/H tiers Rainbow; full white; uniform max
LIGHT-VIGN-DMG / META Edge vignette; shape backup Eats center/UI; color-only low HP
LIGHT-EMISS-HIER / SPEC-LOT Emissive for danger/interact; consistent spec Every prop emissive; shiny lottery

Motion tells in stills (MOT-*)

Freeze-frames leak motion grammar.

Receptor PASS FAIL
MOT-BLUR-READ / SMEAR Direction clear; vector smear Omnidirectional mush
MOT-TRAIL-SOUP / GHOST Short trails; sparse ghosts Trail soup; clone soup
MOT-SQUASH / STRETCH / VOL Volume law holds Inflating blob
MOT-ANTIC Wind-up readable No telegraph
MOT-SHAKE-OFF / TIER / ROT UI re-anchors; tiered; minimal roll HUD floats; coin=boss; sickness roll
MOT-HITSTOP Micro-freeze Slideshow freeze
MOT-TWEEN-OS / UI-BOUNCE Settles on grid; ≤~1.05/0.95 Buttons off baseline; oversized bounce

Co-occurrence → SLOP

Slop cluster COMP/LIGHT/MOT co-tells
SLOP-NEON LIGHT-BLOOM-ABUSE + LIGHT-BLOOM-HIER=0
SLOP-RIM LIGHT-RIM-OWN=0
SLOP-SMOOTH LIGHT-SPEC-LOT=0 + ORIG-STRUCT soft
SLOP-EQUAL COMP-FOCAL-SINGLE=0 + COMP-MASS-EDGE=0

Sources

  • GDC / Game Developer — leading lines (Ethan Carter), level design tips, animation principles
  • GameJuice — screen shake, shader juice / hit flash
  • LearnOpenGL — physically based bloom
  • GamineAI — top-down combat VFX readability (lanes, occupancy)
  • Magnopus / composition craft — value grouping, focal clarity

Source: SKILL.md on GitHub

1 warning1mo3 checks · Risk MEDIUM
  • Gen Agent Trust Hub1mo

    The skill captures screenshots and interfaces with the Godot editor by executing system commands and modifying project configuration files. It includes a temporary editor bridge staged as a Godot plugin, which introduces risks related to command execution and dynamic module loading.

  • Socket1mo

    No alerts

  • Snyk1mo

    Risk: LOW · No issues

Signed by skilld at 90634df. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 3 weeks ago.

Activeupdated 2 months ago

README badge

README badge for thedivergentai/gd-agentic-skills/godot-agent-vision