skilld
All skills
thedivergentai avatar

/godot-agent-vision

@90634df
by Divergent AIthedivergentai/gd-agentic-skills783 stars
48

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.

Ask your Agent

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

This session only. Nothing lands on disk.

referencesvision-review-rubric.md

≈750 tokens on demand. Your agent reads this file only when SKILL.md points to it.

Vision review rubric (rollup + gates)

Primary scoring instrument: taste-receptors.md — Taste Receptor Atlas v2 (237 micro-receptors × 0–2 across 13 families).

This file is the rollup, hard gates, and routing sheet. Do not score only the coarse V1–V8 bands and call it done — that is a FAIL of the skill itself.

Score visible pixels only. Default capture short-edge 512; --detail when TYPE/micro receptors fail for illegibility; 2–8 frame sequence when MOT/hit-peak ambiguous.

Core formula

Good visual review = Σ taste receptors − AI pretty centroid

Grade % of applicable receptor max
A ≥ 90%
B 75–89%
C 58–74%
F < 58%

Hard gates: see taste-receptors.md.

MANDATORY load order

  1. Capture WebP(s) → Read.
  2. Open taste-receptors.md — score every applicable row.
  3. Deepen as needed:
  4. Emit family subtotals + zero list + ordered fixes keyed to receptor IDs.
  5. Optional: roll up to V1–V8 /120 using the map in taste-receptors.md.

Coarse V bands (secondary only)

V Theme Max Source families
V1 Hierarchy 15 HIER + COMP
V2 Pressure contrast 15 CTRST + LIGHT (pressure)
V3 Typography 20 TYPE
V4 Affordances 10 AFF
V5 Spacing 10 SPACE
V6 Anti-slop 15 SLOP
V7 Originality 15 ORIG + MOT craft
V8 Identity 20 ID + COLOR + ICON

Legacy A-bar 108/120 ≈ receptor grade A (≥90%) when all families applicable.

Mode routing

Content Score families
HUD / menu / pause HIER CTRST TYPE AFF SPACE COLOR ICON COMP SLOP ID (+ LIGHT if FX)
Asset / icon sheet TYPE (labels) COLOR ICON ORIG ID SLOP
Gameplay viewport HIER CTRST SPACE COLOR COMP LIGHT MOT SLOP ID (+ TYPE/AFF if UI)
Marketing + game pair ID mandatory + SLOP + ORIG + TYPE

Incomplete review = invalid

A review is invalid if it:

  • Reports only PASS/FAIL or only V1–V8 without receptor zeroes
  • Skips Family TYPE when text is visible
  • Skips Family SLOP on any UI/marketing frame
  • Praises “looks good” without % + zero list

Source: SKILL.md on GitHub

Third-party checks

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

Provenance

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