Dread Visual Skills

Reusable Codex skills for technical web work that should be clear, interactive, evidence-backed, and genuinely well made.

Install

git clone https://git.dread.technology/forgejo-agent/dread-visual-skills.git
mkdir -p ~/.codex/skills
cp -R dread-visual-skills/skills/* ~/.codex/skills/

Skills

build-interactive-system-maps

Architecture maps, dependency graphs, geographic network globes, runtime paths, trust boundaries, and agent-system diagrams. Includes a system-map brief, graph review gate, geographic measurement rules, progressive world-to-site-to-path interaction, and a sanitized revision example.

Examples: project lanes, Dread agent bus, research publication pipeline, and tmux control plane.

build-operational-dashboards

Dense monitoring, research, and repeated-action surfaces organized around an operating decision instead of equal-weight metric cards. Includes a dashboard brief and state/review gate.

Examples: research control surface and product dashboard.

art-direct-technical-experiences

Interactive scientific, audio, hardware, model, and research stories where the real mechanism or evidence is the visual subject. Includes a scene/state contract and art-direction checklist.

Examples: spatial beamforming lab and narrated research result.

verify-visual-artifacts

Browser-driven screenshot, interaction, responsive, console/network, accessibility, and Canvas/WebGL verification. Includes a receipt template and severity model.

Paired example: desktop DGit surface and mobile DGit surface.

Desktop	Mobile

Use

Mention the skill by name in a request, or ask Codex to use the matching installed skill. Each skill defines the design contract before implementation and requires an evidence receipt before completion.

The screenshots demonstrate patterns, not templates to copy. Read RATIONALE.md for the principles and tradeoffs behind the pack.