Agent Skills for the Design Award Paper Trail

What it does

design-judge-skills ships six agent-installable modules that handle the unglamorous logistics of industrial-design awards. Each skill owns a distinct phase—searching verified winner databases, evaluating design maturity, matching projects to the right award and category, drafting entry text from user materials, and running a final compliance check against current official rules. A shared support package keeps the taxonomy and source registry consistent across the workflow.

The interesting bit

Most AI design tools generate images; this one generates paperwork. The project explicitly refuses to estimate win probabilities or mimic secret judge preferences, instead anchoring every claim to official sources and separating hard facts from model inference. That kind of epistemic humility is rare in agent tooling.

Key highlights

• Covers major awards including iF, Red Dot, IDEA, Core77, James Dyson, and GOOD DESIGN AWARD Japan.
• Built-in descriptive background draws on 22,125 observed works from iF, Red Dot, and IDEA—used for context, not scoring.
• Skills are composable: trigger design-award-pipeline for a full workflow, or invoke design-award-search or design-evaluation independently.
• Deadlines, fees, and eligibility rules are verified at runtime from official pages; nothing is hardcoded for a specific year.
• Most skills are marked Beta; only design-award-search is currently Stable.

Caveats

• The 22,125 observed records provide descriptive background only; the README is explicit that they do not change core scores or predict outcomes.
• Five of the six user-facing skills are still in Beta, so boundary cases remain likely.
• Requires an agent environment that supports SKILL.md frontmatter (Claude Code, Codex, OpenClaw, Hermes, etc.).

Verdict

Industrial designers and studios already drowning in award-season admin should look here. If you are hoping for an AI that guarantees a Red Dot win, this is explicitly not it.