skills.rest
speplinski
speplinskiCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

design-optimize

Re-audit proposed design surfaces against baseline heuristics and return updated severities.

Updated Apr 21, 2026
Search Tags:#design#optimization#heuristics#audit#decision-support#design-review#pareto-dominance
One-click install
npx skills add https://github.com/speplinski/hackathon-opus-47 --skill design-optimize-speplinski
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: design-optimize
Source: https://github.com/speplinski/hackathon-opus-47/tree/main/skills/design-optimize
Command: npx skills add https://github.com/speplinski/hackathon-opus-47 --skill design-optimize-speplinski

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Re-audit skill for L8 of the Auditable Design pipeline. Given a baseline surface description (before_snapshot), a proposed alternative surface (after_snapshot), and the list of heuristic violations the baseline scored (with their baseline severities), re-audits the proposed surface against the same heuristic list and returns per-heuristic severities on ADR-008's anchored 0-10 scale. The module wraps the result in an OptimizationIteration record and runs a Pareto dominance check (plus weighted-sum fallback) against the baseline to decide whether the proposed design is an accepted iteration. Use when the user asks to re-audit a proposed change, score a design iteration, or evaluate whether an alternative resolves the baseline violations.

Core Features & Use Cases

  • Re-audit a proposed design surface against a baseline heuristic list and return updated severities per heuristic.
  • Enforce ADR-008 anchored severities and apply Pareto dominance with weighted-sum fallback to decide acceptance.
  • Use when evaluating design iterations, comparing before/after snapshots, or validating design changes against a fixed heuristic set.

Quick Start

Provide the before_snapshot, after_snapshot, and the baseline heuristics list, and I will re-audit and return the per-heuristic severities.

Frequently Asked Questions about design-optimize

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I re-audit a proposed design change against a baseline heuristic list?▼

To re-audit a design change, provide the before_snapshot, after_snapshot, and baseline heuristic violations. The audit re-evaluates the proposed surface against those fixed heuristics and returns updated per-heuristic severities on an anchored 0-10 scale.

What is Pareto dominance in design optimization and how does it decide acceptance?▼

Pareto dominance checks whether a proposed design iteration improves heuristic severities without regressing others. If dominance is inconclusive, a weighted-sum fallback calculates overall scores to decide whether the design change is accepted.

How do I score a design iteration using ADR-008 anchored severities?▼

Score a design iteration by comparing before and after snapshots against a fixed heuristic list. The evaluation applies ADR-008 rules to produce anchored 0-10 severities per heuristic for the proposed surface.

When do I need to run a design review with Pareto dominance checks?▼

Run a design review with Pareto dominance checks when evaluating whether an alternative design resolves baseline heuristic violations. It validates if the iteration improves severities without regressions before acceptance.

Does the design re-audit work without baseline severity scores?▼

No, the design re-audit requires baseline heuristic violations with their severities. Without the baseline list, there is no anchored reference to compare the after_snapshot against or run the Pareto dominance check.

What happens when a Pareto dominance check is inconclusive for a design iteration?▼

When Pareto dominance is inconclusive, a weighted-sum fallback calculates overall severity scores. This fallback determines whether the proposed design iteration is accepted despite lacking clear Pareto dominance.

Related Skills

Explore More Design & Creative Skills→
speplinskispeplinski

design-tweak

Propose minimal surface tweaks to reduce residual design-heuristic severities.

Community
Advanced
jkitchinjkitchin
34

python-multiobjective-optimization

Optimize multiple conflicting objectives in Python using pymoo, platypus, or DEAP.

Community
Advanced
matt-rileymatt-riley
1

agent-governance

Enforce governance policies on AI agent tool usage with audit trails.

Community
Advanced
TykoDevTykoDev
9

gatekeeper-design

Audits design documents and outputs structured reports with severity-ranked fixes and routing verdicts.

Community
Advanced