Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add ulises-jeremias/agent-toolkit --skill design-improvementgit clone --depth 1 https://github.com/ulises-jeremias/agent-toolkitWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ulises-jeremias/agent-toolkit/design-improvement)<a href="https://agentmods.dev/skills/ulises-jeremias/agent-toolkit/design-improvement"><img src="https://agentmods.dev/badge/skills/ulises-jeremias/agent-toolkit/design-improvement/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/ulises-jeremias/agent-toolkit/design-improvement"><img src="https://agentmods.dev/badge/skills/ulises-jeremias/agent-toolkit/design-improvement.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 2 findings, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Excessive Agency · line 112 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
- medium Excessive Agency · line 154 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00060 | $0.02616 |
| Opus 5 | $0.00030 | $0.01308 |
| Sonnet 5 | $0.00012 | $0.00523 |
| Haiku 4.5 | $0.00006 | $0.00262 |
Grade A, and why
design-improvement scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 168 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Design Improvement (WHAT)
Close the loop from assessment → implementation → rendered verification. Use this skill when the user wants to act on design-assessment (or frontend-design-review) findings, improve an existing interface, or run a browser-grounded iteration loop.
Composition, not duplication: Orchestrates design-assessment, frontend-design, frontend-design-review, web-design-guidelines, playwright-cli / chrome-devtools, figma and accessibility — delegates work, does not copy their instructions.
Default guardrails
- Consume assessment findings — start from a
design-assessmentscorecard /frontend-design-reviewoutput (Blocking/Major/Minor + evidence links). If no assessment exists, offer to rundesign-assessmentfirst; do not invent findings. - Reuse the single evidence model:
observation / impact / severity / effort / confidence / evidence / screens / recommended fixplus 1–5 scale,3 = Defined,Not assessed, confidenceHigh/Medium/Low, andoutput-handshakebefore final artifacts. Do not introduce72/100or other synthetic scores. - Do not assume code is good without rendered evidence (screenshot/recording via browser). Implement → run → capture → inspect → compare → iterate. Where browser/vision unavailable, use text-only heuristic and mark Low confidence — browser/vision unavailable.
- Do not auto-push to default branch or create tickets/docs without explicit approval. L2 controlled mutations for code changes; app runs isolated (no prod).
- Generic-AI lens is contextual, not a checklist — a pattern is not bad merely because AI often produces it; judge appropriateness for this product's subject and tone.
Safe vs ambiguous changes
Triage before implementation:
| Safe / mechanical — low judgment, implement directly with review | Ambiguous / product-defining — requires human direction, propose then confirm |
|---|---|
| Spacing/token inconsistency, broken responsive reflow, contrast violation (WCAG failure), missing focus state, overflow/clipping, misaligned design-system token, dead/incorrect ARIA, duplicate primary action | Brand direction, major visual identity, information architecture / navigation model, product-specific interaction changes, large typography or palette direction, pricing/offer hierarchy |
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 10d ago First seen · 168 lines · 60 tokens per session scan A 54ba177c2844
design-improvement is a skill published in the GitHub repository ulises-jeremias/agent-toolkit (16 stars, last pushed today), licensed MIT. It adds 60 tokens to every session and 2,616 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
vibe-to-ui
Design systems, motion, mood boards, spatial layout, visual assets, Design Context profiles, and a global Inspiration Library from screenshots, website URLs, inspiration images, music, or fuzzy aesthetic intent. Classifies page archetype, explores 3 product-aware directions before locking tokens (unless exact…
ui-ux-design
Create production-grade frontend interfaces with strong UX and visual craft. Use when building web components, pages, dashboards, forms, landing pages, modals, or any UI. Use when user says 'build a form', 'create a dashboard', 'design a component', 'make a landing page', or asks for UI/UX work. Use even when the user…
margaret-design
Figma bridge, both directions: generate a new Figma design or diagram from a prompt/spec, or pull an existing Figma file/selection into this codebase (as code, a design-context summary, a Code Connect link, or a motion implementation). Use when the user asks to create/generate something in Figma or FigJam, wants a…
design-logo
Generate a commercial-grade SVG logo for a project by reading its codebase and iterating with the user.
ndv-perceptual
Perceptual cognition module. Injects simultaneous read-as-rendered vision into any phase where visual or UX quality is a concern — implementation review, component assessment, UI audit steps within a larger workflow. Source: ndv-design (Pixel).
ux-audit
Production-grade UX audit and design guidance skill. Transforms vague design feedback into actionable, implementable recommendations. Two modes: guide (principles + do/don't rules for modern interfaces) and review (structured audit with prioritized fixes). Covers task-first UX, information architecture, CRAP visual…