UI Ecosystem Recommendations
Match an established frontend direction with compatible UI tools, libraries, examples, and skills before implementation starts.
Basic info
- Name
- UI Ecosystem Recommendations
- Description
- Use after frontend technical/design choices are complete and Codex is about to build, redesign, polish, or evaluate a UI. Trigger when the user asks UI work to recommend or use stack-matched MCPs, plugins, skills, component libraries, design systems, templates, official examples, or ecosystem tooling to improve frontend beauty and avoid hand-rolled UI. Also match Chinese requests like "ui 根据前端选型推荐 mcp/插件/skill", "前端选型完成后不要手搓", "提升前端设计效果/美观", "根据技术栈选 UI 生态". Do NOT trigger before frontend direction exists, for backend-only work, generic docs lookup, pure bug fixes, or UI review that only needs web-design-guidelines.
ui-ecosystem
Turn an established frontend direction into a practical ecosystem plan so the UI can borrow mature components and patterns instead of rebuilding everything.
When to use it
Recommend UI tools
My framework, styling approach, page type, and design direction are already known, and I want stack-matched libraries, tools, skills, plugins, or official examples.
Build with the ecosystem
I want a page built or polished, and expect approved ecosystem choices to be used in the actual implementation rather than left as a recommendation list.
Audit ecosystem use
I have an existing interface and want to identify custom controls, icons, motion, forms, tables, or charts that should rely on mature project-compatible primitives.
Not for
Use strategy work before the product shape or frontend direction exists. Use dedicated review workflows for accessibility-only audits, generic design compliance, backend bugs, tiny CSS fixes, or installation of one already selected package.
What it produces
A UI_ECOSYSTEM_PACKET is produced before any frontend edit or dependency installation.
- Context summary: records the framework, styling, design system, page type, constraints, and local files inspected.
- Decision matrix: labels each candidate
adopt,activate,absorb,spike, orreject, with fit, risk, and action. - Implementation plan: lists planned installations or activations, the remaining custom surface, verification methods, and unresolved items.
- Mode-dependent side effects: recommendation mode is read-only; implementation mode may edit UI files and install authorized dependencies; audit mode edits only when fixes are requested.
- Verification evidence: runs relevant type checks, lint, tests, builds, and browser inspection when rendered UI changes.
- Never does: it does not invent tool names, install without authority, copy unlicensed templates, or let ecosystem tools override project, accessibility, or design-system rules.
Prerequisites and boundaries
Prerequisites
The frontend framework, styling approach, page or application type, and enough design context must already exist. Project rules, manifests, components, styling configuration, and current official documentation need to be inspectable.
Discovery requirements
- Available MCPs, plugins, and skills must be discovered before they are named as usable.
- Existing project libraries and design-system primitives take priority over new dependencies.
- License, compatibility, bundle impact, runtime impact, and accessibility risk are checked before adoption.
Nearby responsibilities
| Need | Use instead |
|---|---|
| Choose product shape or frontend stack | strategy-first-development |
| Apply TranFu brand rules | tranfu-website-design |
| Review accessibility or design compliance | web-design-guidelines |
Subtle boundary
“Recommend tools” stops after the packet. “Build,” “polish,” or “improve the UI” continues into implementation, but dependency installation still requires clear authorization.
Version info
Local public Skill catalog snapshot, showing only public-safe fields.
Skill files
(8)SKILL.md
SKILL.md · Markdown