vue-skills

Agent skills for Vue 3 development.

🚧 Early Experiment

This repository is an early experiment in creating specialized skills for AI agents to enhance their capabilities in Vue 3 development. The skills are derived from real-world issues and best practices, but might be incomplete or inaccurate due to hallucinations.

Please give feedback when encountering any issues.

🚧 This is a community project

I created this project to explore how AI can improve Vue 3 development. If it proves valuable, I plan to propose transferring it to the Vue organization so it can benefit the wider community.

Installation

npx add-skill hyf0/vue-skills

Usage

For most reliable results, prefix your prompt with use vue skill:

Use vue skill, <your prompt here>

This explicitly triggers the skill and ensures the AI follows the documented patterns. Without the prefix, skill triggering may be inconsistent depending on how closely your prompt matches the skill's description keywords.

Available Skills

vue-best-practices (17 rules)

Vue 3 development best practices covering TypeScript configuration, component typing, tooling troubleshooting, and testing patterns.

Type	Count	Examples
Capability	15	Component props extraction, vue-tsc strictTemplates, Volar 3.0 breaking changes, @vue-ignore directives
Efficiency	2	HMR in SSR, Pinia store mocking

Rule Types

Rules are classified into two categories:

Capability: AI cannot solve the problem without the skill. These address version-specific issues, undocumented behaviors, recent features, or edge cases outside AI's training data.
Efficiency: AI can solve the problem but not well. These provide optimal patterns, best practices, and consistent approaches that improve solution quality.

Methodology

Every skill in this repository is created through a rigorous, evidence-based process:

1. Real-World Issue Collection

Skills are sourced from actual developer pain points encountered in production.

2. Multi-Model Verification

Each skill undergoes systematic testing:

Baseline test: Verify the model fails to solve the problem without the skill
Skill test: Confirm the model correctly solves the problem with the skill
Deletion criteria: If both Sonnet AND Haiku pass without the skill, the rule will be deleted

3. Model Tier Validation

Model	Without Skill	With Skill	Action
Haiku	Fail	Pass	Keep
Sonnet	Fail	Pass	Keep
Both	Pass	-	Delete

Acceptance criteria: A skill is only kept if it enables Haiku or Sonnet to solve a problem they couldn't solve without it.

Related projects

vueuse/vueuse-skills - Agent skills for VueUse development
onmax/nuxt-skills - Agent skills for Nuxt development

License

MIT

hyf0/vue-skills

README