Development Agent Skills

create-skill teaches agents how to author new Agent Skills with valid SKILL.md frontmatter, trigger descriptions, and optional scripts. Install via npx skills add or copy to ~/.cursor/skills/create-skill/.

Can help with

• →User asks to create a new skill
• →Authoring SKILL.md files

create-rule helps agents write Cursor rules (.mdc files) for project-specific AI behavior. Place rules in .cursor/rules/ for automatic context injection.

Can help with

• →Creating coding standards rules
• →Project-specific AI guidance

code-review applies structured review checklists covering correctness, maintainability, tests, and team conventions.

Can help with

• →Reviewing pull requests
• →Code quality audit

review-bugbot performs structured code review focused on bugs, edge cases, and regressions. Ideal before merging PRs or after significant refactors.

Can help with

• →PR review requests
• →Finding bugs in diffs

skill-creator is a meta-skill that scaffolds, validates, and packages Agent Skills following the agentskills.io specification.

Can help with

• →Creating new skills from scratch
• →Validating SKILL.md

mcp-builder guides agents through creating MCP servers, tool schemas, and client configuration for Cursor and Claude.

Can help with

• →Building MCP servers
• →Adding custom tools

cloudflare-deploy guides Cloudflare Pages, Workers, D1, KV, and R2 setup with wrangler.toml, CI/CD, and edge patterns.

Can help with

• →Deploying to Cloudflare
• →Workers setup

sdk skill covers Cursor SDK integration for running agents from scripts, CI pipelines, and backend services using @cursor/sdk.

Can help with

• →Cursor SDK integration
• →Programmatic agents

ci-investigator analyzes failed CI checks on pull requests and returns concise root-cause summaries with fix suggestions.

Can help with

• →CI failure on PR
• →GitHub checks failing

test-writer creates tests that cover real behavior and edge cases, avoiding trivial assertions and matching project test conventions.

Can help with

• →Adding tests
• →Test coverage gaps

create-hook helps author hooks.json and hook scripts that run on Cursor agent lifecycle events like before/after tool calls.

Can help with

• →Cursor hooks setup
• →Event-driven automation

api-design guides RESTful API design, OpenAPI documentation, versioning, pagination, and error handling patterns.

Can help with

• →Designing new APIs
• →OpenAPI specs

docker-deploy helps write multi-stage Dockerfiles, docker-compose setups, and container deployment configs for production.

Can help with

• →Dockerizing apps
• →Container deployment

i18n-setup configures i18n routing, translation files, hreflang tags, and locale detection for global web apps.

Can help with

• →Adding i18n to a project
• →Multi-language sites

vibe-code-reader is a community-submitted Agent Skill: 专为快速理解 vibe coding 项目设计的代码解读 skill。当用户上传或提供一个由 AI 生成（vibe coding）的项目，想理解项目结构、业务逻辑、技术架构，或接手别人的 AI 生成代码库时，必须使用本 skill。 触. Install with `npx skills add https://github.com/DylanMay/vibe-code-reader-skill` or copy to your agent skills directory.

Can help with

• →When the user needs help with vibe code reader
• →一个帮助人类程序员快速读懂 AI 生成代码（vibe coding）的 Claude Skill。 接手别人用 Cursor / Windsurf / Lovable 生成的项目时，往往面对一堆能跑但看不懂的代码——没有注释、没有文档、不