ReddRadar
Topic: ai-agents
18,135 skills in this topic.
-
cost-aware-llm-pipeline
LLM API 使用成本优化模式——基于任务复杂度的模型路由、预算跟踪、重试逻辑和提示词缓存。
xu-xiang/everything-claude-code-zh 383
-
continuous-learning-v2
基于本能（Instinct）的学习系统，通过钩子（hooks）观察会话，创建带有置信度评分的原子本能，并将其演化为技能（Skills）、命令（Commands）或智能体（Agents）。v2.1 版本增加了项目作用域（project-scoped）的本能，以防止跨项目污染。
xu-xiang/everything-claude-code-zh 383
-
continuous-learning
Automatically extract reusable patterns from Claude Code sessions and save them as learned skills for future use.
xu-xiang/everything-claude-code-zh 383
-
content-hash-cache-pattern
使用 SHA-256 内容哈希缓存高昂的文件处理结果 —— 与路径无关、自动失效且服务层分离。
xu-xiang/everything-claude-code-zh 383
-
liquid-glass-design
iOS 26 灵动玻璃（Liquid Glass）设计系统 — 适用于 SwiftUI、UIKit 和 WidgetKit 的具有模糊、反射和交互式变形效果的动态玻璃材质。
xu-xiang/everything-claude-code-zh 383
-
jpa-patterns
JPA/Hibernate 模式，涵盖 Spring Boot 中的实体设计、关系、查询优化、事务、审计、索引、分页和连接池。
xu-xiang/everything-claude-code-zh 383
-
java-coding-standards
Spring Boot 服务的 Java 编码规范：命名、不可变性、Optional 使用、流（streams）、异常、泛型以及项目布局。
xu-xiang/everything-claude-code-zh 383
-
iterative-retrieval
逐步优化上下文检索以解决子智能体（subagent）上下文问题的模式。
xu-xiang/everything-claude-code-zh 383
-
eval-harness
Formal evaluation framework for Claude Code sessions implementing eval-driven development (EDD) principles
xu-xiang/everything-claude-code-zh 383
-
continuous-learning-v2
Instinct-based learning system that observes sessions via hooks, creates atomic instincts with confidence scoring, and evolves them into skills/commands/agents.
xu-xiang/everything-claude-code-zh 383
-
tdd-workflow
在创建新功能、修复 Bug 或重构代码时使用此技能。强制执行测试驱动开发（TDD），要求包括单元测试、集成测试和 E2E 测试在内的测试覆盖率达到 80% 以上。
xu-xiang/everything-claude-code-zh 383
-
continuous-learning
Automatically extract reusable patterns from Claude Code sessions and save them as learned skills for future use.
xu-xiang/everything-claude-code-zh 383
-
coding-standards
Universal coding standards, best practices, and patterns for TypeScript, JavaScript, React, and Node.js development.
xu-xiang/everything-claude-code-zh 383
-
golang-testing
Go testing patterns including table-driven tests, subtests, benchmarks, fuzzing, and test coverage. Follows TDD methodology with idiomatic Go practices.
xu-xiang/everything-claude-code-zh 383
-
golang-patterns
Idiomatic Go patterns, best practices, and conventions for building robust, efficient, and maintainable Go applications.
xu-xiang/everything-claude-code-zh 383
-
frontend-patterns
Frontend development patterns for React, Next.js, state management, performance optimization, and UI best practices.
xu-xiang/everything-claude-code-zh 383
-
security-review
在添加身份验证、处理用户输入、操作机密信息（Secrets）、创建 API 端点以及实现支付/敏感功能时使用此技能。提供全面的安全检查清单和模式。
xu-xiang/everything-claude-code-zh 383
-
strategic-compact
建议在逻辑间隔处进行手动压缩（Manual Compaction），以在整个任务阶段保持上下文（Context），而非随意的自动压缩。
xu-xiang/everything-claude-code-zh 383
-
regex-vs-llm-structured-text
在解析结构化文本时，用于在正则表达式（Regex）和大型语言模型（LLM）之间进行选择的决策框架——优先使用正则表达式，仅针对低置信度的边界情况引入 LLM。
xu-xiang/everything-claude-code-zh 383
-
clickhouse-io
ClickHouse database patterns, query optimization, analytics, and data engineering best practices for high-performance analytical workloads.
xu-xiang/everything-claude-code-zh 383
-
iterative-retrieval
Pattern for progressively refining context retrieval to solve the subagent context problem
xu-xiang/everything-claude-code-zh 383
-
backend-patterns
Backend architecture patterns, API design, database optimization, and server-side best practices for Node.js, Express, and Next.js API routes.
xu-xiang/everything-claude-code-zh 383
-
postgres-patterns
PostgreSQL database patterns for query optimization, schema design, indexing, and security. Based on Supabase best practices.
xu-xiang/everything-claude-code-zh 383
-
project-guidelines-example
xu-xiang/everything-claude-code-zh 383