金色时刻 · 15:48
让好资源,触手可及
AGENT SKILL

windsurf-cascade-agents

windsurf-cascade-agents 是面向效率工具、数据与分析的 Agent Skill,来自 jeremylongshore/claude-code-plugins-plus-skills。查看技能说明、操作步骤与使用约定,并通过原项目了解详细用法。

Create custom Cascade agent configurations for specialized tasks. Activate when users mention "custom cascade agent", "specialized ai agent", "domain-specific cascade", "agent configuration", or "custom ai behavior". Handles custom agent creation and configuration. Use when working with windsurf cascade agents functionality. Trigger with phrases like "windsurf cascade agents", "windsurf agents", "windsurf".

SKILL.md

下载 MD ↓
namewindsurf-cascade-agents
descriptionCreate custom Cascade agent configurations for specialized tasks. Activate when users mention "custom cascade agent", "specialized ai agent", "domain-specific cascade", "agent configuration", or "custom ai behavior". Handles custom agent creation and configuration. Use when working with windsurf cascade agents functionality. Trigger with phrases like "windsurf cascade agents", "windsurf agents", "windsurf".
allowed-toolsRead,Write,Edit,Bash(cmd:*),Grep,Glob
version1.0.0
licenseMIT
authorJeremy Longshore <[email protected]>

Windsurf Cascade Agents

Overview

This skill enables creation of specialized Cascade agents tailored for specific domains or tasks. Custom agents can be configured with domain knowledge, specialized prompts, and focused capabilities. Use cases include security review agents, API design agents, performance optimization agents, and documentation writers that understand your project's specific conventions and requirements.

Prerequisites

  • Windsurf IDE with Cascade Pro or Enterprise
  • Understanding of prompt engineering principles
  • Domain knowledge to encode in agent context
  • Project documentation and conventions documented
  • Test scenarios for agent validation

Instructions

  1. Define Agent Purpose
  2. Create System Prompt
  3. Configure Context Sources
  4. Set Activation Triggers
  5. Test and Refine

See {baseDir}/references/implementation.md for detailed implementation guide.

Output

  • Custom agent configurations in registry
  • Domain-specific system prompts
  • Context files with specialized knowledge
  • Activation trigger mappings

Error Handling

See {baseDir}/references/errors.md for comprehensive error handling.

Examples

See {baseDir}/references/examples.md for detailed examples.

Resources

来源:jeremylongshore/claude-code-plugins-plus-skills · MIT 许可 · 许可原文 · MD 为来源页保存的历史版本