agents

This skill acts as a comprehensive framework for developing and managing agentic AI patterns within your local development environment. Designed for software engineers, devops practitioners, and AI architects, it facilitates the generation of standardized skill files that enable AI agents like Claude Code, Cursor, Cline, Windsurf, and Copilot to understand project-specific conventions, tooling, and architectural patterns. By automating the creation of skill structures, it ensures that your AI assistants move beyond generic behavior to provide highly context-aware, multi-step assistance tailored to your unique codebase.

• Claude API integration: Provides standardized patterns for streaming message APIs, system prompt design, and managing complex user contexts for reliable reasoning.

• Multi-agent compatibility: Supports a wide range of popular AI agents including GitHub Copilot, Continue, Roo, Amp, Gemini CLI, and Goose through centralized config management.

• Pattern-driven generation: Uses automated generators to scan project directories, analyze dependencies (package.json), and output consistent markdown-based skill definitions.

• Tool calling support: Enables structured function definitions and complex workflow orchestration, allowing agents to execute specific tasks like file system scanning or code generation.

• Verbose logging and diagnostics: Built-in debugging capabilities, including real-time output streaming to stderr and dry-run detection, ensure visibility into the agent's decision-making process.

• Requires an ANTHROPIC_API_KEY environment variable for advanced generation tasks involving Claude models.

• Best practices include strict separation of prompt engineering files (src/prompts/) and generator logic (src/generators/) to maintain maintainable version control.

• The skill encourages 'context over tokens' by intelligently filtering relevant codebase files to avoid input overflow, ensuring higher quality responses for reasoning tasks.

• Always use asynchronous file system operations (fs/promises) when building custom agent patterns to avoid blocking event loops in node-based workflows.

• Constraints: Primarily designed for Node.js 18+ environments using native ESM modules for performance and compatibility with modern AI CLI tools.