Builds specialized, production-ready AI agents using a rigorous 5-phase methodology and domain expertise loading.
The Agent Creator skill implements an industry-leading 5-phase SOP for developing high-performance AI agents within the Claude Code ecosystem. By combining expertise loading, meta-cognitive extraction, and evidence-based prompting, it allows developers to build agents that are consistently reliable, domain-aware, and seamlessly integrated with MCP servers. This skill facilitates the creation of project-specific agents with deeply embedded domain knowledge, ensuring high-quality performance with less than a 3% failure rate in production environments.
主な機能
01Meta-cognitive extraction to identify heuristics and decision-making frameworks
025-Phase SOP Methodology (v2.0) for systematic, repeatable agent development
03Cross-session persistence via memory-mcp for agent specs and metadata
04Phase 0 Expertise Loading for deep domain-specific context and pattern recognition
05Evidence-based prompting integration including Self-Consistency and Plan-and-Solve
066 GitHub stars
ユースケース
01Developing project-specialized agents with deeply embedded technical domain knowledge
02Rewriting and optimizing existing agent prompts to improve consistency and reliability
03Designing complex multi-agent workflows requiring sequential or parallel coordination
