affaan-m/everything-claude-code

Everything Claude Code (ECC) is a comprehensive agent harness performance optimization framework that transforms AI coding assistants into production-ready autonomous developers. Born from an Anthropic hackathon winner, it ships with 28 specialized agents, 125+ skills, and 60+ commands that work across Claude Code, Codex, Cursor, and OpenCode from a single repository. The framework addresses a real gap: raw AI model capabilities don't translate directly into reliable coding output. ECC bridges that gap with token optimization (reducing context waste), persistent memory across sessions, verification loops that catch errors before they ship, and security scanning that flags vulnerabilities in generated code. Version 1.9.0 (March 2026) introduced a selective-install architecture — instead of loading everything, developers pick only the language ecosystem they need (TypeScript, Python, Go, Java, Rust, C++, or Perl). This keeps the agent harness lightweight while still providing deep, language-specific optimization through dedicated code reviewers and build resolvers for each ecosystem. Key agents include: a Planner for task decomposition, an Architect for system design, language-specific code reviewers (TypeScript, Python, Go, etc.), a security reviewer for vulnerability scanning, build resolvers for CI/CD debugging, and a loop operator for complex multi-step workflows. The skills library covers TDD workflows, backend/frontend development patterns, continuous learning, strategic context compaction, and analytics integration. With 110K+ stars and 14K+ forks, ECC has become the de facto standard for Claude Code power users who want more than the default agent behavior. The MIT license means you can fork and customize it for your team's specific needs.