Skip to main content

​
Everything Claude Code

everything-claude-code is a plugin and performance configuration system designed to hyper-charge AI developer agents traversing the Ibex repositories. Originally an Anthropic hackathon winner, it’s tailored to guide AI assistants toward better code patterns, strict security, and modular best practices.

​
Architectural Role

This repository is not a deployable service in the Kubernetes/AWS topology. Instead, it is a development-time asset used by internal engineering and LLM tools (like Claude Code, Cursor, OpenCode) to standardise workflows.

​
High-Level Working

The repository manages subagents, rule directives, and operational skills:
  • Agents: Contains 28 specialised subagents (e.g., code-reviewer, security-reviewer, planner) that can be instantiated by developers to handle specific tasks strictly.
  • Skills: Features 116+ defined workflow skills ranging from React UI patterns to strict Golang and TDD constraints.
  • Hooks: Implements trigger-based logic allowing the agent to halt or change direction if it detects a violation, such as checking for .env secrets during commits.
  • Security Validation (AgentShield): Automatically scans AI configuration contexts against 100+ static analysis rules to prevent prompt injections or unsafe container execution.
Developers within the Ibex ecosystem install this plugin directly into their IDE/CLI environments via npx ecc-install or the marketplace, ensuring all generated code across the microservices adheres to the same architectural layout.