affaan-m/everything-claude-code

Everything Claude Code describes itself as a performance optimization system for the kind of tooling that wraps an AI coding assistant, with named targets including Claude Code, Codex, Cursor, OpenCode, and Gemini. The README frames it as more than configs: a system of skills, instincts, memory optimization, continuous learning, security scanning, and research-first development, grown out of more than ten months of daily use building real products. The project is MIT licensed and credits an Anthropic hackathon win. The repo ships a catalog the README counts at 48 agents, 182 skills, and 68 legacy command shims, with rule sets spanning 12 language ecosystems. The latest release, v2.0.0-rc.1, adds a Tkinter desktop dashboard, operator-focused agents such as brand-voice, customer-billing-ops, google-workspace-ops, and workspace-surface-audit, plus media tooling for Manim and Remotion video work. An in-tree Rust control-plane prototype under ecc2/ exposes commands like start, sessions, status, resume, and daemon, and is described as alpha rather than a general release. Version 1.9.0 introduced a selective install pipeline through install-plan.js and install-apply.js, backed by a SQLite state store, six new language reviewers and build resolvers covering TypeScript, PyTorch, Java, and Kotlin, and reliability fixes for an observer loop. Companion npm packages include ecc-universal and ecc-agentshield, the latter tied to the security guide. The README says the repo is raw code and points readers to three companion guides. The Shorthand Guide covers setup and philosophy, the Longform Guide covers token optimization, memory persistence via hooks, evaluation loops, and parallelization through git worktrees, and the Security Guide covers attack vectors, sandboxing, sanitization, and CVEs. Someone who already uses an AI coding tool every day and wants a prebuilt layer of skills, rules, hooks, and security tooling, rather than building one from scratch, is the intended reader.