Project Overview

The landscape of AI-assisted development has seen an explosion of configuration files, prompt templates, and MCP server setups, but most remain siloed to a single tool or workflow. Everything Claude Code (ECC) takes a different bet: that the patterns for optimizing AI agent interactions are largely transferable across harnesses, and that the community should share them as a reusable system rather than reinventing them per project. Born from an Anthropic hackathon win and evolved over 10+ months of daily production use, ECC has grown into a comprehensive performance system that includes skills, hooks, memory persistence mechanisms, security scanning via AgentShield, and even a desktop GUI dashboard. What makes it unusual is its explicit cross-harness architecture — it works with Claude Code, Codex, Cursor, OpenCode, Gemini, and others, which suggests the authors prioritized portability over deep integration with any single platform. The project sits in an interesting middle ground between a personal dotfiles repository and a structured framework, and with 174,950 stars and 170+ contributors, it has clearly struck a chord with developers who want their AI tooling configurations to be reproducible and shareable across teams and tools.

What It’s For

ECC is for developers who find themselves repeatedly tweaking system prompts, MCP configurations, and hook scripts for AI coding assistants, and want a more systematic approach. It solves the problem of fragmented, non-portable configurations that break when switching between Claude Code, Cursor, or Codex sessions. The system is particularly well-suited for teams that need consistent agent behavior across multiple developers and CI/CD pipelines — the included memory persistence hooks and continuous learning patterns automate what would otherwise be manual context management. Where it differs from simpler prompt collections is its emphasis on verification loops, parallelization strategies using Git worktrees, and subagent orchestration patterns that address the context window problem. However, the tradeoff is evident: this is a substantial system to adopt, with multiple npm packages (ecc-universal, ecc-agentshield), a GitHub App, and its own dashboard application. Teams looking for a lightweight, drop-in configuration may find the scope overwhelming, while those building production AI workflows will appreciate the depth.

How to Use It

The primary entry point is through the Hermes setup guide, which walks through initializing the ECC system within a project. Rather than a single install command, ECC operates as a layered system: you start with the foundational configuration, then selectively enable components like AgentShield for security scanning, memory hooks for cross-session persistence, or the dashboard GUI for visual monitoring. The cross-harness architecture means the same configuration files can be shared across Claude Code, Codex, Cursor, and OpenCode sessions, with harness-specific adapters handling the differences in how each tool interprets rules and hooks. A notable design choice is the use of Git worktrees for parallel agent operations — this avoids the conflicts that arise when multiple AI agents modify the same working directory, but requires understanding Git worktree semantics. The release notes for v2.0.0-rc.1 highlight the new operator workflows, which suggest ECC is moving toward supporting more autonomous, multi-step agent operations rather than just single-session optimizations.

Initializes ECC in the current project, setting up the core configuration files and directory structure

npx ecc-universal init

Runs AgentShield security scanning against the project to detect prompt injection vulnerabilities and unsafe configurations

npx ecc-agentshield scan

Launches the Tkinter-based desktop GUI for visual monitoring and management of ECC components

python ecc_dashboard.py

Recent Updates

Latest Release: v2.0.0-rc.1 (2026-04-01)

Surface refresh with new dashboard GUI, operator workflows, and ECC 2.0 alpha features including cross-harness architecture improvements

The project shows strong momentum with 174,950 stars and active npm package downloads (ecc-universal weekly downloads are tracked via the badge). The shift toward operator workflows and a desktop GUI suggests the maintainers see ECC evolving from configuration management into a full agent orchestration platform. The cross-harness support for Codex and Gemini indicates a bet on multi-model workflows rather than Anthropic exclusivity.

Sources & Attributions

[1] The project has 174,950 stars on GitHub — github.com/affaan-m/everything-claude-code [2] ECC v2.0.0-rc.1 was released in April 2026 — affaan-m/everything-claude-code@v2.0.0-rc.1 [3] The project has 170+ contributors — github.com/affaan-m/everything-claude-code [4] ECC won an Anthropic hackathon — github.com/affaan-m/everything-claude-code [5] The system supports Claude Code, Codex, Cursor, OpenCode, and Gemini — github.com/affaan-m/everything-claude-code