ECC

// Overview

ECC packages a fixed engineering loop — plan, test, implement, review, verify, remember, improve — as an installable plugin for coding-agent harnesses, so the process lives in the tool instead of being rebuilt in every prompt. Its own summary of what ships: 68 agents specialised for planning, review, build repair, security, architecture and domain work; 286 skills covering TDD, research, security, docs, frontend, data, ML and operations; 94 legacy command shims kept while the project moves to a skills-first surface; hooks and memory for enforcement, session summaries and context control; selectively-loaded rule packs; and AgentShield, a scanner for prompts, hooks, MCP configuration, permissions, secrets and agent files.

It works best with Claude Code, has a supported Codex path, and provides capability-limited adapters for Cursor, OpenCode, Gemini, Zed, GitHub Copilot, Antigravity, Qwen and other harnesses — the repository publishes a support-status matrix and is explicit that these are not at feature parity. ECC has three public identifiers that are not interchangeable: the GitHub source repo `affaan-m/ECC`, the Claude marketplace/plugin id `ecc@ecc`, and the npm package `ecc-universal`.

There are several install paths and the documentation is emphatic that you should pick exactly one per harness: the guided `npx ecc-universal setup`, Claude Code's native `/plugin` commands, a declarative `extraKnownMarketplaces` block in `~/.claude/settings.json`, or Codex's native repo-marketplace plugin. Stacking two methods into the same harness duplicates skills, commands and hooks; installing once into several different harnesses is fine. The repository is MIT-licensed, with a separately-sold hosted GitHub App (ECC Pro) for private repositories.

// What it does

• 68 agents and 286 skills covering planning, review, build repair, security, architecture, TDD, docs, data and operations
• Hooks and memory for enforcement, session summaries, continuous learning and context-window control
• AgentShield scanning of prompts, hooks, MCP configuration, permissions, secrets and agent files
• Selectively-loaded rule packs (`rules/common` plus a language or framework pack) — Claude plugins cannot distribute rules, so they are copied in by hand
• Native plugin installs for Claude Code and Codex, plus capability-limited adapters for Cursor, OpenCode, Gemini, Zed, Copilot, Qwen and others
• `ecc list-installed`, `doctor` and `repair` for diagnosing and restoring an ECC-managed setup
• Declarative install via `extraKnownMarketplaces` / `enabledPlugins` in `~/.claude/settings.json`

// Getting started

Choose one install path per harness — the guided npm setup is the recommended default. It needs Node.js 18 or newer; the Claude plugin path also needs Git and Claude Code 2.1 or newer on PATH.

npx ecc-universal setup
# or, to cover several harnesses in one flow:
npx ecc-universal install --guided

/plugin marketplace add https://github.com/affaan-m/ECC
/plugin install ecc@ecc

codex plugin marketplace add affaan-m/ECC
codex plugin add ecc@ecc
codex plugin list --json

git clone https://github.com/affaan-m/ECC.git
cd ECC
mkdir -p ~/.claude/rules/ecc
cp -R rules/common ~/.claude/rules/ecc/
cp -R rules/typescript ~/.claude/rules/ecc/

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Reach for it when you keep re-typing the same plan-test-review discipline into every agent session
• Reach for it when you want a large curated skill and sub-agent library instead of assembling one repo by repo
• Reach for it when you need the same agent setup mirrored across Claude Code, Codex and a third harness
• Reach for it when you want an agent configuration audited — AgentShield scans hooks, MCP config, permissions and secrets

// How ECC compares

ECC alongside other open-source agent skills & plugins tools AI/TLDR tracks, ranked by GitHub stars.